PodSearch

Cortex

134: AI Art Will Make Marionettes Of Us All Before It Destroys The World

 

00:00:00   Well, we have a lot more terrifying things to discuss about AI.

00:00:05   A lot has happened since the last time we spoke, Myke.

00:00:08   I don't presume that I'm the center of the universe.

00:00:12   But there is an almost Truman Show-like aspect sometimes to life,

00:00:17   and this is one of those things where I feel like,

00:00:20   you know, I was seeing enough about A.I.R. around the place,

00:00:23   like, you know, there was a reason that we spoke about it,

00:00:26   because it was coming into our worldview more and more.

00:00:28   Right.

00:00:29   But in the time between these last two episodes,

00:00:31   I feel like it has taken over the world.

00:00:34   I can't move for people talking about it.

00:00:37   It's coming up on podcasts that have got nothing to do

00:00:40   with this stuff, like not even tech related shows

00:00:43   they're talking about.

00:00:43   Like Idina sent me a link to one of her favorite podcasts,

00:00:46   it's a podcast called Comic Lab,

00:00:48   which is about cartooning and making comics.

00:00:52   They did like a whole episode talking about it.

00:00:54   And it makes sense for them a little bit

00:00:56   because they are comic artists, right?

00:00:57   They are artists.

00:00:58   - But it's like, why now?

00:01:00   - Yeah, I think this is definitely

00:01:03   the biggest Truman Show-like effect

00:01:05   for anything that we have ever spoken about.

00:01:08   And I think it's just that we happened to hit it

00:01:10   at the exact right moment before a bunch of things exploded.

00:01:15   - Yeah, it's like realistically, I know why this happened,

00:01:18   but it's still such a strange feeling.

00:01:20   This happened because we spoke about it

00:01:22   because it was starting to become a thing

00:01:23   that we couldn't avoid, right?

00:01:25   So that's why then it's just like a snowball effect.

00:01:28   Oh, I know, I know. And like listeners may remember that I had a conference that I said

00:01:33   I was going to go to in September. I was like, "Oh, guess what?" One of the major topics

00:01:38   like everyone was discussing at the conference. It was a bunch of AI stuff. Like my whole

00:01:42   life just exploded into this in the past month. And I think also with the last episode of

00:01:48   the show, there was like a little bit of confusion over our positions because there's just

00:01:53   like there's so much that this touches on, right? Like this, we just kind of talked about

00:01:57   it off the cuff. And, you know, I really think this this absolutely touches on almost everything

00:02:04   in the world. So I think to try to be clear, it's like, with AI arts, and the language

00:02:10   models in particular, in the short term, I think there's a lot of interesting discussions

00:02:14   to be had about how it affects technology and how it affects the economy and how it

00:02:19   affects ethics. But in the long term, we're seeing the beginning of the end of the human

00:02:26   race.

00:02:27   Okey-dokey.

00:02:28   So…

00:02:29   Is this the clarifying you were looking to do?

00:02:32   Yeah, just so people can understand where I'm coming from.

00:02:35   There's lots of interesting discussions to be had where you can take many different

00:02:39   positions, but ultimately I think this all points in one direction, which is not very

00:02:44   good.

00:02:45   But yes, in the short term, since the last time we spoke, so much has happened, and I

00:02:52   I think one of the first things that you sent to me was that Dolly is now open for the public

00:02:57   to use.

00:02:58   Yeah.

00:02:59   And the thing that I think really helped kick it off was, because last episode we were talking

00:03:04   about how you, oh, you need these like massive computing clusters to do this kind of stuff.

00:03:08   I don't know.

00:03:09   It was like 10 days after that episode went up, the very first desktop versions of these

00:03:15   things came out.

00:03:17   And the one I've been playing with is called Diffusion B, which allows you just on your

00:03:23   Mac as long as it's one of the new M processors so that it's fast enough, it allows you to

00:03:28   start typing in whatever and generating AI art.

00:03:32   And so it's like, oh, okay, this went from supercomputing clusters only down to, oh,

00:03:40   it can happen on your desktop, which is I think the first time that lots of people actually

00:03:45   started to play with it firsthand and I think that's kind of what kicked off for lots

00:03:51   of people.

00:03:52   "Oh this is really real, like I can play with it and I can see what it does."

00:03:58   The first publicly used by us piece of AI art went up which was running our Cortex logo

00:04:05   through some AI art stuff.

00:04:08   - How did you do that?

00:04:09   - That was done with, oh god what was it called, it was done with Dream I think is one of the

00:04:13   projects but you can upload a piece of art and have it like iterate on that piece of art.

00:04:20   - Oh, okay. That makes sense then. Because I wonder, it's like, how did you get it to do the Cortex logo?

00:04:24   - This is also one of these things like, even Diffusion Bee, which is the thing that you can

00:04:28   play with on your laptop, it started out as, oh you can only just put in words and then it will

00:04:33   generate images, but even that one now, you can give it an image and then ask for modifications

00:04:38   on that image, but the dream program just lets you put in a piece of art and then it

00:04:43   just iterates on it in some way.

00:04:45   Part of the intrinsically horrifying thing about all of this art is that a bunch of examples

00:04:50   had to be rejected because the AI knew that it was a brain and so ended up making it look

00:04:57   very like, "Oh, it's a horrifying mass of meat that's also a brain."

00:05:04   My favorite.

00:05:05   Which brings us to AI Art.

00:05:07   Why you so terrifying?

00:05:09   So last time we spoke, another one of these things was, "Oh, obviously this is coming

00:05:14   for video.

00:05:15   It's just a matter of time.

00:05:16   It's just a question of how long does it take computationally to do these sorts of things?"

00:05:21   And it's like, "Oh, is that going to happen next year?"

00:05:24   No, it happened two weeks after the show went up.

00:05:28   And I have something for you to take a look at, which was the first version of this, which

00:05:34   is this was made a little bit manually.

00:05:37   It was made with stable diffusion.

00:05:39   But the prompt is a journey through a boy's life.

00:05:43   Again, this is not true video, but it's like an,

00:05:48   I don't know how to describe it.

00:05:50   It's a-- - It's like,

00:05:50   I'm watching it now, it's like a flip book.

00:05:52   - Yeah, it's a bit like a flip book.

00:05:53   So just take a look at it

00:05:54   and tell me what you think about this.

00:05:56   - I mean, it's weird.

00:05:58   Like, it's not much of a video to me, really.

00:06:01   Like, I can see it's just like a bunch,

00:06:02   you know, you're just cycling through a bunch of images.

00:06:05   It's got weird,

00:06:05   I just got weirdly religious there for a minute.

00:06:07   What is happening here?

00:06:09   Like what is going on here?

00:06:10   Did an AI generate these and put them together?

00:06:13   Or like, are they generating them from each other?

00:06:16   Like what is going on?

00:06:17   How is this being made?

00:06:18   - Yeah, so I read a little bit of what the author said.

00:06:21   And the impression is, he started with a general image

00:06:26   that's called a journey through a boy's life.

00:06:28   And then each of the next frames

00:06:31   is based on the frame before it,

00:06:34   given that same piece of information.

00:06:36   It's going in some really weird places.

00:06:37   We just went to war for a while,

00:06:40   and now there's some kind of eldritch horror occurring.

00:06:43   Okay, I mean, it's really weird.

00:06:45   I watch something like that

00:06:46   and knowing where it's coming from,

00:06:48   and it doesn't make me feel great in a way,

00:06:53   because of the weird places the AI takes itself.

00:06:59   And there was an uncomfortability in that.

00:07:02   I do not know why we spent a significant portion of that

00:07:07   in various wartime.

00:07:10   And no one can really answer that question.

00:07:14   I assume, there's no way to know why the AI went that route

00:07:17   and stuck in it for quite some time.

00:07:19   - So as a description for the listeners,

00:07:22   I mean, I think it's sort of horrifying

00:07:26   and it's very dirty looking for some reason,

00:07:29   But it is like a sequence of images that start with a child and the child slowly grows up

00:07:35   and you just see a bunch of things like a kid at the desk doing his homework and it

00:07:40   sort of transitions.

00:07:42   The kid just gets increasingly older and older and then yes, it's young men at war for a

00:07:46   while and I presume this is one of these side effects of what's in the training database

00:07:51   and it has some concept of, "Oh, based on all of the images I've seen, what has a boy

00:07:57   been doing in his 20s, and so it generates like a ton of war imagery.

00:08:03   I also suspect that's why at that moment it also gets weirdly religious, like a lot of

00:08:07   crosses appear, because I presume that that's coming from graveyard stuff.

00:08:12   And it goes all the way through to ending with like a, you know, a dead body laying

00:08:18   on a table with sort of muddy blood coming down from it.

00:08:22   This was the first thing that I saw that I thought, oh, this is a video.

00:08:27   And I would also legitimately say, oh, this is a piece of art.

00:08:31   Like you could, you could display this in an art museum and it wouldn't be out of place.

00:08:38   And what it made me think of was, I don't know if you've ever seen the animations from

00:08:42   Pink Floyd's The Wall.

00:08:44   That's what this made me think of.

00:08:45   Oh, it has that kind of feeling to it.

00:08:49   And it has a really horrifying animation style.

00:08:53   And I went back and wanted to rewatch some of the animation sections of that movie.

00:08:59   And boy, it was an interesting experience, because in comparison, suddenly the walls

00:09:06   seemed remarkably undetailed.

00:09:10   And I just felt like I could only see, oh, it's so simple.

00:09:14   And like that is perhaps one of the most complicated examples of like hand drawn animation is the

00:09:20   wall.

00:09:21   Oh God compared to this.

00:09:22   Oh, yeah, they just can't possibly put the detail in every single frame that exists in

00:09:28   this one thing.

00:09:30   And so I just I just had a real feeling of, wow, what a jump.

00:09:34   It's sort of addressing the same idea, but in 100,000 times more detail.

00:09:41   So anyway, that'll be in the show notes for people to take a look at and then I don't

00:09:45   know another seven days after that meta and a few other companies announced true text

00:09:54   to video projects and I have some links in the show notes that you can click on and so

00:09:59   this is called make a video and if you if you take a look at some of these links, I

00:10:04   think this stuff looks more like what stable diffusion does.

00:10:10   - Okay, this looks like, take all of the generators

00:10:15   and make them do animation, right?

00:10:17   Like it looks, it has a similar look to,

00:10:21   like the quality of some of the imagery is like,

00:10:25   has the telltale signs that this was made

00:10:28   by an AI art generator.

00:10:32   - Yeah, I think the key characteristic

00:10:34   of a lot of this stuff is it's still quite dreamlike,

00:10:38   because a lot of the details aren't there. There's a lot of areas where it's kind of fuzzy.

00:10:44   But I think that this again is the thing where a lot of the detractors of this kind of thing said,

00:10:50   "Oh, you'll never be able to do video." And you go, "Well, yeah, let's just wait. Let's just

00:10:57   wait and see how long this takes." And we go, "Okay, here we go. Here's the first versions

00:11:03   of video where you can just type the words "a dog wearing a superhero outfit with a red cape flying

00:11:09   through the sky" and it makes that. Like it makes a little video of that thing. And there's another

00:11:15   video project which I'll try to find for the show notes for viewers later which was this, but it's

00:11:21   like a multi-scene description. So you can say things like "a woman's swimming in the ocean,

00:11:27   she dives under the water and picks up a starfish on the ground and then returns to the shore and

00:11:32   And it's able to keep that concept straight the entire time and construct the whole scene

00:11:39   instead of just an image that's moving.

00:11:42   So the speed on this in the space of what, five weeks has been absolutely breathtaking.

00:11:52   This all seems inevitable.

00:11:55   I will say I am surprised at the speed like you, right?

00:11:58   that we've, from the last episode to now, how much has happened?

00:12:03   ALICE The thing that I just want to draw people's

00:12:05   attention to that gets overlooked in the AI art discussion is, people are talking about

00:12:12   the art because it's visual.

00:12:15   You can see it, I think it's the most understandable thing that's occurring.

00:12:21   But to me, the much bigger deal is the less visible language model stuff.

00:12:28   And so it is the same idea, but for words where AI is responding to some text prompt

00:12:38   with an answer of some kind.

00:12:42   And we talked in more texts about AI explaining jokes like how an AI can explain a joke listeners

00:12:49   can go check that out.

00:12:50   But I have an example that I didn't mention that a listener pointed out from that paper

00:12:54   that I had missed, which I think is much more impressive.

00:12:57   So again, you can do this same thing.

00:13:00   Ai doesn't understand words.

00:13:03   And so I'm going to give an input a thing that the user told the computer and then the

00:13:09   output how the computer answered.

00:13:12   So here is the input and this is this is very strange.

00:13:15   It's almost like a puzzle just so you can understand what I'm what I'm saying here.

00:13:20   Input Michael is at that really famous museum in France looking at its most famous painting.

00:13:26   However, the artist who made this painting just makes Michael think of his favorite cartoon

00:13:32   character from childhood.

00:13:35   What was the country of origin of the thing that the cartoon character usually holds in

00:13:40   his hand?

00:13:43   So do you think you have an answer to that question?

00:13:47   No.

00:13:48   Okay, here is what the AI said.

00:13:52   The most famous painting in the Louvre is the Mona Lisa.

00:13:56   The artist who made the Mona Lisa is Leonardo da Vinci.

00:14:00   Leonardo da Vinci is also the name of the main character in the cartoon Teenage Mutant

00:14:04   Ninja Turtles.

00:14:06   Leonardo da Vinci is from Italy.

00:14:08   The thing that Leonardo usually holds in his hand is a katana.

00:14:12   The country of origin of the katana is Japan.

00:14:16   The answer is Japan.

00:14:17   Okey dokey.

00:14:18   Right?

00:14:19   I mean, part of my issue there is I forgot it was Leonardo da Vinci who made the Mona

00:14:22   Lisa but I would not have remembered that Leonardo the turtle from Ninja Turtles

00:14:27   held a katana because I thought that he had like a kendo stick or something that

00:14:31   is horrifying right so because this is because this is actually a good example

00:14:37   right because my human brain forgets things mm-hmm the AI brain doesn't it

00:14:46   just has all the information or can get it I assume I don't know where from but

00:14:51   But like the difference between pitching me against the machine there is I just couldn't

00:14:57   remember a couple of key pieces of information, which now you tell me I did know them.

