PodSearch

Connected

490: Ivy League Ferret

 

00:00:00   (upbeat music)

00:00:02   Hello and welcome to Connected episode 490.

00:00:12   My name is Steven Hackett and I have the pleasure

00:00:14   of being joined by Mike Hurley.

00:00:17   Hey Mike.

00:00:17   It is a pleasure.

00:00:19   I'm so happy to be here.

00:00:21   And I also have the pleasure of being joined

00:00:23   by Federico Vittucci.

00:00:24   Ciao Federico.

00:00:25   Ciao, it is also a pleasure for me.

00:00:27   We're all very pleasured.

00:00:29   I think.

00:00:30   People say this is the most pleasurable podcast.

00:00:33   They say that.

00:00:34   They say that.

00:00:35   We got, a lot of people are saying that.

00:00:38   They send us email to like,

00:00:39   "Hey, your podcast is so pleasurable."

00:00:42   Such so pleasurable for me, I love it.

00:00:45   You actually would not believe the amount of emails we get

00:00:48   where people say, "This is a pleasurable podcast."

00:00:51   Yeah, yeah.

00:00:52   So take that other tech shows.

00:00:55   Take that other tech shows.

00:00:55   Speaking of which, podcast in the shower.

00:00:59   Podcast in the shower.

00:01:01   So I ran a little poll on Mastodon asking,

00:01:04   "Do you listen to podcasts in the shower?"

00:01:08   Can you remind me why?

00:01:10   Because I feel like I can only half remember

00:01:12   why this is a thing that you wanted to do.

00:01:14   I think I mentioned that I do it

00:01:16   and someone in Discord or one of you

00:01:18   thought I was bananas.

00:01:20   We were talking about something with you in the bath.

00:01:25   Yes.

00:01:26   And you had a thing you were doing

00:01:29   with your phone on a shelf, cases.

00:01:34   We were talking about cases.

00:01:38   Yeah.

00:01:39   Maybe.

00:01:40   Why do you use a case?

00:01:41   Because you're worried you're gonna drop your phone?

00:01:43   I don't want it to slide across the sink.

00:01:46   There you go.

00:01:48   The grip, the grip, you wanted grip.

00:01:49   That's all right, you need that grip.

00:01:51   And so 1,247 people voted on Mastodon.

00:01:56   42% of people said yes,

00:02:02   I do listen to podcasts in the shower.

00:02:04   58% of people said no, I do not.

00:02:08   So I think that there should be a third,

00:02:11   it's too late now, which is I would like to.

00:02:13   So I would like to listen to podcasts in the shower,

00:02:18   but in my new shower, in my new house--

00:02:21   Stop the count!

00:02:23   Stop the count, stop the count.

00:02:24   It's time to stop the count.

00:02:26   But I can't, in my new shower,

00:02:28   I can't hear the podcasts anymore is the problem.

00:02:31   'Cause I have like a cubicle,

00:02:33   and I don't have a space in the cubicle to put my phone,

00:02:35   where my phone won't just be literally in a pool of water.

00:02:38   Although there were some replies to this post

00:02:42   where I was like, oh, that's a smart idea.

00:02:43   Like people building shelves, shelves?

00:02:47   Yeah.

00:02:48   Shelving?

00:02:48   But I don't think that would fly at home,

00:02:50   which probably for good reason.

00:02:51   We had a couple of things.

00:02:53   We had someone write in, they're just putting their phone

00:02:56   on top of like the edge of the shower.

00:02:58   That's a terrible idea.

00:02:59   Your phone's gonna explode.

00:03:00   Look, I'm a professional phone breaker.

00:03:03   This phone is going to break.

00:03:05   Other people had like sticky things on the wall,

00:03:08   they mount their phone in.

00:03:09   Look, it's way simpler than that.

00:03:10   All you need is a waterproof Bluetooth speaker.

00:03:15   I have one in the show notes from JBL.

00:03:17   I happen to own the old version of this.

00:03:19   I got it for free in something at some point years ago.

00:03:23   I've run it forever.

00:03:25   It's great.

00:03:26   So I can Bluetooth it.

00:03:27   My phone can stay safe and sound on the edge of the sink

00:03:29   thanks to the little leather nomad thing,

00:03:31   keeping it safe and sound.

00:03:33   And I can listen in the shower.

00:03:35   It's great.

00:03:36   Don't put your phone in the shower with you.

00:03:37   That's weird.

00:03:38   That's too far.

00:03:39   Why?

00:03:40   Why is that weird?

00:03:41   'Cause you're in the shower.

00:03:42   You don't want your phone in there.

00:03:43   What has that got to do with anything?

00:03:45   It's supposed to be clean.

00:03:46   What if your mom FaceTimes you

00:03:47   and you accidentally answer, you know?

00:03:49   Why would you accidentally answer?

00:03:51   Why would you do that? It's slippery.

00:03:51   Your hands are wet, you know?

00:03:54   You're gesturing wildly listening to ATP

00:03:56   and then suddenly you're talking to your mom.

00:03:58   I don't think this would happen to me.

00:04:00   There's only one way to find out.

00:04:03   You call me and see what happens.

00:04:05   I have to have an experiment.

00:04:07   So I'm putting this in the show notes, the JBL thing.

00:04:13   And instead of waterproof, I wrote shower proof.

00:04:16   I mean. That's not right.

00:04:18   It kind of good though.

00:04:19   It is. It kind of is.

00:04:21   But also that's not what people are looking for.

00:04:23   I think shower proof.

00:04:25   Probably not.

00:04:26   So anyways, look, I get it.

00:04:30   Sometimes your acoustics may not be good.

00:04:32   My shower is not very big.

00:04:34   So I can feel like it works.

00:04:36   Maybe mine has a glass door,

00:04:38   but maybe if you're showering like in a, you know,

00:04:41   like a shower tub or using the shower curtain,

00:04:43   maybe your acoustics would be better with that.

00:04:45   I will say, due to experimentation,

00:04:49   the audio is much better in my shower

00:04:52   if I put my towel over the door.

00:04:54   So there's like some soft material,

00:04:56   like hanging over the door into the shower.

00:05:00   Does that make sense?

00:05:02   I don't think so.

00:05:03   Okay, so I have my shower's tiled in

00:05:06   and there's a glass door that swings open.

00:05:09   And usually the towel's hanging on the wall

00:05:11   outside of the shower, right?

00:05:12   So I get out of the shower, like open the door a little bit,

00:05:14   stick my arm out and grab the towel.

00:05:17   But if I take the towel and put it over the door,

00:05:20   so it's like hanging half in the shower,

00:05:23   half out of the shower,

00:05:25   that fabric hanging in the shower

00:05:27   greatly approves shower acoustics.

00:05:29   Okay, interesting.

00:05:32   You don't have to buy a sound panel for your shower.

00:05:35   Just put a towel.

00:05:36   But it wouldn't hurt.

00:05:38   Wouldn't it?

00:05:38   If you want optimal results, that's what you need to do.

00:05:44   That's the end of the shower cast.

00:05:45   Thank you for coming.

00:05:46   I'm so happy.

00:05:47   I have a question for the both of you.

00:05:50   What are you, if anything, using for RSS on Vision Pro?

00:05:55   Yeah, there's no great native experience at the moment.

00:06:00   I'm just using Lire, that's L-I-R-E, in compatibility mode.

00:06:05   That's my main RSS client on other platforms

00:06:08   and the iPad app is available in compatibility mode.

00:06:11   So that's what I use with iNoReader Sync.

00:06:14   I actually think even compatibility mode is fine.

00:06:17   Like that's totally fine.

00:06:18   Yeah, it's fine.

00:06:19   It's fine.

00:06:20   That's what I use.

00:06:22   Typically in portrait mode,

00:06:23   so it's nice on one side of my workspace, yeah.

00:06:26   I am missing currently, big time, Reader and Timery.

00:06:31   They are both heartbreakers for me

00:06:33   that they're not even in compatibility mode.

00:06:36   And it's making me very sad

00:06:37   'cause these are two apps that I would like to use

00:06:41   and I can't and it's bugging me

00:06:44   'cause I think especially Timery because I'm doing stuff

00:06:47   and I'm like, I don't even think about having a timer set.

00:06:50   And then I'm like, oh man, I didn't set a timer.

00:06:53   Well, so I want these apps to exist.

00:06:55   They don't, I don't know what's happening

00:06:57   and either of them, but I want them to be there.

00:06:59   Yeah.

00:07:01   And we'll give a shout out to the one true gamer.

00:07:04   That is John Voorhees.

00:07:05   From OTJ to OTG.

00:07:07   OTG.

00:07:09   O-G-O-G-J.

00:07:10   I feel like this was a thing where,

00:07:13   because Federico was delayed for a week with his Vision Pro,

00:07:17   John had to pick up the mantle and run with it

00:07:20   to make the comprehensive guide to gaming

00:07:22   on the Apple Vision Pro.

00:07:23   It's a really great article.

00:07:24   There's loads of interesting tidbits in there.

00:07:26   There's some things that I've downloaded

00:07:28   that I wanna try out,

00:07:28   like the PlayStation mirroring thing,

00:07:31   which is the thing I'd heard about years ago.

00:07:34   Like this app's been around forever, right?

00:07:36   And, or at least a version of it has been MirrorPlay,

00:07:39   but I've never really used it,

00:07:40   but I feel like this could be a really good reason to use it.

00:07:43   Mm-hmm.

00:07:45   It also, and I think John wrote about this,

00:07:47   I sort of blacked out towards the end of the story

00:07:49   'cause there were a lot of gaming screenshots.

00:07:51   It gets, well, it also, the further the article goes,

00:07:56   the more unconventional (laughs)

00:07:59   I think I would say the methods become.

00:08:02   Yeah, at some point he's talking about his Wi-Fi network,

00:08:04   you know, gaming Wi-Fi, it's all the rage.

00:08:07   Gaming Wi-Fi, everyone's doing it.

00:08:08   But it would be cool if the developer strap would,

00:08:11   like, I want that thing to do everything, right?

00:08:14   Like we want ethernet from it,

00:08:16   we want video in and video out.

00:08:19   Like, just put some ports on things.

00:08:20   It doesn't ever hurt anybody.

00:08:22   Yeah, I mean, if only,

00:08:24   if that theory, the developer strap,

00:08:26   can do more than what it's currently doing,

00:08:29   but it's locked in software,

00:08:30   I really hope that Apple makes that

00:08:32   like a proper Thunderbolt port.

00:08:35   Because honestly, like with ethernet,

00:08:37   it will fix up basically most of the problems

00:08:41   when it comes to game streaming to the Vision Pro, right?

00:08:44   I mean, if you can have ethernet speeds on that thing,

00:08:47   whether you wanna stream PS5 or Xbox or PC,

00:08:50   like, that's gonna be not an issue.

00:08:52   The Vision Pro doesn't have Wi-Fi 6E, unfortunately,

00:08:55   so it's locked to Wi-Fi 6.

00:08:57   And I'm finding that I consistently get worse performance

00:09:02   with Wi-Fi on the Vision Pro

00:09:04   than, say, my iPad Pro, for example.

00:09:06   So I want to believe that maybe in Vision OS 2.0 or something

00:09:10   they will unlock the true potential

00:09:12   of the developer strap.

00:09:13   Let me plug an ethernet adapter in that thing.

00:09:18   Come on, please.

00:09:19   But yeah, it's a great story.

00:09:21   John spent a lot of money and did a lot of research

00:09:24   to find all the possible options right now.

00:09:27   Of course, as soon as we published that,

00:09:29   we noticed that there's a test flight beta

00:09:34   for a native version of Moonlight.

00:09:38   So if you are a PC gamer and you wanna stream games

00:09:42   from your PC to your Vision Pro,

00:09:44   there's a Vision OS test flight for Moonlight,

00:09:47   which is the client for PC streaming.

00:09:50   Gonna leave some links in the show notes.

00:09:52   And the test flight is currently open,

00:09:54   so you can just click a link and start testing it.

00:09:56   We got a couple of small topics to handle here

00:10:00   at the beginning of the show.

00:10:02   The first one is more battery, who dis?

00:10:05   So this is a report from Chance Miller.

00:10:08   I'm just gonna read this

00:10:10   because I don't understand how this happened.

00:10:14   This is Chance.

00:10:15   Apple initially said that the iPhone 15's battery

00:10:18   would retain up to 80% of its original capacity

00:10:21   at 500 complete charge cycles.

00:10:24   Now the company tells 9to5Mac that the iPhone 15

00:10:27   can retain 80% of its original capacity

00:10:30   at 1,000 complete charge cycles.

00:10:33   That's awesome.

00:10:35   That's great news for longevity of these devices.

00:10:37   But why is Apple discovered this now?

00:10:41   I have so many questions.

00:10:43   - It is peculiar.

00:10:45   - So I guess what we are seeing here

00:10:47   is that sometimes a recount does work.

00:10:51   - Is that? - Yeah, I guess.

00:10:53   - I think so. - Yeah.

00:10:54   - I think that's what it's saying.

00:10:55   - I think so.

00:10:56   - Yeah.

00:10:57   - This comes in conjunction with some changes

00:10:59   coming in iOS 17.4.

00:11:01   It's like, boy, does 17.4 win the award

00:11:05   for the biggest point update ever to iOS?

00:11:07   Like, I think it might.

00:11:08   It's just, there's so much stuff going on.

00:11:10   But a change coming to the battery

00:11:13   and battery health screen.

00:11:15   Which remember, this was added several years ago.

00:11:17   There was the dust up about Apple throttling

00:11:21   the performance of phones.

00:11:22   There was a class action lawsuit.

00:11:24   All these things happened.

00:11:26   Now in 17.4, if you go into the battery screen and settings,

00:11:31   you will see two new menus.

00:11:34   One called battery health

00:11:36   and one called charging optimization.

00:11:38   And in that change, you can now see,

00:11:41   without having to drill down, your battery health status.

00:11:45   So if it says normal, then you're all set.

00:11:49   And if you want more detail, then you can tap in further

00:11:52   and you can still see the percentage, right?

00:11:54   So Mary actually just went through this

00:11:56   with her iPhone 13 mini,

00:11:58   where her battery life was really bad.

00:12:00   I went into the battery health thing

00:12:03   and it turns out she was at like 78% original capacity.

00:12:08   And that was low enough where the Apple store

00:12:10   replaced the battery for free for her,

00:12:13   under AppleCare, which was really cool.

00:12:15   This phone now has also has gotten a new screen

00:12:17   and a new battery.

00:12:18   It's like a whole new iPhone 13 mini.

00:12:21   But this makes it a little bit easier,

00:12:22   a little bit less digging to kind of see what's going on.

00:12:24   And I think that's good because before,

00:12:27   you just got a percentage without much context, right?

00:12:30   It's like, well, is 80% good?

00:12:32   Is 80% bad?

00:12:33   I don't know.

00:12:34   And so they're trying to, I think, better educate people

00:12:37   about how their batteries are actually doing.

00:12:41   - So nine to five Mac on fire right now.

00:12:44   - Yes.

00:12:44   - It's like left and right piece of information.

00:12:47   This one, they got through Waze,

00:12:51   some information about some sizing of the next iPads.

00:12:56   And I wanna specifically just talk about the iPad Pro

00:12:59   'cause it's the most interesting

00:13:01   because we have comparisons.

00:13:03   Essentially the piece of information to know

00:13:06   is that the iPad Pros are all gonna be getting

00:13:08   a little bit bigger by a few millimeters,

00:13:10   but also thinner.

00:13:12   So the 11 inch will be going from 5.9 to 5.1 millimeters.

00:13:19   And the 12.9 inch will be going from 6.4 to five millimeters.

00:13:24   What is really weird to me here is that the 12.9

00:13:31   will be 0.1 millimeters thinner than the 11.

00:13:35   - Yeah, that I think is gonna feel amazing in the hand.

00:13:40   - And also the 12.9 Pro is one millimeter thinner

00:13:45   than the new 12.9 Air.

00:13:48   - Yeah.

00:13:49   - That's also interesting.

00:13:50   - But the Air sticker and the Pro.

00:13:52   - Yeah, it's ironic, right?

00:13:55   - Yeah, it's the Air is thinner

00:13:58   than the current Pro is as well.

00:14:02   And I expect that's because it's maybe the mini LED,

00:14:06   maybe that is more complicated than just a regular LCD screen

00:14:11   or LED screen, whatever they use.

00:14:13   LED screens, I have no idea anymore.

00:14:15   I'm so lost because I assume that we use LED,

00:14:19   mini LED and OLED, right?

00:14:21   So I think that the LED screens will be thinner

00:14:24   than the mini LED screens, but what 9to5 is saying

00:14:28   that everything is gonna be thinner

00:14:30   because OLED is thinner in general.

00:14:34   So that's how they can get thinner.

00:14:35   - Yeah.

00:14:36   - But the fun thing then is that the iPad Pro,

00:14:40   I guess we'll take the crown again

00:14:42   because I think the iPad Pro currently,

00:14:45   or at least maybe the LED version of the iPad Pro

00:14:49   was the thinnest Apple product ever made.

00:14:51   And this will now top that.

00:14:54   - Yeah, I mean, five millimeters makes it thinner

00:14:57   than any iPod touch and even that weird last iPod Nano

00:15:02   that kind of looked like iOS.

00:15:03   Like not the square clippy one, but the very last one.

00:15:07   And that thing, I've got one, like I pulled it out today.

00:15:09   It's really thin.

00:15:11   I think it's gonna be a dramatic change to the iPad,

00:15:14   especially when you're not using it in the case.

00:15:18   - The thing that just always boggles my mind,

00:15:20   like I know this fact, but I had to Google it again

00:15:23   because I still feel like I can never believe it,

00:15:25   that the current iPad Pro is thinner

00:15:28   than the original iPod Nano.

00:15:30   Doesn't make any sense.

00:15:33   - No, that's incredible.

00:15:35   - Right?

00:15:35   The iPod Nano was 6.9 millimeters thick.

00:15:38   Wild, it's wild.

00:15:40   - Yeah, but going to five millimeters,

00:15:43   like that's a major milestone, I think.

00:15:47   Like it seems incredible that it can be that thin.

00:15:52   You know, go OLED, I guess.

00:15:54   So for these iPads, I guess, so for these iPads,

00:15:57   let's sort of run down what we know.

00:15:58   We know that there's gonna be a bigger iPad Air.

00:16:02   They are gonna stick to their usual sizes,

00:16:05   11 inch and 12.9 inch for the iPad Pro.

00:16:10   They're gonna be OLED,

00:16:11   and they're probably remaking the Magic Keyboard

00:16:14   or at the very least making a higher end,

00:16:18   more pro keyboard trackpad accessory.

00:16:21   So that's all we know for hardware.

00:16:24   I am really curious to see

00:16:28   how they're gonna pitch these devices beyond these changes.

00:16:32   Like, should we expect

00:16:36   any change from the software of these iPads?

00:16:41   And also, I had a thought today.

00:16:44   Why, so I guess, no, let me rephrase in a different way.

00:16:49   I think Sidecar on the iPad

00:16:53   should go away, hold on,

00:16:57   and should just be called Mac Virtual Display.

00:16:59   Because aren't they like the same thing?