00:15:02   It's weird.

00:15:03   Yeah, or even when I first read this input, it's just phrased in such a strange and vague

00:15:09   way.

00:15:10   Yeah, I do wonder why they did it that way.

00:15:12   That is odd to me.

00:15:13   I think the whole purpose of why it's framed in this really strange way is to give the

00:15:19   minimal amount of information that you can regarding what the actual answer is. They

00:15:25   don't even say like the most famous museum in France, it's a really famous museum in

00:15:29   France. Right? And also they're like, they're trying to get an answer, but they're asking

00:15:34   a bunch of questions that require recalling the previous answer, right? Right. It's a

00:15:40   multi-stage thing to think through. I think the particular one that's really killer here

00:15:44   is the artist who made this painting just makes Michael think of his favorite cartoon

00:15:49   character from his childhood. For the AI to make the connection, "Oh, it's Leonardo

00:15:55   da Vinci. Leonardo is one of the Teenage Mutant Ninja Turtles."

00:15:58   - Because there's no concept of how old the person is, right?

00:16:01   - Yeah, there's no concept of how old the person is. Simply being able to draw that

00:16:06   out of like, what painting in France would make someone think of a childhood cartoon?

00:16:13   a crazy abstract thing to think of, but the machine got it. So I think this like input

00:16:19   output is a an example of again the AI art stuff is interesting. The text stuff to me

00:16:27   is where a lot of the much more concerning stuff comes from and there's a little bit

00:16:32   of a less visceral demo, but I actually found this the most striking thing I have found

00:16:38   in language models, but it's a little bit hard to describe. So there's an article

00:16:42   called "Using GPT-3 to Pathfind in Random Graphs". GPT-3 is one of these language models.

00:16:50   It's the most advanced one that people have access to. Although at the conference I was

00:16:56   at I got to see some of the not publicly released stuff, which was much more terrifying, but

00:17:00   we'll just leave that alone for now. GPT-3 is the much older one. And this is the most

00:17:07   impressive thing that I've seen. So this is a little bit hard to describe on a podcast,

00:17:13   but there's this problem in mathematics, which is called the traveling salesman problem,

00:17:16   which is say, Oh, you're a salesman and you have to get to 10 cities in the United States,

00:17:23   what's the shortest possible path to travel between those 10 cities in any order. And

00:17:28   it's one of those things like it seems like it should be easy, but it turns out this is

00:17:32   just incredibly difficult to actually solve in a reasonable way. Just a hugely computationally

00:17:38   intense problem. So someone basically got the idea of, hey, why don't I see if GPT three

00:17:44   can solve the traveling salesman problem. So I think you need to go read the paper to

00:17:49   see the details. But let me just describe it in a general way, where the person gave

00:17:55   an input to GPT three that says something like this. There's 17 locations, we need to

00:18:03   find the shortest path between these locations. Location one is connected to location seven

00:18:09   and 13. Location 13 is connected to locations nine and two. So they just wrote out like

00:18:16   a bunch of here's all of the connections. You're currently at location three, find the

00:18:21   optimal path to location eight. And GPT three just did it. Basically, half of the time it

00:18:30   was able to find an optimal or near optimal path, just given a bunch of random locations

00:18:36   and the connections between them. And what's just what's like, that may not sound like

00:18:42   much but what needs to be understood here, what is absolutely mind blowing is that GPT

00:18:47   has just been given a ton of like text documents from which to derive the world.

00:18:54   And inside of it, somehow it has the concept of locations and connections

00:19:01   and what does it mean to try to find the path to a different location.

00:19:06   Somehow it's thinking of what is the optimal solution.

00:19:10   And again, like it's not programmed for this.

00:19:12   This is an incredibly difficult math problem, but it has an idea of what

00:19:17   of what's going on. And so I've just seen more and more people are really trying to

00:19:22   push the edges of these language models to say, what do you understand? And the answer

00:19:30   keeps coming out to be much more than you might have imagined that there's a better

00:19:34   understanding of something in the world that has just been derived from dumping tons and

00:19:42   tons of text files into this big database and building a neural network on top of it.

00:19:47   It is genuinely, genuinely terrifying. And I'll just say like, I don't think I can talk

00:19:54   in much detail about what I have seen. But I saw a couple demos of the next generation of this stuff

00:20:01   in person. And it was extremely alarming what you could ask it to do. And it could give reasonable

00:20:08   answers to. And I think just like the AI art stuff, it has clearly crossed into the realm

00:20:17   of creative in a way that I think people just wouldn't have expected.

00:20:22   Like writing fiction?

00:20:23   Exactly, yes. That's exactly the kind of thing. Some public similar projects that are

00:20:28   easier to talk about are, there's a few cases of auto-generated text adventures. So

00:20:35   Exactly the thing that we do, right, for the bonus episodes at Cortex where we work through

00:20:39   a text adventure that someone has created, there are now projects that do that, where

00:20:44   you can just play an infinite text adventure where it keeps spitting out like, "Oh, you're

00:20:49   in this room and here are your options of what you can do." And if you select an option,

00:20:53   like it'll just go forever in a coherent way. It's like, "Oh my!" So yeah, I think

00:21:01   the AI art stuff is flashy. The language model stuff is what people are going to be quite

00:21:08   surprised at how much it might take over soon. And that includes a lot of things that people

00:21:14   would not expect like computer code. There are already some public examples of this of

00:21:20   you express in human language, the computer code that you want created, and it is able to create

00:21:27   that code for you. So we're rapidly encroaching on computers programming themselves territory.

00:21:34   Mykey - Definitely want the computer to think for itself. That's what I'm looking for.

00:21:39   Alex - I mean look, look, we can get to the Doom stuff later. Like there's so much other

00:21:43   stuff to talk about.

00:21:44   Mykey - Wait a second, what do you mean later? Am we not already in it? I feel like we're

00:21:47   in it. What are you talking about?

00:21:48   Alex - No, no Myke. I don't think this is the Doom stuff at all. This is just timeline

00:21:52   stuff.

00:21:53   Mykey - Oh okay. May I follow up at this point?

00:21:54   "Hey, what's happened since we last spoke?"

00:21:57   And a lot has happened since we last spoke.

00:22:00   And this is why at the beginning I kind of divided,

00:22:03   "Oh, there's stuff in the near term,

00:22:05   and there's stuff in the long term."

00:22:07   And we're still just in the near term conversation, Myke.

00:22:09   Like we're not even remotely close

00:22:11   to the long term conversation.

00:22:13   (laughs)

00:22:15   Hey, this isn't an ad,

00:22:17   but we're putting it in between the ad sounds

00:22:20   because Myke and I have been talking

00:22:23   for forever about AI.

00:22:26   And we forgot that we need to tell you about the subtlety

00:22:31   and the subtle sweater going on sale

00:22:33   sometime before the 90 minute mark of the show.

00:22:36   So we're just breaking into our own conversation now

00:22:40   from the future.

00:22:41   - This is us from the future coming back to tell you

00:22:44   that we talk about AI for an hour and a half

00:22:46   accidentally in this episode.

00:22:48   And we realize as purveyors of fine merchandise,

00:22:52   That is a terrible way to structure the show.

00:22:55   If you want to let people know

00:22:57   that the incredibly popular and beloved

00:22:59   Subtlety and Subtle Sweater is back

00:23:01   for its one time a year sale,

00:23:04   to leave that 90 minutes into the episode,

00:23:07   we are doing a bad job selling our products.

00:23:10   So if you go to cortexmerch.com right now,

00:23:15   you will find until November 8th, 2022,

00:23:20   a limited time sale of the subtlety and subtle sweater.

00:23:24   We are bringing back all of our beloved colors,

00:23:27   the original blue, black, green, and red.

00:23:31   We are adding a new color this year, gray.

00:23:35   - The best color.

00:23:36   - Now I am really into this.

00:23:39   So we did some general merch at Relay

00:23:42   and we added a bunch of colors for some stuff we were doing,

00:23:45   including just a light gray sweater option

00:23:48   for one of the shirts that we did.

00:23:49   And as soon as I got it, I was like,

00:23:51   oh God, I need this in a subtlety.

00:23:53   Because that light gray color,

00:23:56   it's just like the traditional sweater color, right?

00:23:59   So we now have that available

00:24:01   and look so good with the light blue stitching.

00:24:04   So that's available in tees and in sweaters

00:24:08   with all of the other colors,

00:24:10   red, green, blue, black, and gray.

00:24:13   You can get any of them.

00:24:14   This year gray,

00:24:15   I am going to be replacing my original blue sweatshirts,

00:24:19   I've decided.

00:24:20   - Ooh, okay.

00:24:21   - 'Cause I mean, I've had those for like four years now

00:24:24   or something, so I'm gonna get some new ones.

00:24:26   - You're probably right.

00:24:27   From where I'm recording right now,

00:24:28   I can look into my closet and,

00:24:30   this is not an exaggeration,

00:24:31   one third of my closet is subtle sweaters, subtle jeans.

00:24:36   (laughs)

00:24:38   - It's the same for me.

00:24:39   I'm wearing a green tee, the green sleeper hit.

00:24:43   The green is so good.

00:24:44   I'm buying a bunch more green.

00:24:46   - Green looks great.

00:24:46   - Like this happens to me every year.

00:24:48   We get a new color, I buy one of each,

00:24:49   and I'm like, "God damn it, I wish I put more of them."

00:24:52   And then I also have to wait a year to get more of them.

00:24:56   But that green, so good.

00:24:58   - For me actually, surprisingly, it was the red.

00:25:01   Like I bought some of the red just like,

00:25:03   "Oh, it's good to have a complete set for me."

00:25:06   I wear it a surprising amount,

00:25:07   and I would never have picked that as a color for myself.

00:25:09   But yeah, the subtle sweaters and the subtle tees,

00:25:13   They're seriously so comfortable.

00:25:16   It's we get just a ton of positive feedback from people who really like them, which is

00:25:21   part of why we realized we've got to break into the show and remind you, "Hey, they're

00:25:27   on sale.

00:25:28   If you want them, you need to get them now.

00:25:31   People love them.

00:25:32   It's just this one time sale.

00:25:34   So go and order them right now."

00:25:37   Yes.

00:25:38   I cannot impress upon enough that you do this.

00:25:40   We will not have another episode come out to remind you to do this before the sale is

00:25:45   over.

00:25:46   It's a three week sale from when the episode goes out.

00:25:49   November 8th is when it's done and it's gone for a year.

00:25:53   So we only do this once a year.

00:25:55   So if you want them, you need to go and get them at cortexmerch.com.

00:26:00   Cannot impress upon you enough as well how good these things look.

00:26:02   Like a couple of days ago I was wearing the red one and I was walking towards a glass

00:26:06   door and I was like, "Damn that looks good."

00:26:10   It's just a good, our brand on a shirt like that

00:26:14   or on a sweatshirt like that, it just looks so professional.

00:26:18   I'm so happy we did this as a thing.

00:26:21   They are so great.

00:26:22   - I also feel like I'm trying to do the impossible thing,

00:26:25   which is it happens every year that when the sale is over,

00:26:28   we get contacted from people who are going like,

00:26:30   "Oh, I want the shirts, how can I buy them?"

00:26:33   And I'm like, "I'm trying to talk to you right now,

00:26:36   "the person who's going to be sad in five weeks

00:26:39   when they can't get them. You need to do this now. We're still going to be talking about AI later,

00:26:46   but know if you ever want these shirts, which you definitely do, if you want to wear the most

00:26:51   comfortable sweater you have ever worn, go to cortexmerch.com right now and get yourself some

00:26:59   fantastic clothing to wear. You will not regret it. Everybody loves these things and they're

00:27:04   sad when they're not on sale. So, CortexMerch.com, pause the podcast right now.

00:27:10   Or let me give a secondary thing. If you're like, "But Gray, Myke, I'm driving!"

00:27:15   Right?

00:27:16   Here's what I'll tell you you can do. Imagine in your mind now something you see at the

00:27:21   end of your commute. It might be a billboard, it might be like a sign in your parking space,

00:27:26   it might be your garage door, whatever. Imagine that thing right now, tie in your brain the

00:27:32   image of that thing with cortexmerch.com. Think about it right now. Say it in your mind

00:27:37   a few times. So when you arrive at your destination, you have set yourself a reminder.

00:27:42   Look, Myke's trying to pull some fancy memory palace stuff over here. You know, my method

00:27:48   is much more direct. I just say, "Hey Siri, remind me in three hours to go to cortexmerch.com."

00:27:55   Hey Google, remind me in three hours.

00:27:58   (laughing)

00:28:00   Hey Google, remind me in three hours

00:28:02   to go to cortexmerch.com.

00:28:04   (laughing)

00:28:06   We're breaking into your life now.

00:28:08   What are you gonna do?

00:28:09   (laughing)

00:28:11   Cortexmerch.com.

00:28:14   - Okay, and now back to us in the past talking about AI.

00:28:17   - Are we gonna have that long-term conversation?

00:28:21   - Yeah, yeah, yeah, we will, we will.

00:28:23   But like, I think there's also just leftover from last time,

00:28:28   there's still just like a bunch of stuff in the near term

00:28:31   that might be worth revisiting.

00:28:34   One of the other things that came up last time is,

00:28:38   how do these models work?

00:28:40   Like how do they even begin to start creating anything,

00:28:44   whether it's poetry or videos or computer codes,

00:28:48   like how does this work?

00:28:50   And fundamentally, these AI systems are made by just hoovering up a ton of information

00:28:57   in the relative domain and feeding it into the system for the system to be trained upon.

00:29:03   And I do think one of the most concerning short-term questions about that is like,

00:29:11   what does it mean to use the public work that people have done, whatever that is?

00:29:19   you've written a book and you've published it. You've gone on to Stack Overflow and you've helped

00:29:24   answer hundreds of people's questions about computer code. You've been on DeviantArt for

00:29:30   years and you've made images and like that stuff has been sucked into a computer somewhere so that

00:29:36   it can then produce imitations or produce new work based on what you have done. I think that's like a

00:29:44   That is just a really difficult question.