00:17:02   Well, Sidecar extends the display,

00:17:04   and I guess Mac Virtual Desktop replaces it,

00:17:07   but they're definitely very closely related.

00:17:10   It just seems strange

00:17:11   that we have two similar Mac display technologies,

00:17:15   and one is called One Way,

00:17:16   and another is called Mac Virtual Display.

00:17:18   But I think it is that similarity,

00:17:20   the fact that they are similar, not the same,

00:17:22   is it's difficult, 'cause as Stephen said, right?

00:17:25   Yeah, I get it.

00:17:26   You can't mirror, I don't think, right?

00:17:29   Or maybe you can, but you can also, with Sidecar,

00:17:33   have it as a second display, right?

00:17:34   Which you cannot do with Vision.

00:17:36   True, true.

00:17:37   But regardless, anyway, some hardware changes,

00:17:41   and we know nothing about,

00:17:43   and if there's gonna be any software-related improvements

00:17:47   on these iPads, which I don't think there will be.

00:17:50   No, at least not when they announce them,

00:17:53   if we're still expecting a March timeline, right?

00:17:57   You're three months away from any changes.

00:18:01   I would be, I mean, we're gonna talk about iOS 18

00:18:05   in a little bit.

00:18:06   I would be surprised if there was anything

00:18:09   like super specific for iPad and iOS 18,

00:18:11   'cause it looks like Apple was one very, very clear focus,

00:18:15   but we'll find out.

00:18:17   Yeah.

00:18:18   I kinda wish that he had the measurements

00:18:22   for a new iPad Mini.

00:18:23   Can I just put that out there?

00:18:24   Those numbers aren't floating around.

00:18:26   Yeah, 'cause, Stephen, you know what?

00:18:27   It's gonna be exactly the same,

00:18:28   'cause if we get anything, it is a spec bump.

00:18:31   Yeah, I think I've said this,

00:18:33   but my iPad Mini got sort of down-cycled

00:18:36   to be a kid's iPad, 'cause they were using,

00:18:38   one of the kids was using an iPad, I don't know,

00:18:42   seventh gen, something really old,

00:18:44   and I replaced it with an iPad Mini,

00:18:45   and I got a refurbished 11-inch iPad Pro,

00:18:48   but I really miss the iPad Mini.

00:18:51   I kinda think I messed up, but.

00:18:53   I adore, adore my iPad Mini.

00:18:57   It's going to be very hard to resist

00:19:00   an OLED 11-inch iPad Pro.

00:19:02   Oh yeah, no, I'm gonna get that.

00:19:03   I think I am gonna get an 11-inch iPad Pro OLED for sure.

00:19:08   Like, that's gonna be my new iPad.

00:19:11   'Cause I use, so remember when I had COVID?

00:19:13   You remember when I had COVID?

00:19:15   I don't know if you guys could feel that.

00:19:15   Oh, you do, you do?

00:19:17   Remember I had COVID that one really inconvenient time?

00:19:19   Yeah.

00:19:20   To try and make myself feel better,

00:19:23   I ordered an iPad from Best Buy.

00:19:26   Oh yeah, 'cause you wanted to use Stage Manager,

00:19:28   and that was when it was still limited to,

00:19:32   whatever it was limited to.

00:19:33   The M processors, and I had an iPad

00:19:36   with an A processor in it.

00:19:38   And so I was like, oh, I'll just get that.

00:19:41   So we use that iPad for watching TV on,

00:19:46   if we're eating something, if we're eating dinner,

00:19:47   maybe we'll catch up on a show, and we watch that, right?

00:19:50   So that works for us.

00:19:51   But that is a 64-gigabyte iPad.

00:19:54   Boys, that is-

00:19:57   Not enough.

00:19:58   An impossible thing to manage.

00:20:00   Because I have like 28 gigabytes of iMessage.

00:20:05   And I can't do anything about it.

00:20:07   That's our fault, mostly.

00:20:09   Oh, but it's just like, I wanna keep it, right?

00:20:11   'Cause all of my other devices can store that.

00:20:13   But this cannot.

00:20:14   I had to do some real like gymnastics a couple of days ago,

00:20:19   just to update the iPad.

00:20:21   It was a nightmare, just trying to be able

00:20:24   to update the thing.

00:20:25   So yeah, but so I would like to,

00:20:27   I think I would like to replace that iPad at some point.

00:20:30   And I don't know, maybe it just becomes my iPad

00:20:33   and I give up on the mini life, but we'll see.

00:20:36   Oh, don't wanna do that.

00:20:37   Could I introduce you to a concept

00:20:40   called the multi-pad lifestyle?

00:20:42   It's on my face now.

00:20:43   That is true.

00:20:45   It's on my face.

00:20:46   Multi-pad on my face.

00:20:48   Yep.

00:20:49   This episode of Connected is brought to you by NetSuite.

00:20:56   If you have a business, you can probably relate to this.

00:20:59   Things get to a certain size

00:21:00   and then cracks start to emerge.

00:21:02   Things you used to do in a day are suddenly taking a week.

00:21:07   You have too many manual processes

00:21:08   and you don't have one source of truth.

00:21:11   It's never a good feeling

00:21:12   when everything in your business is disjointed.

00:21:15   Too many processes in too many places.

00:21:18   You want clarity.

00:21:19   You want one place where all the important stuff happens.

00:21:22   The solution to untangling that disjointed feeling

00:21:25   is NetSuite.

00:21:26   NetSuite is a software company

00:21:28   that has developed a cloud-based business management platform

00:21:32   to help your team deal with key business processes

00:21:35   like enterprise resource planning and financials,

00:21:38   CRM, e-commerce, inventory, and more.

00:21:42   I know if I have too many manual processes

00:21:44   running around at work,

00:21:45   it makes getting work done feel harder than it needs to be.

00:21:49   And if you want to make sure the cracks

00:21:51   don't emerge in your business,

00:21:52   there are three numbers you should know.

00:21:54   36,000, 25, and one.

00:21:58   There have been 36,000 businesses

00:22:01   that have upgraded to NetSuite by Oracle.

00:22:03   That makes NetSuite the number one cloud financial system,

00:22:06   streamlining accounting, financial management,

00:22:09   inventory, HR, and more.

00:22:11   And 25 because NetSuite turns 25 years old this year.

00:22:15   That's 25 years of helping businesses to do more with less,

00:22:18   close their books in days, not weeks, and drive down costs.

00:22:22   And one because your business is one of a kind.

00:22:25   So you get a customized solution for all your KPIs

00:22:29   in one efficient system with one source of truth.

00:22:32   Manage risk, get reliable forecasts, and improve margins,

00:22:36   everything you need to grow all in one place.

00:22:40   Having all the information you need in one place

00:22:42   makes it so much easier to make decisions.

00:22:45   I know as a business owner what a difference that can make

00:22:48   and how much easier everything operates

00:22:50   when information is available.

00:22:52   It really means that you can make smart decisions

00:22:55   and you can make them faster.

00:22:57   So go download NetSuite's popular KPI checklist

00:23:00   designed to give you consistently excellent performance.

00:23:03   It's absolutely free and you can find it

00:23:05   at netsuite.com/connected.

00:23:08   That's NetSuite, N-E-T-S-U-I-T-E,

00:23:13   netsuite.com/connected to get your own free KPI checklist

00:23:18   and to make sure your business is one

00:23:20   that continues to thrive.

00:23:22   Our thanks to NetSuite for their support of the show

00:23:24   and Relay FM.

00:23:25   - So Jason published the Six Colors report card for 2023

00:23:33   over on sixcolors.com, the website,

00:23:35   and we spoke about it on upgrade

00:23:39   and me and Jason both gave our scores.

00:23:41   But one of the great things about this now

00:23:43   is people start publishing their own report cards,

00:23:46   all of the information that they gave.

00:23:48   And so you two both published yours as well.

00:23:51   And so I wanted to give you both the opportunity

00:23:54   to pick out any specific highlights.

00:23:56   We probably won't go through it all.

00:23:57   I don't think we have enough time in this week's episode

00:23:59   to go through everything.

00:24:01   But I don't know if either of you wanted to pick out

00:24:03   any specific scores that you gave and some reasoning

00:24:06   for those.

00:24:07   - Yeah, my one out of five to the iPad, maybe.

00:24:10   - Okay, yeah, I'd like to know about that.

00:24:12   I also want to know about your wearables

00:24:14   because our scores are very similar.

00:24:17   - What did I say about that?

00:24:18   - Wearables you gave two out of five

00:24:20   and I gave four out of five.

00:24:21   So I also want to hear about that.

00:24:24   - Yeah, so the wearables, I'm kind of bored

00:24:27   with the AirPods Pro.

00:24:28   They are serviceable.

00:24:30   They do what they're meant to do,

00:24:32   but it just feels like they've been doing

00:24:34   the same thing forever now.

00:24:35   And I know that in tech years, forever,

00:24:39   can also be three years, four years,

00:24:42   but it feels that way.

00:24:43   They feel, I don't know, they feel kind of boring to me.

00:24:46   I cannot stop using them because they are the best earbuds

00:24:50   if you are an Apple person.

00:24:54   But I just wish that Apple did something new.

00:24:56   More sensors, more, I don't know, more battery life,

00:25:00   more, you know, a different shape.

00:25:02   I look, for example, at,

00:25:04   have you guys seen those new Bose Ultra open earbuds

00:25:09   that actually don't go into your ears?

00:25:12   - Yeah.

00:25:13   - Yeah, I don't know.

00:25:14   I don't think Apple is ever gonna do that sort of stuff,

00:25:17   but I just wish they had a little more creativity

00:25:20   in terms of the whole wearable category.

00:25:22   And of course, the AirPods Max.