00:29:47   - Human inspiration.

00:29:49   So, one of the things that a lot of people brought up

00:29:52   is how is this any different to being inspired

00:29:56   by someone's work and creating your own work?

00:29:59   I feel like it is quite different, but what do you think?

00:30:03   - I think it's different, but it is hard to articulate

00:30:13   why in a coherent way?

00:30:16   - I feel like I have something that I think is pretty core

00:30:19   to me, but I also, I know a lot of people don't agree

00:30:23   with it.

00:30:24   - What is that?

00:30:25   - The difference to me is the skill required in acting

00:30:29   on the imitation is the thing that I actually think

00:30:33   is valuable.

00:30:34   - What do you mean by that sentence?

00:30:36   - Let's imagine we'll go with painting, right?

00:30:40   It's just a simple thing that we can all understand, right?

00:30:42   how somebody paints a picture.

00:30:45   If you look at a painting,

00:30:49   we'll talk about the previously famous Parisian art piece,

00:30:53   the Mona Lisa, right?

00:30:54   - Mm-hmm.

00:30:55   - And you wanna make your own Mona Lisa.

00:30:57   To be able to take the inspiration from that piece

00:31:01   and do it yourself, it is an imitation of previous work.

00:31:04   But you had to do it, you had to practice

00:31:09   and get the skill and build up your own level of skill

00:31:13   to perform that work.

00:31:16   Now, yes, all you have done is imitate it,

00:31:18   but if you get even 50% close,

00:31:21   you've made something that's interesting.

00:31:22   You've now learned the skills that you can go out maybe

00:31:26   and produce your own work,

00:31:27   but you've built the actual skill, the practice.

00:31:31   You've built the skill.

00:31:33   That is what I think is the thing that concerns me most

00:31:37   about this type of work.

00:31:39   is that I worry that the skills will get lost.

00:31:43   And I accept that some people do not value that

00:31:48   the same as me, but that's where I come from with this,

00:31:52   where I think that there is an inherent humanity

00:31:56   in these mostly inconsequential skills

00:32:01   that we value important as humans.

00:32:04   Practices, traditions, all of this kind of stuff

00:32:09   I hold those kinds of things dear and my concern with this a lot of this stuff is

00:32:16   we may lose this part in larger numbers like we may lose this part of our

00:32:24   humanity to more people if the creation of art is so simple. Like for example I

00:32:30   saw a comment on our YouTube video this morning when we put up and this

00:32:36   commenter had said that one of the things that they love about the idea of AI art is

00:32:41   that there are movies that people want to exist that don't currently exist and they

00:32:49   can type, they would maybe in the future be able to type a prompt into an AI art generator

00:32:55   and it would create that movie for them to watch. Honestly, I can't think of anything

00:33:02   worse than that. Why? Because there's no art in there, there's no passion in there,

00:33:08   there's no drive from a creator, from everyone involved in the creation of a

00:33:14   movie to come together and work towards something good. Now you may sit and think

00:33:20   to yourself, "Myke I don't agree with what you're saying right now" and that is

00:33:23   perfectly fine but I just want people to understand from my perspective the

00:33:28   The creation of art is as important, if not more important, than the piece that is at

00:33:34   the end.

00:33:35   And I think if all we end up with is just a bunch of pieces at the end, we will lose

00:33:39   so much of the humanity in this work.

00:33:43   The ideas that somebody might have that sparks off something to create a different shot in

00:33:48   this way.

00:33:49   Like that is what I actually hold to be so important to who we are as the human race,

00:33:56   rather than just here is media to consume.

00:34:00   So I don't know.

00:34:01   I don't know if I'm expressing myself clearly,

00:34:05   but I just want people to understand

00:34:07   that the thing that I care about is the creation of the art.

00:34:12   And it's not even just about jobs.

00:34:15   I just worry that we will lose this part of who we are.

00:34:20   And one of the things that makes us different

00:34:25   to every other species on this planet

00:34:28   is this kind of thing that we do,

00:34:32   sometimes for pleasure, mostly for pleasure.

00:34:34   There is business in it, but people like to make things

00:34:38   because they just like to make them.

00:34:40   That's something I find to be so beautiful

00:34:42   and I don't get the same sense of pleasure

00:34:46   out of typing six complex sentences into a text field

00:34:50   to then look at an image.

00:34:52   Maybe I'm old-fashioned.

00:34:55   If I'm trying to summarize your position, because I often have a hard time when

00:35:01   people use language like it's a fundamental part of our humanity.

00:35:06   Like I'm not, I'm never quite sure what that means, but I guess I'm trying to

00:35:11   summarize your position as like, you think it is just a fundamental good that humans

00:35:16   are producing art and that part of that process is the skills that are required to be learned

00:35:26   in order to make that art.

00:35:28   - I think that there is an importance in it.

00:35:30   I can't tell you why, but it just feels, I don't know, there's a lot of emotion in it.

00:35:37   And I just want people to understand that I'm not sitting here like, "Oh no, my job's

00:35:42   gonna go away."

00:35:44   Because realistically, it's not going to in my lifetime.

00:35:47   I feel pretty confident about that.

00:35:49   We'll get to why in a little bit.

00:35:51   I have an example.

00:35:53   I think we'll kind of like clear up why I'm not concerned about my own job.

00:35:57   I'm concerned more about creativity as an idea, something that I care greatly about,

00:36:05   about people being creative, even if it's just for fun.

00:36:10   And I'm just not sure that I like this idea of creativity will just ultimately become

00:36:18   the same thing, which is how good can you be at writing a prompt?

00:36:22   Like that doesn't feel creative to me.

00:36:25   The act of the act, the process doesn't feel like it exists to me in the same way anymore.

00:36:32   I don't know.

00:36:33   I agree. I find it strange this argument that the new artistry will be in creating the prompts.

00:36:40   There's something I don't know there's something very odd to me about that argument.

00:36:43   It kind of reminds me of a while ago when AI systems started becoming the best

00:36:50   chess players in the world. There was this what to me always seemed like an absolutely bizarre idea

00:36:56   that the chess that there was I forget what they called it, but it was it was this concept of like,

00:37:03   "Oh, the best chess player will be a hybrid chess player, that it will be a person who is being

00:37:09   advised by the computer." And there was a period in time where that was true, that a tag team of a

00:37:15   human and a computer could beat the best computer and they could beat the best human. That always

00:37:21   seemed to me like a strange artifact, like this won't exist forever. This just happens to be the

00:37:25   situation right now, but I see no reason why the computers won't just ultimately outclass the human

00:37:32   And the human will just be like a monkey, right, adding absolutely nothing to this incredibly

00:37:37   complicated game that's taking place.

00:37:39   And it's not the best comparison, but I feel that there is something in this concept that

00:37:46   people have of artistry will be the prompts that's the same.

00:37:51   If the computers are getting so good at text, at the same time that computers are getting

00:37:56   so good of interpreting text to create art, why will they not just meet?

00:38:00   Yeah, maybe that's what it is, is it's like, why do you think you will be the best

00:38:07   at coming up with the sequence of words that generates the most interesting art?

00:38:12   Yeah.

00:38:13   I just don't—

00:38:14   Because similarly, it's the same argument of like, "Oh, the computer will make better

00:38:18   art than any human ever can," right?

00:38:21   Like that's kind of the thinking of like, "Oh, or just as good or good enough."

00:38:24   Then why do you think computers won't be just as good as you are at creating prompts?

00:38:29   Yeah.

00:38:30   maybe that's, you've kind of sharpened it up there. I find that argument strange. And even if,

00:38:37   let's say that for whatever reason that was never – it turned out that just that wasn't true.

00:38:41   It's built into the laws of the universe that humans are just great at writing prompts in ways

00:38:47   that machines will never be, which I think is strange. But let's just say it was true.

00:38:50   At least in my experience of playing around with Diffusion Bee, I just, I agree with you. I don't

00:38:57   I don't think there was really anything creative

00:38:59   about what I was doing.

00:39:00   - It's problem solving, it's creative problem solving,

00:39:03   but that's like a completely different thing

00:39:05   to what I care about, which is the process of practicing

00:39:10   and getting good at a thing.

00:39:12   Like I think that that's really important.

00:39:14   I think I'm gonna do it again.

00:39:16   I think it's part of the human experience, I do.

00:39:17   And it doesn't need to be that everybody becomes a painter,

00:39:20   but we all have these things in our lives

00:39:22   that we practice and get better at, right?

00:39:25   It's like, why don't I just give my game controller

00:39:28   to a robot and then just watch what it does?

00:39:31   - Hmm.

00:39:32   - That's where we're going.

00:39:33   It's like "WALL-E", right?

00:39:36   This is how we get to "WALL-E".

00:39:37   This is how "WALL-E" happens,

00:39:40   that we're all just sitting in these chairs.

00:39:42   Like that's, right now is the beginning

00:39:45   of the path to "WALL-E".

00:39:47   - Well, "WALL-E" if we're lucky, but yes.

00:39:49   (laughs)

00:39:50   - You know, it's just like,

00:39:52   I feel like we have to do things.

00:39:54   whatever it is, some kind of thing

00:39:57   that we enjoy the process of.

00:39:59   And I feel like if all we're doing is saying

00:40:02   that this stuff is just gonna replace filmmaking,

00:40:05   it's just like, I don't wanna watch those movies.

00:40:08   I really don't, but I like to be sold in something.

00:40:11   I like to believe that a human was involved

00:40:13   in the endeavors that I'm consuming.

00:40:15   Like, but again, maybe I'm old fashioned

00:40:18   and I'm fine to accept that,

00:40:20   but I'm just trying to get across

00:40:22   like why these things are important to me,

00:40:24   why I am so passionate about it.

00:40:29   Yeah, I sort of have a minor point and a major point

00:40:33   on that topic.

00:40:34   I think last time, I made some offhanded remark of,

00:40:37   like, very soon in major productions

00:40:40   where we're going to see stuff that's AI generated

00:40:43   and not know it.

00:40:45   And I didn't realize at the time,

00:40:47   but like, oh, that was already true.

00:40:49   Same.

00:40:50   Didn't know about this.

00:40:51   Yeah, I had watched, uh, the Obi-Wan Kenobi show on Disney and I didn't really think about it,

00:40:58   but when I had watched Rogue One, I was very aware of listening to James Earl Jones do Darth Vader's

00:41:05   lines of like, "He's still Darth Vader, but he's getting too old for this." Like, you can just hear

00:41:12   it in someone's voice that they're just older. That's just what happens, uh, as your vocal

00:41:16   chords physically change. And when I watched Obi-Wan Kenobi, I never thought about it.

00:41:21   And I also didn't think about the fact that I didn't think about it. And I realized, since

00:41:26   we recorded that episode, oh, I didn't think about it, because all of his lines were done

00:41:31   with AI. There's a program called Respeacher that will take a voice actor's lines and redo

00:41:39   them in the voice of someone else. And what's really quite remarkable about it is it is

00:41:46   isn't just what you sort of think like voice modulators on the phone like, oh, I you know,

00:41:51   I can increase or decrease my pitch, but it still sounds like me. Everybody has vocal

00:41:56   tics and things like no, no, no re speech or will put in the vocal tics of the other

00:42:01   person like it's not just making your words sound in their voice. It's making them sound

00:42:08   like them. Just as an interesting thing to note, though, I did I did go back on YouTube

00:42:13   and I watched some of the original line deliveries of James Earl Jones in the original Star Wars.

00:42:17   I was like, "Ooh, he is better here!"

00:42:20   Like he has like funny little things that he does with a bunch of words that make those

00:42:23   line deliveries really great.

00:42:25   And there isn't as much of that in the Obi-Wan Kenobi show.

00:42:28   I will say when I watched Obi-Wan, there was something about Darth Vader's performance

00:42:33   that I didn't like.

00:42:35   And I assumed it was because James Earl Jones was getting older.

00:42:40   But it turned out it wasn't the case.

00:42:42   But there was something that felt missing to me,

00:42:45   and I just put it down to,

00:42:46   and I remember saying to Adina at the time,

00:42:49   "I can't believe they're still getting this guy to do this.

00:42:51   Why don't they just get like a voice actor?"

00:42:54   (laughing)

00:42:55   - Right, leave James Earl Jones alone.

00:42:57   - Yeah, well, it was just kind of just like,

00:43:00   this isn't gonna last forever with him,

00:43:03   so like they need a path.

00:43:05   And they created one, and it did a fine job,

00:43:07   but it still felt like it was missing something to me.

00:43:10   I can't put my finger on what it was.

00:43:11   It just didn't feel right.

00:43:13   I don't know if this was part, whatever.

00:43:14   But what I will say, this particular implementation of AI,

00:43:19   I'm fine with, because there are practical reasons for it.

00:43:22   Of like, if we want Darth Vader to sound the same,

00:43:26   which I think ideally we do, right?

00:43:28   Like you could just get someone else to do it,

00:43:29   but I would prefer it if he still sounded

00:43:31   like James Earl Jones,

00:43:32   'cause it's like an existing character.

00:43:34   And James Earl Jones signed this away

00:43:37   as an individual while still alive.

00:43:39   - Right, which is key.

00:43:40   you can do this, you can take my voice and do whatever you want

00:43:44   plus working voice actors continue to get jobs

00:43:48   being Darth Vader then they change the voice. For me there are enough

00:43:52   pieces of this puzzle where I'm like I am fine with this

00:43:56   because people are involved in it. It's if Disney just were like

00:44:03   we've just decided we don't want to hire him anymore

00:44:06   and we've created a thing and we're just gonna type

00:44:10   some text in and the AI's gonna spit out

00:44:13   the Darth Vader lines, I will be like,

00:44:15   "Mm, I don't like that, that doesn't feel good to me."

00:44:18   But the way they have done this whole thing going around,

00:44:21   I'm kind of fine with.

00:44:23   Like in the same way that I do find it kind of funny really,

00:44:26   that they continue to have Mark Hamill on set

00:44:30   during the Mandalorian stuff to be Luke Skywalker.

00:44:35   But then they completely digitally replace him with the younger version of him.

00:44:39   Well they do multiple things. They have him do a thing, then they have a look-alike actor

00:44:43   do a thing, and then they put a digital recreation of his face on the young actor's face.

00:44:48   "Really there is no point in him being here, right? We just like to have him around? I

00:44:53   don't know, right?"

00:44:54   But like, there is something funny to me about that, but at the same time, I'm like, at

00:44:58   least it is a respect of the person they are digitally recreating. Everyone is in on this

00:45:04   that's fine. You know?

00:45:07   I suspect that they're also using re-speacher to make Mark Hamill's voice sound like young

00:45:12   Mark Hamill, because that's again, like, he does not sound like he did when he was in

00:45:15   his 20s.

00:45:16   But he also doesn't sound like him now either, right? So they're doing something.

00:45:20   Yeah, something is happening there. But this is also one of these key things of, like,

00:45:24   okay, it's interesting to realize I had already watched an entire show where a major part

00:45:28   of it was AI created and I didn't even notice, or it didn't even cross my mind. But it also

00:45:34   touches on like what you're saying here is some of the key differences between what's

00:45:40   happening in different parts of the world.

00:45:43   And a lot of the AI art developments on the internet are just like a like a crazy Wild

00:45:49   West where people are just grabbing whatever they can.

00:45:52   And it's like, that makes me extremely uncomfortable.

00:45:55   Like, okay, so again, I will just say in general ways, I saw a demo of a Siri like voice assistant

00:46:03   that was significantly better than things that currently exist.

00:46:07   Oh, that's— it was very interesting, but it was also using the voice of a famous actress

00:46:14   just in their demo mode.

00:46:16   And I got very uncomfortable about that, 'cause it's like…

00:46:21   She's a— she's a person who doesn't know that you've made a machine that can make her

00:46:26   say anything.

00:46:27   - Waaay!

00:46:28   Hooray!

00:46:29   That's what we want!

00:46:31   - Right?

00:46:32   Yeah, and it's like, "No, no, no!" And so a lot of the AI art stuff just feels like that,

00:46:37   like it's just grabbed everything that exists. And the thing that I keep thinking of is,

00:46:44   like it makes a marionette of everyone who's ever put out any kind of art in the public.

00:46:52   And that's like a horror, right? That's just a completely horrifying concept that,

00:47:00   if you have a social media timeline where you've posted stuff, guess what?

00:47:05   There's enough information where someone can make a marionette of you, and it can do anything.

00:47:10   And right now, like, this can happen without someone's permission.

00:47:13   And yeah, it's totally why, like, the Darth Vader stuff I have no problem with,

00:47:17   because presumably James Earl Jones was, like, fantastic.

00:47:20   - I saw an article with that. That was, like, I read about that.

00:47:22   It was the case that he approved this, signed the rights, like, he's good with it.

00:47:26   There's a big distinction between giving permission for something like this to be done

00:47:32   and just the power of AI to make a marionette of anyone.

00:47:39   And there's something really horrifying about that.

00:47:42   Now I think it becomes, like other people brought this up as a kind of question, there's

00:47:48   a boundary that's crossed between doing it to a living person, which is super bad.

00:47:54   It's like, yeah, I just think that's really awful and immoral. It's less bad when the person's dead,

00:48:02   but it can still be bad if it's recent. I don't quite know what recent means in this context, but

00:48:09   clearly when we get to a point where we're talking about modifying Leonardo da Vinci's art,

00:48:18   I don't think there's anything bad there about like making a marionette of Leonardo da Vinci,

00:48:23   Making a machine that can spit out a ton of Leonardo da Vinci style paintings

00:48:27   I don't think it's possible to ascribe some kind of rule onto this

00:48:31   You just feel it when you feel it like I can't tell you what the right time period is

00:48:37   and it's also different in every case and

00:48:39   You know goes on and on forever like

00:48:44   The fast and furious movie right where Paul Walker had just died, right and it was like months later when the movie came out

00:48:53   out but they used some digital recreation of his face to be able to

00:48:57   give him like a tribute and send-off. And I think that there is a

00:49:02   little like it's awkward I think that movie came out too soon really but the

00:49:08   whole family kind of seemed to agree and there's something about like that was

00:49:12   the thing. Same as like Princess Leia right? Yeah that's what I was thinking.

00:49:16   Princess Leia's send-off was the next one after Carrie Fisher had died.

00:49:20   Whereas still I feel like we're on a fine line here because it's weird.

00:49:24   But it was recreated from stuff that she did participate in shooting and at least there's

00:49:29   like a tribute.

00:49:31   But then if like Princess Leia just continued being a character after that, be like no you've

00:49:35   pushed it too far now.

00:49:36   Right?

00:49:37   Because now that's the making of the marionette right?

00:49:40   She's dead now but you're continuing to use her.

00:49:43   That would be weird.

00:49:46   I don't think it's possible to put rules on the death of a person. I just know that I

00:49:52   don't find it particularly comfortable for someone to profit off, either financially

00:49:59   or publicity-wise, somebody who is dead. Like taking their actual work, and again, divorcing

00:50:07   the work required in the copying, right? And just straight up, like, "I took this thing

00:50:14   and I made this thing out of it.

00:50:16   And I didn't really have to do anything

00:50:19   other than put this image into the AI.

00:50:23   There's uncomfortableness for me in that.

00:50:25   - Yeah, I just wanted to highlight that there's a...

00:50:28   Like with many discussions, you can run into areas

00:50:33   where there's some kind of spectrum.

00:50:34   - Yeah, like someone taking the works of Shakespeare

00:50:37   and putting that into an AI and making more Shakespeare,

00:50:39   I don't care about that.

00:50:40   But I can't tell you why I don't care about it either.

00:50:43   I just don't.

00:50:44   Yeah, I mean, I do want to also add that this, this dance is around a concept

00:50:49   that I've been, I've been thinking about for years and I have, I have a hard time

00:50:52   articulating, but for now I will just call it as the importance of saying no in art.

00:50:58   And I think a lot of artistic projects or the life of creative people are defined as

00:51:08   much by what they did as what they didn't do.

00:51:10   A thousand no's for every yes.

00:51:12   Yeah, there is a way to put it, but I think that having spoken to creative people, a lot

00:51:19   of them have some internal set of things that they don't do that's not obvious in the

00:51:27   work that they create, but it lends a kind of character to their work.

00:51:31   I've also just recently went through this where I killed a what was going to probably

00:51:37   be like a 20 or 30 minute video that was all storyboarded and written out.

00:51:42   And part of the reason I killed it is because I realized this was a project to say no to,

00:51:48   that like, yes, I could put it out and I think people would like it, but it violated a couple

00:51:54   of my own internal, like, I don't want to do this sort of thing in this way, even if

00:52:01   no one in the audience would have noticed.

00:52:05   As like, man, that was a hard decision.

00:52:06   But I think that that's what matters.

00:52:09   And so even while I think that there is there's nothing morally wrong about making new Shakespeare

00:52:16   plays or new Leonardo da Vinci paintings with AI, I do think that there's something about

00:52:23   this concept of making an artist do anything that you want demeans the limits that that

00:52:35   person put on their own things.

00:52:37   - And I'm also like not sure why you even want it.

00:52:41   - Well, I think this is also just a side effect

00:52:43   of how easy it can be and will be to create this stuff.

00:52:46   Where just like out of curiosity, you type some prompts in.

00:52:50   I guess-- - Curiosity is different

00:52:52   though, right? - But I mean, it's like

00:52:53   when this gets good enough that it can just like

00:52:55   make a movie where some things happen.

00:52:57   I don't know, this idea of the importance of saying now

00:53:00   is also connected to a thing that I see with a lot of,

00:53:02   particularly TV shows and long running creative projects,

00:53:07   where I think of this as you can see the audience influence the creators that never works.

00:53:14   And there's lots of shows where it's like, "Oh, this is the moment. I can see now that the

00:53:19   creators are aware of the audience." And with creative projects, the audience will ask for

00:53:27   things that they want that they don't realize in the long run actually destroys or makes worse the

00:53:34   thing that they want. And like that's sort of the I think that can sound super snobbish,

00:53:39   but I think it's true. And I've seen it in enough projects was like the audience wants

00:53:42   something and the creator then goes, Oh, I'll do that thing and the audience responds and

00:53:48   you get into this little positive feedback loop of doing the things that the audience

00:53:51   wants and none of those decisions mattered individually, but cumulatively they can make

00:53:57   a thing much worse in a way that's hard to pin down. And I just think there's something

00:54:02   in AI art where even when there's no problem, I think people can like destroy the things

00:54:10   that they love because there's no one to say no. There's no creator who says, "These

00:54:16   are the limits of my thing." And again, I keep thinking of the Miyazaki movies as a

00:54:22   particular example of this. Like if you can make a Miyazaki movie about anything you want,

00:54:28   It kind of destroys what those movies are unless moved by the argument about it's important

00:54:34   for humans to do these sorts of things.

00:54:37   But I can easily imagine a situation where even if the AI is able to make amazing art,

00:54:44   it's actually kind of worse for everyone involved, even though it's a thing that the audience

00:54:50   has asked for or that people go like, Oh man, I wish I could continue the series of movies

00:54:55   forever and I can do it by typing into the machine and it will make the movie for me.

00:54:58   It's like you'll ruin the thing you love by doing that, by getting what you want all the

00:55:03   time without a creative mind to say like, "No, this is the limits of this project" or "No,

00:55:10   this is when the story stopped."

00:55:11   Yeah, I mostly agree with that statement. Like, we've spoken about this a lot, right?

00:55:17   Like that idea of the point where the artist becomes aware of the audience. We speak about

00:55:24   this a lot in regards to like just personally TV show recommendations and like you know

00:55:29   like I know you are very sensitive to this like it's something that I'm familiar with

00:55:33   like you point this out and I'm like oh okay this is an interesting idea for why the show

00:55:37   may have gotten bad in that season like it's not really something I would have thought

00:55:40   of and I will say that I mostly agree with your thinking here but not completely I think

00:55:48   my kind of interpretation of this idea that you have is like, just to add the word like,

00:55:55   mostly to it.

00:55:56   Yeah, yeah.

00:55:57   Right?

00:55:58   Like that I don't, and I'm not saying that you say this, but the way that it sounds makes

00:56:01   it sound like a definitive, like that listening to the audience or seeing what the audience

00:56:07   want can never make something better, but I don't think it's that way, and I'm not sure

00:56:12   you think it's that way, but I just want to state that, right?

00:56:15   Where like-

00:56:16   Yeah, yeah.

00:56:17   there is definitely feedback which helps make something better and I do this all

00:56:23   the time but there are sometimes people will ask for me to talk about a thing

00:56:28   and I just know it won't be interesting because I'm not that person to do that

00:56:35   right and that that tends to be a lot of it for me of like you're telling me you

00:56:39   want a thing I know you won't want it because I know it's not gonna be

00:56:43   interesting because either I don't find it interesting or I don't have the

00:56:49   knowledge or whatever it might be you know like that there is that part of the

00:56:53   artist of oh god I just called myself an artist there is that part of the

00:56:58   creative person right where they are aware of what they're good at and what

00:57:03   they're not and what they think will be ultimately enjoyable for this type of

00:57:08   stuff that they create yeah and then they go out and make it obviously like

00:57:12   this is a different kind of thing.

00:57:14   I also think it's the bounds of the project.

00:57:18   And I found, I just recently stumbled upon to me what is just the most perfect

00:57:24   example of the artist saying no, which is also a little bit heartbreaking for

00:57:30   reasons that will be obvious when I, when I mentioned who it is, J.R.R.

00:57:33   Tolkien.

00:57:34   I just discovered, I can't believe I never knew this.

00:57:37   He started a book that was set after the Lord of the Rings.

00:57:42   And so he wrote the first couple of chapters of this story.

00:57:48   And it's quite interesting, but like he,

00:57:51   basically he then wrote about,

00:57:56   he wrote in one of his letters,

00:57:57   you can get these books of all of his private letters

00:57:59   where he discusses a bunch of his thoughts.

00:58:01   He wrote about why he didn't finish this.

00:58:04   And the reason he didn't finish it

00:58:06   is because, oh, in his timeline after The Lord of the Rings,

00:58:10   there's not really much magic left in the world.

00:58:14   He had an idea of who the story would be about,

00:58:16   but basically everybody magic is gone,

00:58:18   and it's just the world of men.

00:58:20   And he said, "Oh, I could have written

00:58:22   "a totally fine adventure book,

00:58:26   "but that's not what the world

00:58:28   "I was trying to create was about."

00:58:31   He was trying to create this mythology.

00:58:34   And writing another story that's what happens

00:58:38   after the mythology, he was like,

00:58:39   "Wait, why am I doing this project?

00:58:41   This isn't really what I want to do.

00:58:43   Sure, we can continue what happens

00:58:44   with a bunch of these characters, but I'm gonna say no."

00:58:47   And so he just like stopped writing it and said,

00:58:49   "No, I'm not gonna write anymore

00:58:51   the stories of what happens after.

00:58:52   I'll just fill in some of the details

00:58:54   of what happened before."

00:58:55   And I think that's an amazing example

00:58:57   in reading his letter about it.

00:58:58   He's very self-aware.

00:59:01   Sure, the audience would love this.

00:59:03   Everybody's asking for like a sequel to Lord of the Rings because it was so great,

00:59:07   but it's also so great because he knew where to stop.

00:59:10   I just think that's an interesting example and it would just make me very sad if people

00:59:15   were typing into AI generator machines like, "Give me Tolkien's next book!"

00:59:20   Like, "Oh no, please don't.

00:59:22   Please don't do that.

00:59:23   It's bad."

00:59:24   This episode of Cortex is brought to you by Squarespace, the all-in-one platform for building

00:59:29   your brand and growing your business online.

00:59:31   You can stand out with a beautiful website, engage with your audience and sell your products,

00:59:36   services, the content that you create, whatever it is, Squarespace has got you covered.

00:59:41   No matter what type of website you're looking to make, whether it's something for a personal

00:59:45   event, maybe you have a business site that you want to create, it doesn't matter, Squarespace

00:59:49   can help you.