00:25:24   They should be a little ashamed

00:25:27   that they haven't upgraded those headphones

00:25:31   in any capacity, both in terms of hardware and software

00:25:35   for the new audio features that we have on the AirPods Pro

00:25:39   in how many years?

00:25:41   Four, five?

00:25:42   So yeah, I think it's a two out of five

00:25:46   because I wanna see some changes,

00:25:48   and the AirPods Max just feel like a forgotten product

00:25:52   at this point.

00:25:52   - Which is a shame 'cause-- - Yeah.

00:25:54   - They're the, my favorite over is

00:25:56   because they get all of the features

00:25:58   that come with being as part of the ecosystem.

00:26:00   - And they are so comfortable.

00:26:02   Like they are the best over here Bluetooth headphones

00:26:05   that I have.

00:26:06   I really, really like them,

00:26:07   but they make me feel bad when I use them.

00:26:09   You know, when I need to charge them

00:26:11   with a lightning cable, just feels terrible.

00:26:14   - Yeah, and I want, I mean, because I use it,

00:26:16   I want adaptive mode.

00:26:18   I want conversations, like I use those features,

00:26:21   and so I want them on all of my headphones.

00:26:25   - So the iPad, I gave it two out of five.

00:26:29   You gave it one out of five.

00:26:32   I'm sure for similar reasons, right?

00:26:34   No hardware updates in 2023,

00:26:37   but also kind of no meaningful iPadOS updates.

00:26:41   I mean, they improved Stage Manager, but--

00:26:43   - Yeah, so that's why I give it a one.

00:26:44   I was really tempted to give it a zero,

00:26:46   but because they, at the very least,

00:26:51   they sort of improved some basic problems of Stage Manager,

00:26:56   I gave it a one.

00:26:59   - Well, there is, for JSON, there is no zero.

00:27:02   - Oh, okay.

00:27:04   - 'Cause that was my, I gave it a one,

00:27:06   instead of, I said in my report notes,

00:27:09   if there was a zero, I would've given it a zero.

00:27:11   - Yeah, yeah.

00:27:13   So it's not just that there weren't,

00:27:17   for the first time in over a decade,

00:27:19   no new iPads last year.

00:27:22   It's the fact that it really seems, okay,

00:27:26   seems from the outside that if you set aside the fact

00:27:31   the final cut and logic finally came to the iPad,

00:27:34   but those are like separate apps,

00:27:35   but as a platform, right, as iPadOS, as its own entity,

00:27:40   as a whole operating system,

00:27:42   it just feels, my perspective, that Apple's,

00:27:47   I don't know, it almost seems like they don't have,

00:27:50   there's no passion behind it anymore,

00:27:52   and it seems to me like that passion

00:27:54   is now going to VisionOS.

00:27:57   It just, which is why I keep saying,

00:28:01   like, I really wanna see how they pitch,

00:28:03   how they present these new iPad Pros,

00:28:06   because if it just, I get this feeling that,

00:28:10   besides Stage Manager, which, again,

00:28:12   is an optional setting that you need to choose,

00:28:15   and it's not even that great, right?

00:28:17   I've been using my 11-inch iPad Pro

00:28:20   with Split View again and Slide Over, and it's glorious.

00:28:23   But besides Stage Manager,

00:28:27   we haven't really gotten any meaningful,

00:28:31   like, desktop-class features on iPadOS,

00:28:35   unless you wanna talk about, like,

00:28:37   slightly customizable toolbars and sidebars

00:28:41   as desktop-class features,

00:28:43   which is what Apple would like you to think,

00:28:45   which, fine, sure,

00:28:46   it's nice to have a customizable toolbar,

00:28:48   but it seems to me like Apple's really putting the passion,

00:28:53   so to speak, behind VisionOS and behind macOS these days.

00:29:01   And, of course, iOS, but iOS is going forever, right?

00:29:05   iOS needs to go forever.

00:29:07   But besides iOS,

00:29:09   which is the central operating system in many ways,

00:29:13   it feels like there's VisionOS, macOS,

00:29:17   really, really...

00:29:19   I mean, look at Mac hardware, look at Mac software, right?

00:29:22   Tons of things happening over there,

00:29:25   and iPadOS is sort of just left there to linger

00:29:29   and to wait for small updates every once in a while.

00:29:33   And I say this as a person who profoundly loves

00:29:36   the iPad form factor and the idea of a tablet OS,

00:29:42   but looking at reality,

00:29:45   is there anything really substantial happening

00:29:51   on that front these days?

00:29:53   And I think the answer is no and has been no for a while.

00:29:56   Yeah, it's hard to argue with.

00:29:59   My score that I wish I could do over is the Apple Watch.

00:30:03   I don't know what I was thinking.

00:30:04   I gave it four out of five, and I think that's too high.

00:30:08   I didn't even write it in my thing.

00:30:10   I was going through this in preparation of the show.

00:30:12   I was like, "I didn't even write anything about this."

00:30:14   The Apple Watch, I guess some of this

00:30:16   was the very end of the year,

00:30:17   maybe even after I turned to my answers.

00:30:19   Maybe that's what it was.

00:30:21   But the whole deal with the blood oxygen sensor

00:30:23   in the United States is ridiculous.

00:30:26   And I think if we look at 2023 as a whole,

00:30:31   that is a pretty big mark against the Apple Watch.

00:30:34   And now, and of course, now it's the 2024,

00:30:37   so this will be on next year's, I guess.

00:30:39   You can't, like if you buy a new Apple Watch

00:30:41   that supports blood oxygen stuff,

00:30:43   it's turned off in the United States

00:30:45   because Apple can't or won't settle with this other company.

00:30:50   And the longer it goes, the more ridiculous it is.

00:30:54   Like, I don't know what that company wants from Apple,

00:30:57   but they need to be able to come to a common ground here

00:31:00   because it's a good feature that people are being,

00:31:05   they're not able to use it,

00:31:07   they're not able to take advantage of it

00:31:08   because this litigation that's going on

00:31:11   that most people don't even know anything about.

00:31:13   - Yep, it's one of those things that I was very aware of

00:31:16   and then completely forgot about

00:31:18   because now we're not talking about it anymore

00:31:20   and there's no more information about it

00:31:21   because it reached a point and stopped.

00:31:23   And like, there will be more in the future,

00:31:25   but it is kind of ridiculous to think about that.

00:31:29   Like, I hope it gets resolved 'cause it's petty,

00:31:36   it feels just like such a small thing.

00:31:41   And I understand why they don't wanna do it,

00:31:44   but if they infringe the patents of this company,

00:31:48   well, then that's just the way it goes, right?

00:31:51   Like if that's how it's found, that's just the way it goes.

00:31:54   - Yep, I think those are my sort of high points.

00:32:00   You know, I think overall, Apple's doing okay.

00:32:05   I think the pain points, most of them aren't new, right?

00:32:10   Like Federico's thing about iPadOS,

00:32:12   you could copy that out and paste it

00:32:14   in an episode of "Connected" from two years ago

00:32:16   or three years ago.

00:32:18   That's basically the same argument, right?

00:32:20   I think the same thing could be said for hardware,

00:32:23   that the hardware, like each year is better and better

00:32:26   and Apple Silicon's been great for the Mac for now.

00:32:29   Coming up on a little over three years,

00:32:32   iPhone hardware, incredible.

00:32:34   I feel like these things that are changing

00:32:38   in sort of the Apple landscape, they change very slowly.

00:32:42   And I feel at least like the overall shape

00:32:46   of the report card and the way I think about it,

00:32:49   like not that much has changed

00:32:51   over the years we've been doing it.

00:32:53   Not drastically, at least.

00:32:55   - Yeah.

00:32:55   - No, there've been a couple of areas like the Mac.

00:33:00   So in 2016, the Mac was a 2.6.

00:33:05   And now it's a 4.2.

00:33:06   - Yeah, I forgot it went back that far even.

00:33:09   Yeah, that's fair.

00:33:11   - Right, so there are a few things like that

00:33:13   that have really swung.

00:33:14   The iPad has taken a similar dip.

00:33:16   It's now at 2.4.

00:33:17   - The iPad is now lower than where the Mac was

00:33:22   during the whole butterfly keyboard era.

00:33:25   - Like Apple Watch goes up and down,

00:33:26   but not really huge swings and similar for Apple TV.

00:33:30   But like the Mac and the iPad is the ones

00:33:32   that have had the biggest like deltas

00:33:35   I think during this time period.

00:33:37   - Yeah, I mean, gosh, what a terrible time

00:33:41   the Mac was having in 2016, '17 and '18, nightmare.

00:33:46   - I mean, it's what it's like to be an iPad user today.

00:33:50   - I guess so.

00:33:51   - They flipped.

00:33:52   - At least your keyboards work, you know?

00:33:55   - Apple doesn't believe in the future of the iPad.

00:33:58   They just want the Mac to be an iPad.

00:34:01   - Well, I guess it's like, all right,

00:34:02   so let's just play this out for a second, right?

00:34:05   - Yeah.

00:34:06   - Now, I guess the story will be that like

00:34:09   how the iPad was gonna replace the Mac

00:34:12   that Vision will replace iPad, right?

00:34:14   I guess that's probably where that story could go.

00:34:19   - For sure.

00:34:20   - I think so.

00:34:21   - Interesting, be fun for you guys.

00:34:24   - Yeah, well, I don't think it's not even that wrong

00:34:29   as an assessment.

00:34:32   I think just looking at the facts,

00:34:34   the fact that Vision OS is based on iPad OS

00:34:39   and lets you open as many windows as you want,

00:34:42   I know that by design,

00:34:43   the fact that you can turn your own room into a workspace,

00:34:48   but it just feels like a version of iPad OS

00:34:55   where people actually improved the things

00:35:00   that were not working on iPad OS.