00:59:50   They've got you covered.

00:59:51   And it starts with one of their beautiful templates.

00:59:54   They're best in class, they're award winning.

00:59:57   It is as easy as browsing the type of site or category of business you want to make a

01:00:01   website for and that will give you the perfect starting place for you to then customise just

01:00:05   a few clicks and really make it feel like your own.

01:00:09   And then once your website expands and maybe you want to add that online store, you can

01:00:13   sell whatever you want, physical, digital goods, it doesn't matter.

01:00:17   They have all of the tools that you need to start selling online.

01:00:20   And then as your website grows and your visitors start pouring in, you can use insights to

01:00:25   grow your business.

01:00:26   If you want to know where your site visitors are coming from, where your sales are generating

01:00:30   from, which channels are most effective for you, you can analyse all of this in Squarespace.

01:00:35   Then when you have the data, you can improve your website and build a marketing strategy

01:00:39   based on your top keywords or the most popular products and content you have available.

01:00:43   I love Squarespace. I have been using Squarespace for nearly 15 years now for various projects

01:00:50   and I wouldn't go anywhere else. When I want to put something online, they're the first

01:00:54   place I go. I know how to use it, it's so easy, everything looks great on every type

01:00:58   the device, they have awesome apps for you to manage everything. It's the full package,

01:01:03   I love it and I'm confident you're going to love it too. But you don't have to just take

01:01:07   my word for it. Go to squarespace.com/cortex and you can sign up for a free trial, no credit

01:01:12   card required. You can go in and build your whole website. Then when you want to launch

01:01:16   it to the world, use the offer code "CORTEX" and you will get 10% off your first purchase

01:01:20   of a website or domain. That's squarespace.com/cortex and then when you sign up, use the offer code

01:01:26   Cortex to get that 10% off your first purchase and show you support for the show.

01:01:30   Our thanks to Squarespace for their continued support of Cortex and Relay FM.

01:01:34   Hey, do you want to talk about this podcast that you found here, Myke?

01:01:38   So I referenced why I'm not worried about my job specifically. This has been going around

01:01:44   a lot. It is an AI Steve Jobs on an AI Joe Rogan episode. And it is a full podcast created

01:01:54   by an AI. I've not really bothered to look into this because I actually think it might

01:01:59   be part of a publicity stunt. But nevertheless, it is a thing that exists, it's 20 minutes

01:02:05   long and you can listen to it and I've like skipped around in it and you know what? It

01:02:09   really sounds like Steve Jobs. Like there are points where you can hear that it's not

01:02:13   real right? But like and ultimately my takeaway from this is who wants this? Like genuinely

01:02:19   like who wants this? Who wants to hear what Steve Jobs might have said to a fake Joe Rogan?

01:02:25   For me, I don't feel concerned about my job because you can take a version of me and a version of you

01:02:33   and we can have them make podcasts forever, but I feel like if that's what you want, I don't know

01:02:43   how much you could enjoy the content because it's not real. It's not real conversations.

01:02:48   That's what I do for a living, is real conversations between real people about

01:02:52   things that they care about and my assumption is the majority of people that listen to my shows

01:02:58   want to hear that rather than let's imagine what to ai might be talking about instead for me that's

01:03:06   just so broken from what i imagine people want the content for and look if you are that person you

01:03:14   don't need to tell me of course i know you exist right like i'm sure that there are people that

01:03:18   would like to just have us on in the background so they could go to sleep and it's just like we're

01:03:22   We're just going to have an infinite amount of episodes forever.

01:03:25   But realistically, I don't imagine that that's going to be a thing that people

01:03:32   would genuinely care about enough in the way that people might care

01:03:35   about the content that I make.

01:03:36   So just coming across this one episode, it honestly made me, even though it

01:03:42   showed me it can be done, it made me feel more secure in my own profession.

01:03:45   But honestly, in the last episode, I wasn't worrying about my own profession.

01:03:50   It was more about the idea of creativity in people.

01:03:53   That's what I get more about.

01:03:54   - You said this to me and I took a listen.

01:03:56   I did the same thing.

01:03:57   I didn't listen to the whole thing all the way through.

01:03:59   I sort of skipped around.

01:04:01   I have a slightly different take on this sort of stuff,

01:04:03   which is it doesn't matter how good Steve Jobs

01:04:07   on Joe Rogan is to listen to.

01:04:10   It's more of just a demonstration of proof of concept.

01:04:14   This thing is possible.

01:04:15   And once you have a demonstration of proof of concept,

01:04:18   guess what?

01:04:18   get better, they don't get worse. So this was just like the first real proof of concept

01:04:24   demonstration of two people having a podcast conversation. I do have to say, listening

01:04:29   to it, I think the Steve Jobs was less good, but like holy moly did it nail Joe Rogan's

01:04:36   way of talking.

01:04:37   That makes sense though, right? The amount of source material for Joe Rogan is almost

01:04:41   infinite really.

01:04:42   Yeah, I just thought like it's actually interesting to listen to because you can almost

01:04:46   hear the fact that there's what, 100 million hours of Joe Rogan talking and, you know,

01:04:52   the database for Steve Jobs is so much smaller. And you can hear that in the two voices. It's

01:04:57   like, especially because I think Rogan has a funny circular way of talking sometimes

01:05:02   that it's like, I would never not know that that wasn't him. If you told me it was Rogan,

01:05:07   but it was only just an AI Jobs, I would have believed it. But this to me is, again, a good

01:05:13   example of this thing where making a marionette of someone is just bad. It's like, "Hey

01:05:19   guys, whatever you think of Rogan, he's still a person and it's real bad to make

01:05:23   like a fake show where he's talking. And it's less bad for Steve Jobs because he

01:05:31   is dead, but it's still close enough that it makes me very uncomfortable." I just

01:05:36   think we're going to see an increasingly large number of these sorts of things where

01:05:40   people can make whatever they want and it's only going to become increasingly easy over

01:05:45   time and I don't know how this project was made in particular but yeah it's the first

01:05:50   example of oh it's an AI podcast with two people where we have enough information to

01:05:57   recreate them in some sense.

01:05:59   Yeah it's a publicist stunt for the company. The company that made this has an AI text

01:06:05   to voice generation system that they are trying to sell.

01:06:08   I mean that's not surprising.

01:06:10   But it's also interesting just how fast so much of this is being commercialized.

01:06:15   I stumbled across a – I thought, "Oh God, how brutal is this?"

01:06:21   But it's a company that makes AI people who will read quite convincingly scripts where

01:06:27   it looks like it's a talking head segment.

01:06:29   And the whole idea is, "Oh, you can have all of your corporate training material delivered

01:06:33   in this way where there's like an AI person who will talk through whatever it is you need

01:06:37   to onboard your new employees. There's something about that to me which is like, it's very

01:06:41   convincing but it also feels like what a horrible dystopian nightmare. It's like you as an

01:06:46   employee sign up with the company is like, "Oh hey, we didn't even take the time to

01:06:52   film a person going through our own training materials. We just gave it to an AI and they

01:06:57   made a fake person who you get to listen to explain your job to you."

01:07:01   Enjoy your training.

01:07:02   I know, it's just so awful.

01:07:04   We value you.

01:07:06   I just want to read from the podcast AI, their kind of description of the show.

01:07:10   "Whether you're a machine learning enthusiast, just want to hear your favorite topics covered

01:07:16   in a new way, or even just want to listen to voices from the past brought back to life,

01:07:20   this is the podcast for you."

01:07:22   - Right. - Alright.

01:07:25   - Do you want to hear our ghoulish marionettes say what we made them say?

01:07:30   - It's just like, if you are a fan of Steve Jobs, which I think is who they are pitching this to,

01:07:35   right? Why do you want to hear him talk about things he never spoke about? Like what do you

01:07:39   value from that other than just like hearing the voice? Like if you just want to hear his voice

01:07:44   just go to youtube and watch like commencement speeches or interviews or whatever. Like I don't

01:07:49   understand why you want to hear him talk about things he never spoke about. It's not his opinion.

01:07:56   There is no opinion of him in this. They've built all of this from things he said but you're just

01:08:03   taking words he used and just putting them together in a new way. It's not actually

01:08:08   his opinion. And I just I find it so strange.

01:08:15   Since the last episode, I have much more strongly onboarded the concept of trying to only read

01:08:23   and consume and listen to and watch media that you know has been produced by a human.

01:08:29   And how did that do for you with Obu?

01:08:31   You know what I mean?

01:08:32   (laughing)

01:08:33   - Yeah, it sneaks in there, but at least it's like,

01:08:36   oh, I know humans made the show.

01:08:38   (laughing)

01:08:39   - Yeah, at least you got that part, right?

01:08:41   - Yeah.

01:08:42   - Maybe not all humans, all of it all the time,

01:08:43   but like enough.

01:08:44   - Yeah, and so again, this idea that went from like,

01:08:48   that's crazy to, oh, I guess this is wise advice to live by

01:08:51   has accelerated quite quickly in my life.

01:08:54   And yeah, it's kind of a weird,

01:08:57   oh, make sure you know it's a person

01:08:59   who wrote or made this thing.

01:09:01   So yeah, I am trying to onboard that as a concept

01:09:05   because I think not onboarding that as a concept

01:09:07   is part of what is going to lead to AI doom for us all.

01:09:12   - Oh God, we still want to talk about AI.

01:09:14   All right, yep, okay.

01:09:16   This episode of Cortex is brought to you by Fitbod.

01:09:19   Between balancing your work life, your family life,

01:09:22   and everything else going on in your life,

01:09:24   it can be hard to make fitness a priority.

01:09:26   Personal fitness shouldn't be about competing with people.

01:09:30   What you need is a program that is tailored for you, working with you and not against

01:09:34   you and that's why you need FitBud.

01:09:37   You don't want to have to look to other people and do what they're doing.

01:09:40   You don't want to have to be told exactly what to do.

01:09:42   You don't want to have to do these boring things every single day.

01:09:45   You don't want that rigid structure.

01:09:47   You want something that is flexible, something that adjusts, that is dynamic, that is your

01:09:51   own fitness plan that you can have access to from anywhere. In a beautiful app of wonderful

01:09:58   HD video tutorials shot from multiple angles so every exercise you want to learn is super

01:10:04   simple. This is all why you need Fitbod. They use data to create and adjust that plan for

01:10:10   you. It's crafted to be unique for you exactly. Their app is super easy to learn. It integrates

01:10:18   with your smartwatch, your Apple watch, your Wear OS watch, it integrates with apps like

01:10:22   Strava, Fitbit and Apple Health. Everything's super awesome. I love Fitbud. It really set

01:10:27   me on a different path to my fitness journey when I started using it. What I really liked

01:10:32   was that there was a variety there. Like that variety was super important to me. Previously

01:10:37   I'd used apps and videos and stuff. It's just the same thing every day on a phone that's

01:10:40   really boring. But I was learning new things, while at the same time it's also mixing up

01:10:46   with those variety of exercises so you're not overworking or underworking anything.

01:10:50   It's really clever. Personalized training of this quality can be expensive. Fitbod is just

01:10:55   $12.99 a month or $79.99 a year but you can get 25% off her membership by going and signing up today

01:11:02   at fitbod.me/cortex. So go now and get your customized fitness plan at fitbod.me/cortex.

01:11:09   That is 25% off that you will get by signing up at FITBOD.ME/CORTEX.

01:11:15   25% off.

01:11:16   Our thanks to Fitbod for their support of this show and Relay FM.

01:11:19   [BEEP]

01:11:20   Yes, I do want to talk about Doom for us all, but partly because it's clearing up a little

01:11:26   bit of confusion from last time as well.

01:11:28   Okay.

01:11:29   So one of the things we discussed last time was stable diffusion making images like,

01:11:36   here is the filming of NASA landing on the moon

01:11:40   and creating fake stuff.

01:11:42   And I think a totally fair criticism of that

01:11:44   that I saw from people is,

01:11:46   hey, Photoshop exists, you know,

01:11:50   like we can make those images now,

01:11:52   we don't need an AI art system to generate them for us.

01:11:56   If you would be convinced by a good Photoshop,

01:11:59   what's the difference from being convinced

01:12:01   by an AI art system about something

01:12:03   that isn't real in the world?

01:12:06   So what I want to portray here is I think Doom comes in like three phases.

01:12:11   Phase one, I think will be just pollution of the public information space.

01:12:18   I think it's sort of a general confusion.

01:12:20   And what makes the difference between AI art and something like Photoshop is the scale

01:12:29   and the cost. So right now, if someone wants to try and put out a bunch of misinformation

01:12:37   or create evidence to back a conspiracy theory, there still needs to be effort that goes into

01:12:43   creating that thing. Like you were saying, Myke, you need to learn these skills as part of your art

01:12:48   for how to create a piece of misinformation to put out in the world. And I think what I kind of

01:12:56   of expect if I'm projecting a bunch of this stuff forward is the scale of this potential

01:13:03   misleading images and misleading text that were AI generated is just vastly beyond what

01:13:10   we can imagine now. And I think in the modern world, like a lot of people have been driven

01:13:16   kind of crazy just from the selection of what information they're presented. Like, you know,

01:13:26   people on social media can kind of like drive themselves crazy by just going down rabbit

01:13:30   holes and being continually presented with information that agrees with them. And that's

01:13:34   just by like, selecting the things that they're seeing that actual human beings have created.

01:13:40   But what I'm kind of thinking might happen here is that when you're able to generate

01:13:45   a huge amount of content. Just as right now, companies intentionally AB test what they

01:13:53   show you for engagement. Not even on purpose, but AI generated content will effectively

01:14:01   be unintentionally AB tested for convincing this. How convincing is this sequence of words

01:14:09   in whatever idea it's trying to spread. And I really do view a lot of this stuff as the

01:14:17   kind of concept of memes, of ideas. They evolve, and they spread, and they mutate, and their

01:14:24   spreading in the world doesn't have anything to do with how true they are, it has to do

01:14:29   with how convincing they are.

01:14:31   Imagine hooking up Twitter's algorithm to a text AI.

01:14:36   Yeah, I think that's going to happen.

01:14:37   every time you pull to refresh, it just gives you a bunch of other nonsense.