00:35:02   Yeah, so.

00:35:04   - This episode of Connected is made possible by Nom Nom.

00:35:10   Don't settle when it comes to your dog's health.

00:35:12   Switch to real, gently cooked dog food

00:35:15   with ingredients you can actually see.

00:35:18   Nutrient-packed recipes designed

00:35:20   by board-certified veterinary nutritionist,

00:35:22   freshly made and delivered to your door.

00:35:25   Nom Nom delivers freshly made dog food

00:35:27   with every portion personalized to your dog's needs

00:35:31   so you can bring out their best.

00:35:33   And it's made with real wholesome ingredients

00:35:35   that you can both see and recognize

00:35:37   without any additives or fillers

00:35:39   that contribute to bloating and low energy.

00:35:42   And that's because Nom Nom uses the latest science

00:35:45   and insights to make real good food for dogs.

00:35:49   Their nutrient-packed recipes are designed

00:35:51   by board-certified veterinary nutritionist,

00:35:53   freshly made and shipped free to your door.

00:35:57   It is so important to make sure

00:35:59   that your dog has the nourishment that they need.

00:36:02   They're valued members of our families

00:36:04   and their needs change over time.

00:36:07   And Nom Nom is ready to help you make sure

00:36:10   you get the best out of your furry friend.

00:36:12   In fact, they've already delivered over 40 million meals

00:36:16   to good dogs like yours,

00:36:17   inspiring millions of clean bowls and tail wags.

00:36:21   Plus Nom Nom comes with a money-back guarantee.

00:36:23   If your dog's tail isn't wagging within 30 days,

00:36:26   Nom Nom will refund your first order.

00:36:29   No fillers, no nonsense, just Nom Nom.

00:36:33   Go right now to get 50% off your no-risk two-week trial

00:36:38   at trynom.com/connected.

00:36:43   That's spelled trynom, N-O-M.

00:36:46   Trynom.com/connected for 50% off.

00:36:51   Our thanks to Nom Nom for their support of the show.

00:36:54   - So Steven, I feel like we have Mac Power Users'

00:36:59   version of you for the rest of the show today.

00:37:02   Like these show notes,

00:37:04   these feel like what you would do for MPU.

00:37:06   So I'm very excited about this.

00:37:08   There's a lot of hack it research going into this topic.

00:37:11   So on behalf of the Connected podcast,

00:37:14   the most pleasurable podcast, I would like to thank you

00:37:17   for doing this pleasurable research.

00:37:20   - It's also in Notion, which is like MPU.

00:37:24   - Yep.

00:37:25   - Okay, so I wanted to spend some time talking

00:37:30   about Apple's AI efforts.

00:37:32   And I've kind of broken this into three sections roughly,

00:37:36   background, rumors, and what they've actually shipped.

00:37:41   And you know, at this point,

00:37:41   it's like a really interesting point in time.

00:37:44   Google, OpenAI, others have kind of put their cards

00:37:47   on the table about what they're doing.

00:37:49   And Apple hasn't in terms of product.

00:37:53   We don't know what AI infused product is gonna be like

00:37:57   from Apple, but they are doing things.

00:38:01   And when I started digging into this,

00:38:04   'cause like you see reports every once in a while,

00:38:05   like Apple released this, they published this paper,

00:38:07   whatever, when you start putting it together,

00:38:10   there's actually a lot of stuff here.

00:38:12   And so while Apple hasn't shipped anything

00:38:14   to the regular end user, that doesn't mean

00:38:17   that there's not stuff going on behind the scenes.

00:38:20   And if anything, all this makes me excited

00:38:22   to see what we could end up with in products,

00:38:25   because clearly they're spending a lot of time and money

00:38:28   in this area, which is good,

00:38:30   because all the competitors are, they should.

00:38:32   - To say follow up, obviously, when we're talking about AI,

00:38:36   what we're talking about is products like the products

00:38:39   that are existing now from the companies

00:38:41   you mentioned earlier.

00:38:42   Obviously, we know Apple has machine learning

00:38:45   in a lot of their things.

00:38:46   We know they put a transformer model in the keyboard,

00:38:49   but none of these are what we're really thinking about

00:38:52   when we talk about quote unquote AI products, right?

00:38:55   These are chatbots, these are things

00:38:57   that do things on the web, like you can have conversations

00:39:00   with all this image generation, all this kind of stuff

00:39:03   that is occurring new right now.

00:39:07   That's what we're talking about.

00:39:09   - Exactly, I think we should start with personnel.

00:39:13   So there was some reporting coming up on a year ago

00:39:18   about some job descriptions that we're talking about,

00:39:22   hiring people in these areas.

00:39:25   And I had kind of forgotten that it's been this long,

00:39:30   but back in, what is it, back in 2018,

00:39:34   Apple hired John Gendrea from Google,

00:39:39   and he is now Apple's Senior Vice President,

00:39:45   comma, Machine Learning and AI Strategy.

00:39:49   He is on the leadership page on apple.com.

00:39:52   And when he came over, obviously a big deal,

00:39:57   he had done a lot of this sort of groundwork at Google.

00:40:02   And I think it's clear that he has been building a team,

00:40:05   not only through these job postings,

00:40:08   but exactly what you said, Mike.

00:40:09   They've already shipped some products,

00:40:11   I'm glad you brought that up.

00:40:12   And I would imagine there's even stuff

00:40:15   that we don't know about, but that is machine learning

00:40:19   or AI and services or in Siri or something

00:40:22   that they haven't really publicized, so.

00:40:24   - Man, talk about a guy who was hired

00:40:27   and then his life changed, right?

00:40:30   Like, I'm sure he was hired for whatever he was hired for

00:40:33   and he was working on that stuff.

00:40:35   And then like two years ago, it was like,

00:40:36   oh, you need a lot more resources now, right?

00:40:41   Like that, I assume his team has exploded

00:40:46   over the last 12 to 18 months.

00:40:49   - I'm sure it has.

00:40:50   And that brings us to some of the rumors.

00:40:53   So we mentioned iOS 18 earlier,

00:40:57   and it really seems like this is going to be a release

00:41:02   that is really focused on generative AI features.

00:41:07   Mark Gurman has talked a lot about this.

00:41:09   He has called it one of the biggest updates ever to iOS,

00:41:12   which is just wild.

00:41:14   There's also reports and rumors that the iPhone 16 line

00:41:19   could get a big bump in neural engine performance.

00:41:23   There's also a rumor floating out there

00:41:25   that the iPhone 16 and 16 Pro will both get a new chip,

00:41:29   that they won't do the thing

00:41:30   they've done the last couple of years.

00:41:32   And if that means new neural engine stuff to all the phones

00:41:36   in conjunction with iOS 18,

00:41:38   like Apple hardware, software services coming together,

00:41:42   right, that's it, that's what they do.

00:41:44   And I think some of these specific rumors

00:41:48   are actually really interesting in what it could mean.

00:41:52   One of the biggest ones has to do with Siri,

00:41:56   where you could use Siri,

00:41:59   hypothetically if these reports are correct,

00:42:02   to kind of do one-off tasks

00:42:03   that maybe you would build a shortcut for previously,

00:42:07   but shortcuts isn't very good, in my opinion,

00:42:09   at the one-offs, right?

00:42:11   You kind of build a shortcut

00:42:12   because you're going to do it over and over.

00:42:13   Same thing with automation on the Mac, right?

00:42:15   I'm not going to make a hazel rule to move MP3s one time.

00:42:19   I'm going to build a hazel rule to move MP3s every single day

00:42:22   'cause I'm making files every day for work, right?

00:42:25   And if this is true and Siri is going to be supercharged

00:42:29   with some sort of LLM,

00:42:31   where you could ask it to do something

00:42:33   based on what's on screen or files you provided

00:42:36   or pictures you provided, something like that,

00:42:39   it kind of in a way like opens automation, in air quotes,

00:42:43   to the masses in a new way,

00:42:45   just like shortcuts did years ago.

00:42:47   And I think that's really interesting.

00:42:49   And I'd love to hear what y'all think about,

00:42:50   like what Siri could look like with these new superpowers.

00:42:55   Yeah, I think there's so much incredible potential

00:43:00   when you start thinking about Siri and shortcuts together

00:43:04   as sort of a service that can understand

00:43:08   what you're asking in terms of features

00:43:10   of the operating system and actions to take

00:43:14   around the operating system and inside apps,

00:43:18   which is kind of similar, I guess,

00:43:19   to what the shortcuts folks are building now

00:43:23   as a separate startup, right?

00:43:25   Remember, what's the name of like a--

00:43:27   - Computer software incorporated.

00:43:30   - Computer software.

00:43:31   Yeah, software Inc, something.

00:43:34   But that's the idea, right?

00:43:35   So I think the biggest advantage that Apple has

00:43:39   compared to say OpenAI, right?

00:43:42   Or these other companies that are doing generative AI,

00:43:45   they have an entire OS or rather an entire ecosystem

00:43:50   of operating systems that can talk to an AI

00:43:55   and provide your own data.

00:43:58   That for me has been the biggest obstacle

00:44:00   in terms of really using something like chat GPT

00:44:03   as a service on a daily basis.

00:44:06   Chat GPT has no idea what's in my,

00:44:08   sure, there are plugins now, blah, blah, blah.

00:44:10   - You wrote one, you did SGDPT.

00:44:13   - Yes, but the idea is the AI that,

00:44:17   imagine if the AI lived inside my iPhone,

00:44:20   inside my computer and knew about my files,

00:44:23   knew how to say open a PDF and resize it,

00:44:27   or stuff like, hey, take the latest five screenshots

00:44:31   and zip them up and send them to John on iMessage.

00:44:34   Like that's the kind of stuff you would previously

00:44:37   write a shortcut for.