01:14:43   That is truly horrifying to me as a thought.

01:14:47   I think we're going to see that.

01:14:48   I think we're going to see that or we're going to see something very much like it.

01:14:51   I mean, look, the AI video stuff that we saw before.

01:14:55   Do you think TikTok won't start doing that the moment that it becomes engaging to people?

01:15:01   That's a really good point.

01:15:02   That's actually the most likely of all of them, I think, to occur.

01:15:06   to just automatically generate 20 second, 30 second videos every time a person refreshes

01:15:12   and just keep doing it for whatever keeps that person engaged.

01:15:16   I think we're going to see that.

01:15:18   Even if it's not the companies directly doing it themselves, you'll have entities on Twitter

01:15:23   where it's like, "Oh, it's a bot, but it's, you know, it's acting like it's a person and

01:15:27   it's just existing in the world and it's doing this kind of unintentional A/B testing for

01:15:32   convincing this."

01:15:33   and the ones that are more convincing for whatever reason will just spread better.

01:15:37   So I really think that is a kind of like, I think a lot of the criticism is people want

01:15:44   to know, like, gray, what like, you love technology, why aren't you behind this stuff? I used to

01:15:50   be a real technological optimist. But I've changed my mind on a bunch of that stuff.

01:15:55   And this is one area in particular, where, and I want to be clear here, I don't think

01:16:01   It's malicious.

01:16:02   I don't think it's necessarily that someone's out there trying to do bad, but I think the

01:16:07   ability to create hundreds of millions of memes in the way of like just a concept that

01:16:15   can spread at the drop of a hat is just bad for the public information space.

01:16:22   And the world has barely survived social media in some ways, like that has made stuff so

01:16:29   so bad with how people think about the world. And this is that just taken to the next level

01:16:34   by a huge order of magnitude. So I suspect that'll be one of the first ways that this

01:16:41   becomes obviously bad over time. Like once it progresses out of the stage of just being

01:16:49   an interesting toy. My hope is that for whatever reason, we're in the part of the technological

01:16:55   development where it looks like it's an exponential graph, but it's rapidly going

01:17:00   to level off at an s-curve, and we discover "oh, there's parts of this that were way

01:17:05   harder than we thought." That's where I hope this is going, but…

01:17:09   MATT: You show me the signs of that one on the next episode, then.

01:17:12   BRIAN well, I'm in this position where I feel

01:17:15   a little bit like "oh, I'm like a crazy person talking into a microphone about this

01:17:19   stuff. But I have spoken to some of the top people in the world in this area and they're

01:17:28   very concerned is the way that I would put it. And talking to people, I was trying to

01:17:34   tease out this concept of where do we think we are on this curve? Is this the start of

01:17:39   the exponential or are there obvious problems ahead? And the answer was pretty universally,

01:17:45   Oh, the as far as we can tell, the exponential has barely begun. Because the thing part of

01:17:51   the reason it's making such fast progress is because the the work being done right now

01:17:57   is still in the realm of Oh, hey, you get into work. And like, what's the first thing

01:18:01   you think of that could make this better? And you try it, and it makes it better. That's

01:18:05   just like an indication that we're at the start of an exponential curve. So yeah, I

01:18:09   think we're gonna have like a bunch of confusion about this stuff. I think that transitions

01:18:14   into genuine economic problems as particularly language models get better and better and better

01:18:23   at doing all the kinds of work that humans do, which is largely knowledge work.

01:18:30   You know, and it's funny, like I made "Humans need not apply" eight years ago now, I think?

01:18:37   And it's been on the back of my mind about, you know, revisiting that at some point.

01:18:41   And I was kind of thinking like, oh, I don't know how relevant this video still is. But having

01:18:48   rewatched it, it's like, Oh, no, I put it back, like to be featured on my channel under one of

01:18:52   like the most watched videos, because it's like, no, no, all this AI stuff and all of the language

01:18:56   stuff like makes this way more concerning. I think the lesson learned there has simply been that

01:19:02   physical automation is slower to progress for a bunch of reasons. But all of the knowledge

01:19:08   worker stuff is coming along very very fast. I just know that there are companies that

01:19:13   are very explicitly targeting low-level knowledge work and then will be progressing further

01:19:19   and further up the chain as fast as they can with better and better language models to

01:19:23   do all sorts of things that people can do. I understand that lots of people just fundamentally

01:19:28   disagree with me on this point of economics that jobs can't be replaced because humans

01:19:34   have infinite needs and always want more things. I understand that argument, I just don't

01:19:41   agree with it, and I think AI just breaks some of the fundamental assumptions that are

01:19:45   built into that model of as we get better machines of all kinds, we just increase the

01:19:51   quality of life and increase our desires. Like, I just don't think that that's universally

01:19:55   true. So yeah, I think we're going to end up with some really major problems in the

01:20:01   the economy, particularly in the knowledge worker part of that economy.

01:20:06   And look, we don't need to talk about it today because we've talked about it enough,

01:20:10   but I think once you start encroaching on AI systems that are good enough to replace

01:20:15   most human work, you really start encroaching on the kinds of things that can lead to the

01:20:21   extinction of the human race.

01:20:24   Maybe that's a little too heavy for today, but that's kind of my having thought it

01:20:28   through of like, what are the three phases of where does this go?

01:20:31   – confusion of the public information space, destruction of the economy, extinction of

01:20:36   the species. It's one, two, three.

01:20:39   Oh boy.

01:20:40   How you feeling, Myke?

01:20:43   Not great.

01:20:44   No?

01:20:45   No, I didn't want to do another hour and a half on this, to be honest. But I guess

01:20:50   this is who we are now.

01:20:52   Yeah. Can I just ask you though, like, how crazy does that sound to you?

01:20:55   No, it doesn't. And that's why I don't like it. I don't have optimism about this

01:20:59   this area of technology. I do not think that this is a thing that will produce much good.

01:21:05   We can leave it there for now then. On with the show. It's only been an hour and a half.

01:21:17   Just some quick follow up from the last thing we talked about on the previous episode. Okay,

01:21:21   on the lighter side of things, what is a podcast, Myke?

01:21:25   I don't know why. Why are we doing this one again, too? What is this?

01:21:30   We just relive the horrors of the previous episode?

01:21:32   I feel so badly for you, Myke. Last episode, you stumbled into a terrible mistake,

01:21:39   which was attempting to describe... but what is a podcast?

01:21:43   Well, the real issue was I put two topics together that seemed related, but they were unrelated.

01:21:50   Right, yeah. This is always the dangers of speaking

01:21:53   extemporaneously and you're just sort of like in the middle of a conversation and you say some

01:21:57   things and so anyway we ended up talking for a while about like but what is a podcast what must

01:22:03   it be in order for it to be the platonic ideal of a podcast and this this caught people's attention

01:22:09   and you can click the link in the show notes a cortex sent in what i absolutely adored is this

01:22:16   oh i saw this podcast alignment chart and i read this and was like i think i agree with absolutely

01:22:22   everything on this chart. This nails what I wanted, I think, to kind of get across,

01:22:27   but maybe didn't do a great job of doing, I don't know.

01:22:29   So in the great tradition of memes, there's a meme where people make the alignment chart for

01:22:36   various things, and I think this started with what is still the fantastic example of the sandwich

01:22:42   alignment chart. So you have two axes in the chart for the sandwich, which is ingredients

01:22:49   and structure. So it ranges from ingredient purist to ingredient rebel, and then you have

01:22:56   structure purist. A sandwich must have the classic sandwich shape, two pieces of bread

01:23:01   with toppings in between. And then like structure rebel, any kind of food enveloped in any way is a

01:23:07   sandwich. Oh, like a hot dog? Yeah, so this is like, is a hot dog a sandwich? And hot dog falls

01:23:13   on the sandwich alignment chart of ingredient neutral and structure neutral. Wait, this is a

01:23:18   - There's a very intriguing way that you've introduced this.

01:23:21   Isn't this just like the chaotic, evil, chaotic, good thing?

01:23:25   - Yes, that's where this comes from.

01:23:26   That's the origin.

01:23:27   - Oh, right. - It is like.

01:23:28   - I misunderstood you and thought you were saying

01:23:29   that the sandwich one was the origin.

01:23:31   I was like, I don't think you're right there.

01:23:33   - No, no, no, no, no, no, no, no, no, no, no, no.

01:23:34   (laughing)

01:23:37   I always forget 'cause I never played "Dea Deas."

01:23:39   - Chaotic, neutral, chaotic, good, that's the thing.

01:23:41   - No, but the two axes, it's evil and good

01:23:43   and then lawful and chaotic.

01:23:45   Is that, that's, that's,

01:23:46   those the two axes on the traditional one?

01:23:47   I think so.

01:23:48   The sandwich alignment chart is the meme, right?

01:23:51   Which is like, oh, here is from which other things birth.

01:23:54   And so what I love on the internet is people will get into an argument about like, what

01:23:59   is thing?

01:24:00   Is a hot dog a sandwich?

01:24:01   Is a pop tart a sandwich?

01:24:03   Is a chicken wrap a sandwich?

01:24:04   And someone will come up with like, here's the alignment chart to try to describe where

01:24:09   all of these things fit in.

01:24:10   And so someone did this for podcasts, the podcast alignment charts.

01:24:15   The two axes are distribution method and media type.

01:24:20   So do you wanna run through some of these, Myke?

01:24:22   Where do you fit on the podcast alignment chart?

01:24:25   - Oh man, I think I was...

01:24:28   - I feel like you were making an argument

01:24:30   for traditional distribution method,

01:24:32   traditional media type,

01:24:33   which is an audio RSS feed is a podcast.

01:24:37   - No, because I don't, I think I am a modernist.

01:24:41   - Distribution modernist, media type traditionalist,

01:24:44   - Any audio that's subscribable is a podcast?

01:24:47   - No, I think I'm modern modern.

01:24:49   - Any audio or video that's subscribable is a podcast?

01:24:52   Oh, so my YouTube channel is a podcast,

01:24:54   that's what you're saying?

01:24:55   - No, you see that's, yeah, no,

01:24:56   that I'm modern traditionalist.

01:24:59   'Cause that's the thing of like,

01:25:00   but it gets into an issue for me is that I watch videos

01:25:04   that I consider podcasts.

01:25:06   It's the any, I would see, I would phrase it as any audio

01:25:10   or video that's subscribable can be a podcast,

01:25:14   That would be my personal definition, I think.

01:25:17   Would be closer to how I feel.

01:25:20   But I think I'm more traditionalist modernist,

01:25:22   which is any audio that's subscribable is a podcast.

01:25:25   No, but then it's got audio books in here.

01:25:27   This is very complicated.

01:25:28   I don't...

01:25:29   (laughing)

01:25:30   I would say any audio or video

01:25:32   that somebody wants to call a podcast

01:25:34   that you can subscribe to can be a podcast.

01:25:36   I think that's how I actually personally sit now.

01:25:39   I think that works for me.

01:25:42   I really love these charts go kind of crazy, right?

01:25:45   I think the problem with the podcast one is a two-dimensional surface is not enough to

01:25:50   express the entirety of what it wants to be.

01:25:53   You need a third axis, which I think is something like consumption intention.

01:26:00   So for me, like, I would be like a traditionalist here, right?

01:26:03   That the consumption intention is audio only.

01:26:08   That to me is a really key characteristic of what is a podcast.

01:26:13   It has to be intended as an audio first experience, which isn't quite captured on this chart.

01:26:20   But then it can be a video.

01:26:22   It can be a Spotify exclusive.

01:26:24   Yeah, see, I don't know.

01:26:28   I am very sympathetic to the position you were expressing last time, which is there's,

01:26:33   or at least I thought it was the position you're expressing last time, that there's

01:26:37   Something about it needs to be generally accessible. I think for me, the example in my mind that

01:26:45   really sticks out is Audible does these things that they call podcasts, which are like little

01:26:50   shows that you can subscribe to in the Audible app.

01:26:52   I don't know why they call them podcasts. I don't get that one at all.

01:26:56   I've listened to some of them, like they're fine, I haven't found one that's amazing

01:26:59   or anything, but they're also, I don't know, there's some part of my brain which

01:27:03   is just like, this is not a podcast at all.

01:27:05   It's just original audio content.

01:27:06   Yeah, it's an audio-

01:27:07   call them audible originals yeah and they call them podcasts and I just wonder if maybe

01:27:12   they should just stick to one of those descriptors but yeah but anyway like those are the ones

01:27:18   that really clang in my brain of like but what's the key thing there and I think the

01:27:22   fact that they're only available on audio really makes them not a podcast to me okay

01:27:27   well then Joe Rogan yeah is that a podcast?

01:27:31   - Ah, damn you, Myke.

01:27:35   - Right?

01:27:37   'Cause I would say yes, it is,

01:27:38   even though it's a Spotify exclusive.

01:27:40   - Curse you.

01:27:44   That is the perfect counter example to my position.

01:27:49   - So that's why I feel like for me,

01:27:51   there is the third axis, which is like intent.

01:27:54   And that's when if we go on this third axis of intent,

01:27:58   I would sit in the modernist

01:27:59   where it could be like any audio or video that's

01:28:01   subscribable can be a podcast,

01:28:04   but it's on creators intent.

01:28:06   - Yeah, but see, look, just to make things hard,

01:28:08   it has to be phrased in order to fit the meme as,

01:28:11   is a podcast.

01:28:13   You can't put in these weasel words of can, right?

01:28:15   The purpose of the alignment chart

01:28:17   is to definitively answer what is a podcast.

01:28:20   - Well, I'll tell you now, it doesn't.

01:28:22   - I really love looking at this chart,

01:28:24   that traditionalist traditionalist is

01:28:26   an audio RSS feed is a podcast.

01:28:29   And then you start moving in these reasonable directions of any audio that's subscribable

01:28:33   is a podcast.

01:28:35   And then as you get into the radicalist ones, it just gets very funny, right?

01:28:39   Any audio available on the internet is a podcast.

01:28:43   Anything that you can subscribe to is a podcast.

01:28:46   And my personal favorite, the radicalist radicalist is that anything on the internet is a podcast,

01:28:53   which made me laugh so hard when I first saw it.

01:28:57   Because that got me.

01:28:59   Because you know what?

01:29:02   That fits with some people that I've come across in my life, you know what I mean?

01:29:06   I also enjoyed the cortexes giggling over this one, where people were like, "This

01:29:11   comment on this Reddit thread is a podcast, right?

01:29:13   Because I'm a radicalist?

01:29:15   Radicalist!"

01:29:16   I thought this was fantastic, and thanks to XD1936 for posting it, it got a really good

01:29:22   laugh out of it and I think yeah it was just a fun way to try to encapsulate the conversation

01:29:27   from last time.

01:29:30   This episode of Cortex is brought to you by Wealthfront.

01:29:32   Is your bank keeping money that could be yours?

01:29:35   If you're earning less than Wealthfront's 2.55% APY, they might be.

01:29:40   Here's why.

01:29:41   Federal interest rates have been going up this year which means banks have had the opportunity

01:29:44   to earn more on your savings.

01:29:46   Where's all the extra money going?

01:29:48   Well according to the FDIC, the average US bank has only raised their rates to 0.17%

01:29:55   this year, while Wealthrun is now offering their clients a rate that's about 15 times

01:29:59   higher with the Wealthrun Cash account.

01:30:02   Of course you have a choice.

01:30:03   You could let your bank keep the extra money they're making, or you could move your savings

01:30:08   to a high yield account like Wealthrun and earn 2.55% APY.

01:30:13   I think that having your savings in order is really just a great thing, it's one less

01:30:17   thing to have to think about, so check it out. It's super easy to sign up, plus you

01:30:21   get unlimited transfers that are completely fee free and up to $1 million in FDIC insurance

01:30:27   through partner banks. There are no account fees, no minimum balance, and if you sign

01:30:31   up at wealthfront.com/cortex you get a free $50 bonus of an initial deposit of $500. Don't

01:30:37   let your bank keep the interest you could be earning. Join nearly half a million people

01:30:42   who already use Wealthfront to earn 15 times more than the average US bank. That's wealthfront.com/cortex

01:30:49   to get started and earn your free $50 bonus. That's wealthfront.com/cortex. This has been

01:30:55   a paid endorsement for Wealthfront. Our thanks to Wealthfront for the support of this show

01:30:59   and Relay FM.

01:31:01   We are hurtling towards the end of the year.

01:31:02   Uh, yes, alarmingly fast.

01:31:05   And this is the final regular episode of Cortex this year.

01:31:09   Don't remind me, Myke.

01:31:10   So if you have more to say about AI, oh god, great, where's AI art gonna be in January?

01:31:16   Oh my god.

01:31:17   By January we're gonna be at step two of my three steps to the apocalypse.

01:31:21   Excellent, can't wait for it.

01:31:23   So maybe that's why like we had to spend the best part of two hours doing follow-up, because

01:31:28   we're not gonna do it now.

01:31:30   Yeah I guess so, let's say that, that was the plan, we didn't just end up talking about

01:31:34   it for forever.

01:31:35   There is a possibility we might do something in Mortax, getmortax.com, I can't say that

01:31:40   for sure because I don't know what we're going to do over the next two months except

01:31:43   for what I know we're going to do which is state of the apps November yearly themes December.

01:31:49   So on that note for yearly themes I wanted to take this as a time to suggest a method

01:31:56   to people for preparing for their yearly theme. If you don't know what a yearly theme is I

01:32:02   will put a video in the show notes where Grey explains what yearly themes are but we're

01:32:06   going to get to it again obviously in December so I don't want to talk about it now. But

01:32:11   I am personally at the point of my year now where I'm preparing my yearly theme. The

01:32:17   way I do this is I have it in my mind to start noting down things that I'm happy about

01:32:25   in my work life and personal life and things that I find frustrating in those as well and

01:32:31   things I would like to change. So I keep this as an Apple note, it's in my yearly themes

01:32:36   Apple note, but I have it just playing in my mind that as I'm bumping into things that

01:32:42   are good or bad that I would like to continue or stop, I write them all down. I just write

01:32:47   them down. I don't need to draw any comparisons to them yet, I just start writing them down.

01:32:52   So I end up with a list of things, maybe 20 things or so, over the space of a few weeks

01:32:58   or a month that are in my head. Some of them are not even things I bump into, but because

01:33:02   I'm in this kind of reviewing mode, things just pop up in my head randomly. I'm like

01:33:07   "Oh, that's a thing, I'll write that down." Then as we get towards November or so, I start

01:33:13   to review this list of things and note the similarities where I can tie some similarities

01:33:18   between things. Be like "Oh, that's actually related to this" or "That is similar to this."

01:33:24   I then use these links as the basis to build my theme.

01:33:27   So what do I think I could do next year to address some of these common issues that I

01:33:34   have or these common things that I would like to do more of?

01:33:38   So if you struggle to come up with creating yearly theme, this is my top tip.

01:33:43   It's how I do mine every year.

01:33:44   I don't know if this is asking for spoilers for the theme episode, which I don't want,

01:33:49   but I just I feel like I would like some concrete examples of the kinds of things that you're

01:33:52   down that you end up thinking about and they don't they don't have to be

01:33:56   examples for this year. I will give you some examples for this year but I'm not

01:34:00   gonna give you so many that I think was born my theme. Good. I want to spend more

01:34:04   time on product creation and design. That's one. I want to listen to more

01:34:09   music. I'm happy with how I've looked after my health and want to do more of

01:34:13   that. I want to be smarter about my scheduling and the days that I'm in the

01:34:17   office. These are the kinds of things I write down.

01:34:20   Okay, so presumably then the way you have this happen is there's some frustration

01:34:27   about the scheduling and then I guess you notice that and add it to this file?

01:34:30   Is that the system? Mm-hmm. Okay. Like all these are at the moment

01:34:35   it's just an outline like bullet points of things, right? And I write a

01:34:39   bunch of notes down, I write some context down, and so then I start reading through

01:34:43   these and I'm like, "Oh hang on a minute, this can relate to this and this.

01:34:47   Or like, "Hey, I've written a bunch of times here that I'm unhappy about this kind of thing

01:34:52   in like a bunch of different ways. So can I address that thing?"

01:34:58   - Hmm, okay, actually, interesting. I think I should frame this thought a bit differently,

01:35:03   because I have a notes file where when it occurs to me things that I may want to discuss

01:35:08   on the theme episode, I make a note of them. I guess this is just a different kind of

01:35:13   the same idea of like things about my own personal theme there.

01:35:17   But no, I like that as a have a place to collect what you notice about your life,

01:35:25   which also then just encourages you to notice more.

01:35:28   Might be a way to try to pitch it to people who are,

01:35:32   as the end of the year comes hurtling towards us,

01:35:35   and if they've never tried a theme before,

01:35:37   might want to think about that over the next two and a half months.

01:35:41   They're going to start a theme at the beginning of the year.

01:35:43   It's like just have a note somewhere to put down things that you notice about your life as a way to get started

01:35:49   And then you have something to look over and try to synthesize later. It's not about any particular moment

01:35:56   Okay. Yeah, that's interesting

01:35:57   I think that this is make something that's more likely to stick on because it's it's actually related to things that you've experienced

01:36:04   Mmm, and pushes it further away from the New Year's resolution idea because New Year's resolutions are typically created from whole cloth

01:36:12   like this just like I have this aspirational idea of myself that I would like to be this different kind of person and I'll

01:36:17   Just go and live that life

01:36:18   Where what I'm suggesting that you do is how do you live right now?

01:36:22   What frustrates you about things that are happening right now? What would you like to be better in your actual life right now?

01:36:29   Mmm and try and think about what that might be

01:36:33   It's related to something that's actually happening rather than something that you look forward to just imagine yourself as a different person

01:36:39   Yeah, that's a good distinction because, again, also the New Year's resolutions have

01:36:44   the feeling to me of homework assignments that you forgot about until the morning of

01:36:48   and it's like, "Oh no, I need to write an essay about what I'm going to do this

01:36:50   year." Whereas this makes me think of—it's actually kind of popping into my head the

01:36:55   way some researchers have tried to study how happy people are by messaging them at random

01:37:01   points throughout the weeks or the years of just like, "Hey, right now, how do you feel?"

01:37:06   And this is closer to that.

01:37:07   It's like a continual process of, hey, notice in your life

01:37:10   how you're feeling and just make a little note of it.

01:37:12   And then later, you can have some stuff to review.

01:37:14   So I like that.

01:37:15   I think that's good.

01:37:16   I think that's a good suggestion for people looking

01:37:18   for a place to onboard where to start instead of just trying

01:37:23   to create a theme out of whole cloth

01:37:25   after they've listened to the theme episode.

01:37:27   [BING]

01:37:28   Oh, great.

01:37:29   There's something I've got to tell you about before we wrap

01:37:32   up today.

01:37:33   Yeah.

01:37:33   I didn't know about this.

01:37:35   somehow seemed to have slipped by everyone. In iOS 16 there is now a Dvorak software keyboard

01:37:44   on the phone. Oh yeah? Let me grab my phone here. It's in the settings app right? You

01:37:52   have to go to settings and then I think it's general keyboard and then whatever your setting

01:38:00   as like English UK or English US or whatever. If you tap that you get the option of QWERTY,

01:38:06   AZERTY, QWERTS and DIVORAC.

01:38:07   Oh I see so it's not a new keyboard it's under English I would never have found that.

01:38:12   You could probably add it as a new keyboard?

01:38:14   No it's not that's what I was just trying to do.

01:38:16   Oh god I hate it.

01:38:18   Oh interesting.

01:38:19   Wow that's horrifying.

01:38:20   Do you still use DIVORAC?

01:38:24   Of course I use DIVORAC Myke.

01:38:26   So your keyboard on your desk is set in the Dvorak layout?

01:38:30   I feel like we haven't spoken about that

01:38:31   in a really long time.

01:38:32   I've kind of forgotten.

01:38:34   - Yes, well, my keyboard, if you look at it,

01:38:37   the current one I'm using is, it has a QWERTY layout,

01:38:40   but the keys are mapped to the Dvorak settings.

01:38:43   - You are a monster.

01:38:45   - No, no, no, I actually, I've decided that's best.

01:38:48   That's the better way to go.

01:38:49   I've stopped getting keyboards.

01:38:51   Look, I think there's only two ways to go.

01:38:53   QWERTY visual layout or blank, right?

01:38:58   Like nothing on any of the keys.

01:39:01   - Why can't you use the Dvorak visual layout?

01:39:03   I think I could build you a keyboard one day

01:39:08   that would be laid out this way.

01:39:10   - Yeah, okay, I guess here's,

01:39:12   I was just trying to articulate why.

01:39:15   After years and years of doing all sorts of different things,

01:39:17   why have I settled on this?

01:39:19   The answer is that keyboard shortcuts are funny

01:39:24   in a lot of applications.

01:39:27   So there's something in the system level

01:39:31   where a keyboard shortcut will either trigger based

01:39:34   on the letter that the key represents

01:39:37   or it will trigger based on the location that the key is.

01:39:41   You can see this explicitly in some things like Final Cut

01:39:45   where you can reprogram the key based on letter

01:39:49   or location. So I guess the way I use it is that it is I'm just not having I'm struggling

01:39:55   trying to think of a specific example that's not final cut, but I know it comes up where

01:39:59   it's useful to be able to see what does the keyboard look like for everyone in the whole

01:40:06   wide world, sometimes for keyboard shortcuts like that does come up where I am glad that

01:40:11   the layout is a QWERTY layout. And since I touch type, I don't look at the keyboard,

01:40:16   It doesn't matter when I'm actually writing what the keys look like at all, which is why

01:40:20   I have gone with a completely blank keyboard sometimes, which I do think is cool, but can

01:40:25   be annoying for keyboard shortcuts.

01:40:27   I have a question for you.

01:40:28   Yep.

01:40:29   Well, because one, I'll come back to the keyboard shortcuts thing because I know how to fix

01:40:32   that.

01:40:33   Depends on what you use.

01:40:34   Anyway, when you are using your Dvorak layout, what is the copy command?

01:40:40   Okay, so for me, it's command I is what it would look like on a query keyboard. That's

01:40:47   copy. Oh, okay. Yes, I know what you're saying. Now you are actually hitting still command

01:40:54   C though is what I'm asking. That's more work, but it's in a different place. Okay, it's

01:40:58   in a different place. So one of the things that exists now, which I don't know if it's

01:41:03   actually a good idea for any other lunatics who might want to switch their keyboard layouts.

01:41:09   One of the things that computers have gotten better at over years is there are now explicit

01:41:13   layouts which are on the Mac, it's called something like Dvorak preserve shortcuts.

01:41:20   So it's like you type in Dvorak, but it totally ignores the keyboard layout for all keyboard

01:41:25   shortcuts.

01:41:26   It would be like command J looking at the Dvorak layout right in that scenario where

01:41:33   it's preserving the location of the key.

01:41:36   Yeah, it's funny, it's like I can't – look, I is C. What's the complication here?

01:41:42   I'm looking at a Dvorak layout on Wikipedia and like, you know, J is in the location that

01:41:48   C is in a QWERTY keyboard.

01:41:51   And so you would hit, I guess, command J, because what it's saying is, the way I've

01:41:56   understood the way you've explained that, is like, it's keeping the physical location

01:42:00   of the key no matter what the key actually says it is.

01:42:03   Yeah, okay. So I just looked it up like what is the actual thing called in the system setting?

01:42:08   So on the Mac, you can set it as something called Dvorak QWERTY command. So I presume

01:42:14   that what it's doing there is whenever you hit command, it ignores the Dvorak layout

01:42:17   and reverts to the QWERTY layout.