00:44:38   But as Steven said, that was a one-off task.

00:44:41   Do you really wanna write a shortcut for a one-off task?

00:44:44   What if you could just write that down as a request

00:44:47   or speak it as a spoken command

00:44:50   and the system did it for you?

00:44:52   Now that's, I think, the power of generative AI

00:44:56   and a large language model that understands

00:44:58   what you're asking and knows how to act on your own data.

00:45:03   That for me is the key point here.

00:45:05   - Is my worry with this, right?

00:45:07   So that's a great thing that you suggested

00:45:10   and I'm sure they could probably use like Intents

00:45:12   as a way to power some of that, right?

00:45:14   Like I have some of the bones that they would need

00:45:17   for like little things that an app can do.

00:45:20   The thing that I worry about is like, oh, save a task to,

00:45:26   you know, remind me to do the same, save a task to this.

00:45:30   I don't wanna say with Todoist, right?

00:45:34   - To Daoist, you mean.

00:45:35   I think it's pretty good.

00:45:36   - Add lunch with Federico to Fantastical, you know?

00:45:41   And that's the way they've done this kind of stuff

00:45:44   in the past.

00:45:45   You can't just say what apps you wanna use

00:45:49   like to the system and it will remember them.

00:45:52   You have to like make it clear,

00:45:54   like I'll save this thing to Todoist,

00:45:57   to remember the milk, to OmniFocus.

00:46:00   And like that's gonna be too much friction in there

00:46:04   where this, what they could implement,

00:46:07   what I hope they do implement is that this model

00:46:10   that they create can have a memory.

00:46:12   - Yes, I was just about to say this model

00:46:14   should have a data storage for your own preferences.

00:46:18   - This is something that I know ChatGPT is starting to do now

00:46:21   that you will be able to give it information about you

00:46:24   and it will remember it.

00:46:26   Like, and it will remember it across all of the chats

00:46:28   that you have, not just like the singular ones.

00:46:30   So like it's an actual real memory that it will have

00:46:34   rather than just needing to reread everything

00:46:36   you've ever said to it to remember something.

00:46:38   So like stuff like that would be good.

00:46:40   And that is very important that they do it that way.

00:46:45   Like if they want this thing to be truly useful

00:46:47   that we get that.

00:46:48   And this is the beginning of that dream

00:46:51   of like the knowledge navigator all over again, right?

00:46:55   So you can have an actual conversation with the computer

00:47:01   it will understand you, give you information

00:47:04   and then you can ask the computer to do things for you.

00:47:06   Like we are potentially very close now to having that

00:47:11   but they really have to make some fundamental decisions

00:47:15   early and correctly for us to get there, you know?

00:47:19   - And my other question is, and I guess, you know

00:47:23   or we can mention this because it's in Steven's

00:47:26   extensive list of links here.

00:47:30   Should you just, you know, are these features

00:47:32   these requests that you may ask a brand new serial AI

00:47:36   just going to be spoken requests

00:47:38   or will you be able to type them out as in a chat bot?

00:47:43   And one of the latest reports that we've heard

00:47:46   is that spotlight is also going to be infused

00:47:49   with these new AI features, which makes total sense, right?

00:47:53   There's times when you're working

00:47:54   and you want to get assistance from an AI

00:47:57   and you don't want to speak to Siri all the time.

00:48:00   So if Apple can do with a brand new spotlight

00:48:04   a version of essentially chat GPT

00:48:06   where you just have a written conversation

00:48:10   that you type out and you get responses from the AI

00:48:15   and you get results from the AI

00:48:17   I think that's going to be excellent.

00:48:18   So that, you know, you can have

00:48:21   this is the sort of advantage again that Apple has

00:48:23   in terms of like this multimodal interaction system

00:48:27   where the same AI could be used from spotlight

00:48:30   could be used with Siri on your phone

00:48:33   with Siri on your Mac, when you're wearing AirPods

00:48:36   when you're using the Apple watch,

00:48:37   like all those places where you

00:48:40   maybe even from the software keyboard

00:48:42   like all those places where Apple could in theory

00:48:45   put in AI features are the sort of advantages

00:48:48   that other companies don't have

00:48:50   because they don't own your computer or your phone.

00:48:54   - Yep.

00:48:56   Another area of reporting that I think is really interesting

00:49:00   and I think a really obvious direction for Apple to move

00:49:03   into is a co-pilot like feature in Xcode.

00:49:07   So this is been rolled out

00:49:09   across a bunch of Microsoft products.

00:49:11   You can get it in Office 365.

00:49:12   You can get it in GitHub.

00:49:14   I think it was there first

00:49:15   where the large language model can predict

00:49:20   and finish writing code for you.

00:49:23   So the language model knows all about Swift

00:49:26   and Swift UI and Objective-C and UIKit and everything else.

00:49:31   Maybe they'll even teach you about AppKit.

00:49:32   You never know.

00:49:33   And they might go that far back

00:49:34   and it then can help assist you in writing your code

00:49:39   and even maybe potentially find bugs in your code

00:49:43   or suggest more efficient ways of doing something.

00:49:47   This seems like something that they've got to do.

00:49:51   This is in other IDEs.

00:49:53   This is in other applications.

00:49:57   And I suspect out of everything on this list,

00:49:59   if we see anything, we would see this first.

00:50:03   - I think the time has passed to just do this only.

00:50:06   Like this felt like we were talking about this last year

00:50:09   at WWDC as a feature they might do.

00:50:12   - Yeah.

00:50:13   - Like, I think if this was something,

00:50:15   like you said, like, yeah, they should do this,

00:50:17   but this now just needs to be one of the many things.

00:50:20   - Yes.

00:50:21   - The time has passed where this is the only thing

00:50:23   you can get away with.

00:50:24   - And it's an area where the specific knowledge set

00:50:29   is smaller than something generalized like Siri, right?

00:50:33   Siri, if it's gonna do all the things we want it to do,

00:50:35   it's got a lot of stuff.

00:50:37   And if it's fenced in basically around Xcode,

00:50:42   then something like this could be really useful, right?

00:50:48   The knowledge, the domain knowledge is smaller

00:50:51   than anything someone might do in an iPhone, right?

00:50:54   So I do expect that we'll see this.

00:50:56   - I'd be really curious to, you know,

00:50:59   we'll revisit this topic in June.

00:51:01   A couple of things on my mind.

00:51:03   One, is this gonna be Swift and SwiftUI only?

00:51:06   Did they train the model on only Swift or also Objective-C?

00:51:11   And two, did they train the model

00:51:13   on private Apple internal code?

00:51:18   Because the thing about GitHub copilot

00:51:21   is that it's trained on open source code

00:51:23   from GitHub, by and large.

00:51:25   I expect Apple didn't follow that route.

00:51:29   And I mean, where does the code that they use

00:51:33   to train the Xcode copilot come from?

00:51:36   And it'd be interesting to see if, you know,

00:51:40   maybe the AI tool in Xcode autocompletes code for you

00:51:44   that was actually written by an Apple engineer

00:51:46   for like, say, Springboard or something.

00:51:49   I'd be curious to know the source.

00:51:50   With this entire initiative, I am really curious

00:51:54   to know the source of Apple's data.

00:51:56   There were also the reports, right?

00:51:58   That Apple was gonna try and strike licensing deals

00:52:01   with publications to train the source of content

00:52:05   for generative AI.

00:52:07   There is also the rumor going around this past week

00:52:10   that Reddit was shopping around a licensing deal

00:52:13   of $60 million to basically say,

00:52:18   "Hey, company, we're gonna give you access to our data.

00:52:21   If you wanna train your generative AI model,

00:52:23   give us $60 million in return."

00:52:26   So maybe Apple could be behind that deal, we don't know.

00:52:29   But I really wanna see how Apple trained

00:52:34   all these different models in all the different domains.

00:52:37   - I remember a report from some time ago,

00:52:40   I think it was a Montgomery report,

00:52:42   that when they were building or starting to build out

00:52:44   their models, I think it's called Ajax internally,

00:52:48   they were really concerned about training it

00:52:50   on their own data.

00:52:52   Like Apple did not want to do that.

00:52:54   And I was concerned internally that it would

00:52:57   give away secrets basically.

00:53:00   So I wouldn't expect this is trained on,

00:53:03   I expect that this will be trained the same way

00:53:05   that they trained the photos machine learning model,

00:53:07   which is they have bought as much data as they possibly can.

00:53:11   So could I imagine they've done a deal with Reddit?

00:53:13   Yes, and I imagine if they've done this,

00:53:15   they've done deals with as literally as many people

00:53:17   as they possibly can to get as much information

00:53:20   as they possibly can.

00:53:21   Which if they have done that,

00:53:23   it does still beg the same questions

00:53:25   that a lot of companies have, which is like,

00:53:27   it's quote unquote, fair use, which I don't agree with.

00:53:31   But this is the world that we live in today.

00:53:33   - Yeah, that's a whole complicated sidebar.

00:53:36   I did sell Apple the entire Mac stories archive

00:53:39   for AI training though.

00:53:41   - Okay, cool.

00:53:42   How much?

00:53:43   - It was like eight bucks.

00:53:46   - Cool, cool.

00:53:47   - The model's gonna know a lot about shortcuts,

00:53:51   but it's not gonna be able to do any of those things.

00:53:53   - I mean, what's new?

00:53:54   The last sort of AI rumor that I think is worth

00:54:00   touching on today is an AI wellness coach.

00:54:05   So the goal would be to keep users motivated

00:54:09   to exercise and monitor their eating habits.

00:54:13   - Clippy, but for workouts.

00:54:14   - Yep.

00:54:16   - Clippy, but for workouts.

00:54:17   - Maybe flippy, 'cause it's like pretty acrobatic.

00:54:20   - Or rippy, so you get strong, you know?

00:54:22   - Yeah, rippy, that's it, that's the one.

00:54:27   Get rippy, baby.

00:54:28   - Get rippy, it's time to get rippy.

00:54:31   Get rippy with Team Cook.

00:54:33   - From the beginning,

00:54:34   the Apple Watch has made suggestions, right?

00:54:36   It's like, hey, stand up.

00:54:37   Hey, if you go, the most frustrating one is like,

00:54:40   it's 11 p.m., hey, if you went for a quick 30 minute walk,

00:54:43   you'd close your ring.

00:54:44   It's like, it's 11 p.m., bro, I'm not going outside.

00:54:47   This, I think, out of all of them, is the most delicate.

00:54:50   I think you've really got it.

00:54:52   You really don't want this AI to hallucinate

00:54:54   and tell somebody to do something they shouldn't do

00:54:57   when it comes to health and wellness,

00:54:59   but we'll see if Apple crosses this bridge.

00:55:01   I'm the most dubious about this one, I think,

00:55:03   actually showing up in the world.

00:55:05   But this one though, this one sells Fitness Plus.

00:55:09   Like this one sells a service, even.

00:55:12   The rest of them, boo.

00:55:14   There's no money in that.

00:55:15   - Imagine if the AI wellness coach

00:55:17   is like a virtual persona

00:55:22   that talks to you. - It could be.

00:55:24   - And he's as creepy as a persona.

00:55:27   - You know, I just had the thought,

00:55:29   it's gonna be really interesting to understand

00:55:32   where and how this information's processed.

00:55:36   Is it, could it even be all on device?

00:55:39   Is there gonna need to be some cloud stuff?

00:55:41   I think there's gonna need to be a mixture.

00:55:43   Maybe that you could, if you're an Apple One subscriber,

00:55:46   you might get priority or speed increases.

00:55:50   - Over your-- - I don't wanna think

00:55:52   about that stuff, well, yeah. - I'm telling you, man.

00:55:54   - But yeah, you're right.

00:55:55   Yeah, this is how this stuff works, yeah.

00:55:57   - The more I hear about this stuff,

00:56:00   the more I understand, it's just raw power.

00:56:02   And I think there's gonna be a lot of these things

00:56:05   where they will not be able to power it on device.

00:56:08   It has to be powered in a data center somewhere.

00:56:11   That's gonna be really interesting if they do that

00:56:13   because they've spent so much time over the last few years

00:56:16   about things that are on device, on device.

00:56:19   And maybe some of it will be able to, but I just,

00:56:22   I don't, you know, looking at the amount of data

00:56:26   that's processed for, say, chat GPT,

00:56:29   and it all needs to be done in these huge data centers,

00:56:32   I can't imagine the chip in my iPhone

00:56:34   is really gonna be able to be comparable with that, but.

00:56:37   - That's gonna be a lot of carbon to offset for Apple.

00:56:41   - Yes, yep. - That'll be interesting.

00:56:44   - This episode of Connected is brought to you

00:56:48   by Squarespace, the all-in-one platform

00:56:51   for building your brand and growing your business online.

00:56:54   With Squarespace, you can build a beautiful website,

00:56:57   engage with your audience, and sell anything.

00:57:00   Squarespace has everything you need all in one place.

00:57:04   I love building on Squarespace.

00:57:05   In fact, I've been working on a couple of new websites

00:57:08   for some of Underscore's apps.

00:57:11   Building it atop Squarespace makes it really easy.

00:57:14   With the Fluid engine, I can design exactly what I want

00:57:17   for both desktop and mobile users.

00:57:20   It's all drag and drop, and it's really easy

00:57:23   to take what I have in my mind and put it on the page.

00:57:27   And organizing the assets for a website,

00:57:30   that can be a pain, right?

00:57:32   You have a lot of pictures.

00:57:33   Maybe you have a lot of app icons you wanna show.

00:57:35   Well, the new Squarespace asset library makes it easy

00:57:37   to upload, organize, and access all of your content

00:57:41   from one place.

00:57:42   It's this one central hub that you upload your files

00:57:46   to your images, and then you can use them

00:57:47   across the Squarespace platform.

00:57:50   Plus, it has built-in analytics.

00:57:52   You can see where your site visits and sales are coming from,

00:57:55   analyze which channels are most effective,

00:57:58   and improve your website building a marketing strategy

00:58:01   based on your top keywords

00:58:03   or the most popular products and content.

00:58:05   Again, all of this is in one place.

00:58:07   You're not bolting a bunch of different things together.

00:58:10   Squarespace has it all under their roof,

00:58:13   and it means that you can go from zero

00:58:15   to a really awesome website in not that much time.

00:58:18   And if you have an older website,

00:58:20   Squarespace is a great place to rebuild it.

00:58:23   I've taken that project on for a bunch of people

00:58:26   over the years.

00:58:27   Taking an old, creaky property

00:58:29   and building a nice new one on Squarespace

00:58:31   is very rewarding.

00:58:33   So go to squarespace.com/connected for a free trial,

00:58:38   and when you're ready to launch,

00:58:39   use the code connected to save 10% off your first purchase

00:58:43   of a website or a domain.

00:58:44   That's squarespace.com/connected and the code connected

00:58:48   when you decide to sign up to get 10% off

00:58:50   your first purchase and to show your support for the show.

00:58:54   Our thanks to Squarespace for their ongoing support

00:58:56   of Connected and all of Relay FM.

00:58:58   So we've talked about rumors.

00:59:02   Now I wanna talk a little bit about what Apple has shipped.

00:59:07   And most recently, it was the,

00:59:09   I hope you're sitting down for this,

00:59:11   multimodal large language models guided image editing,

00:59:15   or MG, MG, MGIE?

00:59:19   MG, why is it not editor?

00:59:22   I don't know, man.

00:59:24   (laughs)

00:59:25   I feel like it kind of falls off at the end there.

00:59:27   It does.

00:59:29   This was covered, this was just a couple of weeks ago.

00:59:33   This is a tool that you can use to edit photos.

00:59:38   And there were some examples like here's a pizza,

00:59:41   make it more healthy, and it added vegetable toppings,

00:59:43   or lighten and enhance the image,

00:59:47   make the screen green instead of white.

00:59:50   I mean, kind of the basic image editing things

00:59:52   that most people may want to do, but may not know how to--

00:59:55   This pizza looks so bad.

00:59:57   Do people edit their photos

01:00:00   to make their pizzas look healthier?

01:00:03   Is that a thing that people do?

01:00:04   It's like, hey, hey, you're sending a text to your wife

01:00:07   or something like, hey, look, I got myself

01:00:10   such a healthy pizza tonight with all these veggies on top,

01:00:14   where in fact you got the dirtiest pepperoni ever.

01:00:18   Is that a thing that people do?

01:00:20   Okay.

01:00:22   I mean, maybe, look.

01:00:23   I guess this is more, it's not the practical application,

01:00:27   it's the theoretical, right, that like--

01:00:29   The theoretical.

01:00:30   You can say make this more healthy,

01:00:32   and its response is to add greenery, which--

01:00:36   And Rippy pops up like, uh-uh, I see what you're doing.

01:00:39   Yeah, Rippy's back, you can't eat that pizza.

01:00:43   Eat this bowl of broccoli instead.

01:00:45   Rippy won't stand for pizza.

01:00:49   And so like, yeah, look, it is ridiculous.

01:00:52   Like, honestly, the results of all of these

01:00:54   are just like really weird to me,

01:00:55   but I guess this is what these kinds of papers are like,

01:00:59   but I guess it's showing that it understood the request

01:01:04   even if the result is strange, right?

01:01:07   Yeah, I think that's what they're after,

01:01:10   and this, I think, is an example of like,

01:01:13   people get this sort of thing, right?

01:01:15   It's kind of hard to say like, oh, series enhanced with AI,

01:01:18   like what does that actually mean?

01:01:19   But if you could tell your computer, your phone,

01:01:22   hey, I have this picture of us,

01:01:24   and can you take that weird person out of the background?

01:01:29   Or, you know, whatever it is, like.

01:01:30   And if you look at what Samsung just did

01:01:32   on a bunch of their phones,

01:01:34   it's very much in this, sort of in this vein.

01:01:37   So it makes sense that Apple would get this out there

01:01:41   and let people be poking around it.

01:01:44   Like, apparently, like if you have the right software,

01:01:46   you can go play with this now.

01:01:47   I looked at it for like three seconds

01:01:49   and then realized this is beyond

01:01:51   what I'm capable of doing today, and I kind of moved on.

01:01:55   But it's out there in the world, you know,

01:01:58   well before it's in a product, right?

01:01:59   This isn't in iOS today, but it very well could be.

01:02:03   - Man, I don't know.

01:02:05   And there's just something to me about like,

01:02:08   that is weird for Apple to be like,

01:02:12   we made a Magic Editor as well.

01:02:14   There's just something about that,

01:02:18   that just feels strange.

01:02:19   This stuff is going to be so hard to sell to the world,

01:02:24   especially to us at WWDC.

01:02:26   Like, to show me all these things that you're doing,

01:02:30   things that today I think are dubious at best

01:02:34   that we've had this technology in our pockets.

01:02:37   And then Apple's like, we put it in photos, potentially.

01:02:40   And it's like, ah.

01:02:41   - I think it's weird because it's in direct contrast

01:02:45   to the fact that Apple usually prides itself

01:02:48   upon the pure art of photography.

01:02:53   And, you know, and like you're basically here

01:02:57   taking photography, and I know that they have

01:03:00   machine learning-based enhancements,

01:03:02   but they don't drastically alter the contents of a photo,

01:03:06   almost as a sacred entity, if you will.