01:42:19   That seems horribly complicated. Yeah, all of this to say, one day, because I want to

01:42:24   build your keyboard one day, just for fun. I would choose a keyboard where I can change

01:42:30   the programming of the keyboard. The boards that I build these days tend to all support

01:42:36   a piece of software called VIA which overrides the layout on the keyboard. So it's not relying

01:42:44   on the Mac. So if you hit any shortcut in any app it's going to register as the key

01:42:53   that is on the Dvorak layout. It doesn't matter that you would try and do some kind of key

01:42:59   binding and software. Does that make sense? Yes. Right? So like one of the

01:43:04   things you're saying is like for example if you hit some kind of keyboard

01:43:08   shortcut in Final Cut it's going to assume that's QWERTY even if you've set

01:43:12   to Dvorak because there's like a software in between the two of them

01:43:15   that's trying to communicate it. But I'm saying the hardware of the keyboard

01:43:20   would communicate to the computer so I am convinced it would get it correctly

01:43:27   it would always hit correctly.

01:43:29   - Okay, I feel like I'm not 100% understanding,

01:43:32   but there's also so many layers here.

01:43:35   One of the other problems is,

01:43:36   I don't know if any of the cortexes

01:43:37   are able to help me out with this,

01:43:39   if anyone's already done this.

01:43:40   Here's the maximum level of crazy, right?

01:43:43   I use the Dvorak layout,

01:43:45   but I've learned all the keyboard commands,

01:43:48   all the basic system commands,

01:43:50   with the Dvorak layout,

01:43:51   so I'm not pressing the regular buttons

01:43:53   that everybody knows.

01:43:55   However, in Final Cut, the way that the programmers laid out all the keyboard shortcuts that do

01:44:03   things, it makes sense physically where they are on the keyboard.

01:44:08   Whereas like Command Copy, Command Paste, like it doesn't really matter if those two

01:44:11   are next to each other.

01:44:13   But there's tons of stuff in Final Cut, which is like trim from the start of the clips,

01:44:17   trim from the end of the clip.

01:44:19   And you want to have those keys like next to each other on opposite sides.

01:44:22   There's a lot of like physicality of this.

01:44:24   So in Final Cut, I've been trying to slowly build up my own custom mapping, which is to

01:44:31   make it so that when I am using the Dvorak layout, it's still acting as though it's a

01:44:37   QWERTY layout just for the Final Cut shortcuts.

01:44:41   So every time I try to learn a new Final Cut shortcut, I try to go in and like, change

01:44:46   what it is.

01:44:47   But I've wondered like, has someone just done this?

01:44:49   is there a QWERTY for Dvorak remapping of all the Final Cut shortcuts that someone has just done?

01:44:55   Because I could tell like I'm getting into a situation of, "Oh, this is a little bit

01:44:59   inconsistent." So anyway, I've just wondered if like a person has done that work for me.

01:45:02   And there's also some weird system stuff that I've done with a few of the shortcuts.

01:45:07   Anyway, there's a bunch of these weird little problems. And I also have this

01:45:10   minor annoyance. Listen, before I say this annoyance, I understand. I understand why

01:45:15   Apple did it. You don't have to leave me comments for why they did it. I

01:45:18   understand. It's still annoying. But it used to be that Apple, when you had

01:45:23   different keyboard layouts, it had a little country flag to represent each of

01:45:28   the layouts. And so the US keyboard layout had a little US flag. And the

01:45:33   Dvorak one was just DV. And they made the change so that they don't use country

01:45:41   flags for languages. Now again, I get it, but it actually causes a huge annoyance because in my menu

01:45:48   bar I want to see which layout the keyboard is using. Is it using Dvorak or is it using US?

01:45:54   Because sometimes I do switch between those two layouts and now I can't visually see instantly

01:46:00   that the keyboard is on US layout versus Dvorak layout and it's maddening because it's just a

01:46:06   a little box and the box either says DV or it says US. And like, no, it was so much better when it was

01:46:13   black DV and that's Dvorak. And then there was a little American flag when it's the US layout.

01:46:18   So it's super frustrating. But anyway, all of that aside, I'm looking at this Dvorak layout on my

01:46:23   phone, and I hate everything about it. I don't think Dvorak was made for phones. I think all

01:46:29   All of the advantages for typing Dvorak with two hands, I don't feel like they translate

01:46:36   for two thumbs on a phone.

01:46:38   I think maybe the QWERTY layout is actually superior for the phone.

01:46:41   Yeah, all of the vowels are next to each other on the Dvorak one.

01:46:45   I'm just trying to type some words here.

01:46:47   By the way, I have two applications for you that can restore those flags.

01:46:52   Oh really?

01:46:53   Yeah, one is called Keyboard Switcheroo.

01:46:56   The other is called Colorful Input Menu Flags.

01:46:59   'cause they're both in the Mac App Store.

01:47:01   - Oh, okay.

01:47:02   Sold.

01:47:03   - Two enterprising developers who created applications

01:47:06   to bring that back.

01:47:08   I'd like to thank Glenn Fleischman at Macworld

01:47:11   for writing the article too.

01:47:12   'Cause I know I'd seen these,

01:47:14   and so I just did a quick Google while you were upset.

01:47:17   And actually on keyboard switcheroo,

01:47:20   they have in their app screenshots,

01:47:23   US, French, German, Dvorak.

01:47:25   So you get a little US flag

01:47:28   a little DV for Dvorak.

01:47:29   Oh good, I'm glad to know I'm not the only person who was annoyed by this.

01:47:32   So I'll definitely check those out.

01:47:34   Thank you.

01:47:35   You can always rely on Mac developers to solve the little annoyances.

01:47:38   Yeah you can.

01:47:39   What I was just thinking here is with iOS, I think Qwerty on the phone, I think it does

01:47:45   make sense.

01:47:46   I feel like the weirdness of the way Qwerty is laid out to sort of slow you down, although

01:47:52   I think that that's overplayed as a story of why it exists.

01:47:55   I think that actually works as an advantage for typing with your thumbs on a tiny phone

01:47:59   screen, but I am glad that this exists for the iPad.

01:48:03   I might actually switch it on the iPad, because that's the place where it's been annoying.

01:48:06   Big keyboard.

01:48:07   And that's more like a larger keyboard there is what you're more used to for Dvorak, right?

01:48:12   Bigger keyboards like a computer keyboard.

01:48:15   Can you get MacBooks in Dvorak?

01:48:17   I don't think so.

01:48:18   I don't think they sell that.

01:48:21   I'm pretty sure you have to just pick between American layout and the English layout when

01:48:26   I'm buying the keyboards, which is the thing I'm always very careful about.

01:48:30   Like, please give me the US layout one.

01:48:32   I do not want the UK layout one.

01:48:34   There's no little "huh" there.

01:48:35   Like the American enter is clearly superior.

01:48:38   I build all of my keyboards in NC.

01:48:40   My laptop is in the British layout, but I think in the future I might try and remember

01:48:46   to just order them in US layout because that's what I'm much more used to now.

01:48:49   Wait a second.

01:48:50   - You build your keyboards in the US layout, NC layout.

01:48:54   - Yeah.

01:48:55   - But you get the MacBook keyboards with the English layout?

01:48:58   - I don't think about it when I'm buying them.

01:49:01   I think I will now in the future, right?

01:49:03   'Cause I'm getting more annoyed because I'm switching

01:49:06   between something I'm not used to.

01:49:08   And the keyboards that I build,

01:49:10   most of them are just more available

01:49:13   and easier to build in NC layout.

01:49:14   So I'm just used to it now, so the US English layout.

01:49:18   But I never think about it.

01:49:19   and obviously the default on the British Mac store

01:49:23   is the British layout, so I don't think about it

01:49:25   and I just buy what's the default,

01:49:28   but I think I wanna start changing it.

01:49:30   It doesn't look like they do Dvorak,

01:49:31   it's not in their options.

01:49:32   - I don't think I'll be using this on my phone.

01:49:35   I think it's worse on the phone,

01:49:36   but I will definitely try it out on the iPad

01:49:38   when the iPad comes out,

01:49:39   'cause that's where it's more of a problem,

01:49:41   but I mostly don't really think about it very much

01:49:44   'cause it's just been so long,

01:49:46   like this has been my entire life.

01:49:48   The last time it caused me problems was when I worked at a school and had to switch between

01:49:53   the QWERTY and Dvorak typing keyboards a bunch for using different computers, but even then

01:49:57   it was not the worst.

01:49:58   Like, I just learned to touch type while not looking and then sort of type while looking

01:50:03   at the keyboard as two different modes that my brain could switch between.

01:50:06   But I wonder why on earth they added Dvorak to the—

01:50:10   Now.

01:50:11   —to iOS 16 now.

01:50:12   It's a very funny, like—

01:50:13   It's strange.

01:50:14   Why did it take 16 attempts?

01:50:15   You know what I mean?

01:50:17   why 16 versions in? Is Dvorak having some resurgence that we're not aware of? Like why?

01:50:23   It is odd, right? Yeah, I'm not even sure if I was suggesting to someone now who had RSI problems if

01:50:31   Dvorak is the way to go. I always forget, I think it's Colmak or something? There's another one

01:50:37   which is definitely worth investigating if you're learning now versus Dvorak. And it's also a

01:50:42   a feature that I don't know why but it took all the way until now for people to notice

01:50:47   it it seems like. There's been a bunch of articles written because it was kind of discovered

01:50:51   but that means it went through the whole beta process without people seeming to know about

01:50:56   it. I don't know.

01:50:57   The thing that's also confusing about it is if they did add... why didn't they add something

01:51:03   like Colmac? I mean look I don't know a lot about programming iOS to add additional keyboard

01:51:08   layouts but if you're going to add something like Dvorak how hard is it to also add the

01:51:14   other popular one of these if you're gonna do it?

01:51:17   Well isn't Dvorak like the popular one though?

01:51:20   I don't know if that's true or not.

01:51:22   I feel like it must be or at least in the requests that they get or some high up executive

01:51:27   like Tim Cook is a Dvorak guy and like he'd had enough you know?

01:51:31   Yeah I mean with Dvorak I just always feel like there's dozens of us.

01:51:36   Dozens!

01:51:37   (laughing)

01:51:38   - You've seen the Godfather, right?

01:51:40   - No. - What?

01:51:42   - I've not seen the Godfather.

01:51:44   I've just never really been that interested.

01:51:46   - Okay, well, in the Godfather,

01:51:49   this is pointless for you now,

01:51:50   but in the Godfather, there is a moment,

01:51:53   I think Godfather 2, I think it's Godfather 2,

01:51:56   which I assume you also have not seen,

01:51:58   because why would you have, right?

01:51:59   - Yeah, that's a correct, yeah, I've seen the third one,

01:52:01   but not the first or the second one.

01:52:03   - You're joking, right?

01:52:04   - I am joking, yes. - Good, good, good, good.

01:52:06   There is a moment where a bunch of things happen

01:52:08   and the line is, "We're settling all family business."

01:52:13   There's a line about settling all family business.

01:52:15   And I feel like that is today's episode.

01:52:17   We have just settled a bunch of business, right?

01:52:20   A lot of follow-up has been dealt with, right?

01:52:23   We've spoken about Dvorak, right?

01:52:25   Like we're just settling the business

01:52:27   before the year ends for us.

01:52:28   Like very good episodes, now we go to the specials.

01:52:31   - Yeah, that's true.

01:52:32   - There are two more pieces of family business

01:52:34   that I would like to settle before we finish today.

01:52:37   One is to thank every Cortexan who donated

01:52:40   to our Saint Jude campaign.

01:52:41   Overall, throughout the month of September,

01:52:44   we raised $706,397.10.

01:52:49   All of us absolutely astounded.

01:52:53   This is the most money we've ever raised.

01:52:55   - It's unbelievable.

01:52:57   It's an incomprehensible amount of money.

01:53:00   So we have now not only passed $2 million raised

01:53:05   in the last four years,

01:53:06   we have now hit $2.2 million raised in the last four years.

01:53:11   - It's incredible.

01:53:15   It's just incredible to think that the relay listenership

01:53:18   has raised that much money for St. Jude.

01:53:22   So yeah, thank you to everyone who donated to the campaign.

01:53:27   That's completely mind blowing.

01:53:29   You know, it seemed for a while that it was like,

01:53:31   okay, we're gonna raise a lot of money,

01:53:33   we're gonna meet our goal,

01:53:34   but we're probably not going to exceed.

01:53:36   Because there is definite economic challenges

01:53:39   right now around the world,

01:53:41   and we were thinking, okay, that's gonna be that.

01:53:43   And then it just, the last couple of days even,

01:53:47   like the last couple of weeks, it just exploded

01:53:49   'cause people were just getting their final totals in.

01:53:51   We had to, we were gonna end it on September 30th,

01:53:54   but we extended it to October 3rd,

01:53:56   because the money was just piling in in the last day.

01:54:00   It was like, all right,

01:54:01   we're gonna leave this open a couple of days.

01:54:02   I think like in the last day or two,

01:54:04   we raised like an extra $110,000 or something.

01:54:07   So it was like, probably should keep this open.

01:54:10   - Yeah, that's worth leaving open an extra day.

01:54:13   - But overall, this year's campaign was fantastic.

01:54:16   We learned a lot and for me,

01:54:18   it was just so incredibly rewarding and fun

01:54:22   to be able to be back in Memphis for the Podcastathon,

01:54:24   which was a great success.

01:54:26   The whole video is on YouTube.

01:54:27   I'll put a link in the show notes

01:54:28   if people don't want to watch it.

01:54:30   A very kind comment to timestamp the whole thing.

01:54:33   So if you want to jump around to different segments

01:54:35   or whatever, that's all in there

01:54:37   in one of the comments in the YouTube video.

01:54:39   But yeah, it was truly fantastic

01:54:42   and we achieved something that I just did not think

01:54:45   was going to be possible.

01:54:46   And once again, the Relay FM community

01:54:48   has gone and surprised us.

01:54:51   Cortexmerch.com.

01:54:52   This is the second thing, final thing.

01:54:54   And what's the final piece?

01:54:56   - Is subtletys and subtle sweaters.

01:54:58   We're reminding you, maybe you have now finished your commute

01:55:00   and you're walking into the office or whatever,

01:55:02   and you didn't do what I asked you to do earlier,

01:55:04   so now you can get your phone out

01:55:06   and you can go to cortexmerch.com

01:55:08   and buy yourself a subtlety or subtle sweater or both

01:55:10   or whatever you want.

01:55:12   - Unlike the St. Jude fundraiser, this will not be extended.

01:55:16   So for realsies, cortexmerch.com.