01:03:09   And here we're talking about these features of like,

01:03:10   hey, I'll change this pizza,

01:03:13   or take out this person from the shot,

01:03:16   or like that fundamentally change the truth of a photo.

01:03:21   And so I think that's why, at least for me,

01:03:23   this all sounds very un-Apple-y at the moment, but.

01:03:26   - But maybe there's no choice.

01:03:28   - Maybe there's no choice.

01:03:30   - No.

01:03:31   - Yeah, I can't imagine, I mean, like you said,

01:03:35   this guy gets hired from Google,

01:03:37   and now he's doing all this.

01:03:38   I'm sure that wasn't the plan,

01:03:39   but maybe there is expectation set in the marketplace, right?

01:03:44   The Pixel can do this. - Oh, there is.

01:03:46   - Samsung stuff can do this.

01:03:47   Maybe they've got to.

01:03:49   - Yeah, that's what I mean.

01:03:50   I don't think that there is a choice.

01:03:52   Like, they can find their way of doing it,

01:03:54   but I think some of these features

01:03:56   are gonna start to become table stakes.

01:03:58   Like Magic Eraser is such a good commercial.

01:04:02   - Yes, yeah.

01:04:04   - Like, it is a fantastic commercial,

01:04:06   and I think they're gonna struggle with that, right?

01:04:10   'Cause who doesn't have that one image in their library

01:04:13   where they really wish they could adjust it?

01:04:14   Like, it's just a thing people feel in their lives.

01:04:18   And I don't think it's in Google Photos today,

01:04:22   but like at a certain point, maybe Google do that.

01:04:24   I don't know, and then it's like makes it,

01:04:26   I don't know if it makes it better or worse,

01:04:28   but it's an awkward situation to be in

01:04:33   for if they take some kind of stance,

01:04:35   which is like, we're never gonna do this.

01:04:38   It's like, I think I might agree with what you're saying,

01:04:41   but from a business perspective,

01:04:43   that would not be a good idea, I don't think.

01:04:46   - Yeah.

01:04:47   So moving on from the multimodal large language models

01:04:52   guided image editing, a related tool is Apple Keyframer.

01:04:57   And so this is an AI animation tool

01:04:59   that lets you take a single image

01:05:01   and then use natural language

01:05:03   to tell the assistant how to animate it.

01:05:07   So the example given in the nine to five article is

01:05:09   you have a picture of a planet,

01:05:11   you say make the planet spin, it builds the frames

01:05:15   and the animation to make that possible.

01:05:17   So related, but different.

01:05:20   And I think making things move,

01:05:23   like it sounds so stupid as I'm gonna say it,

01:05:25   but I really think it's a thing.

01:05:27   That takes the what is a photo, what is real concept

01:05:31   and really makes it more complicated, right?

01:05:34   We saw that with Sora, the open AI video generator

01:05:37   that I'm sure every one of the planet

01:05:39   has seen all this creepy videos.

01:05:41   Like there's something about these tools

01:05:42   making something move that feels like a bigger deal.

01:05:46   And I kind of had that same response

01:05:48   to learning about Keyframer.

01:05:50   - But it could also be really cool

01:05:52   for app and game development.

01:05:54   - Yeah, absolutely.

01:05:56   Like what, yeah, if this were built into Reality Composer

01:05:59   or something like that, right?

01:06:00   It could be a big deal.

01:06:02   - Yes, with playgrounds, like all the sort of tools,

01:06:05   I think it could be really helpful.

01:06:06   - Yeah. - Yeah.

01:06:08   - So some developer tools that they've released,

01:06:11   there's three that I wanna touch on.

01:06:13   Apple MLX, AX Learn library for deep learning.

01:06:18   And then my favorite name, Apple Ferret.

01:06:22   - Ferret. - Like the animal.

01:06:25   - Nice.

01:06:26   - So the first couple MLX and AX Learn library,

01:06:30   these are machine learning AI frameworks

01:06:33   where developers can use the frameworks

01:06:36   to run efficiently on Apple Silicon.

01:06:38   And then the AX Learn library is about taking that

01:06:42   and making it really large scale.

01:06:43   So these are tools--

01:06:45   - Tell me about the Ferret, I wanna know about the Ferret.

01:06:47   - There's something about the Ferret.

01:06:47   - Yeah, well, no one cares about this

01:06:48   'cause you teach Ferret. - AX Learn,

01:06:50   no, give me the Ferret.

01:06:50   - Boo, Ferret, Ferret. - I'll give you the Ferret.

01:06:53   - Ferret, Ferret, Ferret.

01:06:55   - So Apple Ferret is a joint project

01:06:58   between Apple and Cornell University.

01:07:00   - Okay, okay, educated.

01:07:02   - Education and-- - Highbrow Ferret.

01:07:04   Ivy League Ferret.

01:07:06   - He's wearing a little tie, you know?

01:07:08   And it is research into how large language models

01:07:14   can recognize elements within a picture image.

01:07:18   So if you look at the link in the show notes,

01:07:20   it's like a picture of a cat.

01:07:23   And how do we know what the tail of the cat looks like?

01:07:26   - Obviously, the Ferret would know it's a cat.

01:07:28   - Yep, or Ferret.

01:07:30   How do we know where the head of the Ferret is

01:07:31   versus the foot of the Ferret, right?

01:07:34   So you can see how this could play in

01:07:37   with the other things we've spoken about, right?

01:07:39   If you're gonna have an image editor

01:07:41   or an animation editor.

01:07:42   Well, those things need to know what things are, right?

01:07:46   So if it's like, hey, take the tail off this cat,

01:07:50   maybe you don't like-- - No, don't do it.

01:07:52   - Well, you can't do it in real life,

01:07:53   but you could do it in AI.

01:07:55   - No. - No, you shouldn't.

01:07:57   - But it also needs to know for cutting things out of images

01:07:59   and like, okay, what is attached to the cat?

01:08:01   Do cats have tails?

01:08:03   Computers don't know any of this stuff.

01:08:04   Computers are stupid, and so we have to teach them.

01:08:07   And so Apple Ferret is looking at how do we build models

01:08:12   where images understand the components

01:08:16   that make up the subjects in those images.

01:08:19   Apple Ferret.

01:08:20   - I'm sure there are a lot of people

01:08:22   working really hard on this stuff, right?

01:08:23   It seems very, very complicated.

01:08:25   - Very nerdy.

01:08:26   - But whenever I see these papers, so both of these,

01:08:30   why does, like, why don't they just get someone

01:08:33   to help them design the graphics?

01:08:35   Like, what happens?

01:08:37   Like, so on the previous one,

01:08:39   we were talking about with the pizza,

01:08:40   like, I've seen this image a bunch of times now,

01:08:43   and like, why doesn't the text,

01:08:45   why don't the text labels actually line up with the images?

01:08:48   It looks like it was made in Word,

01:08:50   which I'm like 100% sure it was made in Word.

01:08:53   Like, you got the two images,

01:08:55   you got like the two text labels,

01:08:56   and like one of the text labels

01:08:58   is like only over 25% of the image,

01:09:00   and the other one is like flocked 100% to the image.

01:09:03   And then this Ferret one, like, what is going on?

01:09:06   What is that cartoon character in the corner?

01:09:09   Like, what is happening in these images?

01:09:12   - Yeah, I mean, come on,

01:09:14   SVP of AI, hire a graphic designer, you know?

01:09:17   Just one, just one graphic designer.

01:09:19   - No, it's big, graphic design is my passion and energy

01:09:22   in all these images, for sure.

01:09:24   - Or, you know, if you really don't wanna hire someone,

01:09:27   make a machine learning model to do it, you know?

01:09:31   Just like, make someone do it for you.

01:09:34   It's a very strange.

01:09:36   - It is.

01:09:37   - It's very strange.

01:09:38   - It really is.

01:09:40   The last thing I just wanna point out,

01:09:42   Apple has obviously also published research materials.

01:09:45   I picked this one 'cause of the name,

01:09:47   Specialized Language Models for Cheap Inference

01:09:50   on Limited Domain Data.

01:09:52   So it has to do with the other frameworks they've done,

01:09:55   basically looking at cost efficient approaches

01:09:59   to AI development.

01:10:01   And what I thought when I saw this was twofold.

01:10:03   One, this could be great for small developers

01:10:08   or people who are gonna kinda do this in their own app

01:10:10   or on their own servers somehow.

01:10:13   Or if Apple's gonna do a bunch of this,

01:10:16   they want it to be cheaper for them and their data centers

01:10:18   if they do it on the data center.

01:10:20   But cheaper and cost efficient also means smaller.

01:10:23   And maybe this also has to do,

01:10:25   look, I skimmed this paper.

01:10:28   I didn't understand every third word.

01:10:29   It is hard to understand what these people are doing

01:10:31   if you're just a regular person like we are.

01:10:34   But you can imagine this sort of research also goes into,

01:10:38   I wanna run this locally on a phone.

01:10:40   I wanna run this locally on a laptop.

01:10:42   And making this more efficient is cheaper in the data center

01:10:46   and makes it possible along with the other frameworks

01:10:49   that you cut me off about Apple Silicon

01:10:51   and making it as efficient as possible there.

01:10:53   Like these pieces are coming together, right?

01:10:56   That Apple is like putting all these pieces,

01:10:58   putting them together.

01:10:59   And I think products are gonna come out at the other end.

01:11:01   I think we're gonna start to see that this summer.

01:11:03   And that's fun.

01:11:04   And it's fun to see what they're doing ahead of time

01:11:07   'cause normally we don't get this, you know what I mean?

01:11:11   - These people are so smart.

01:11:13   Like I'm looking at that paper you referenced, Steven.

01:11:16   The people that do this stuff, they're just so clever.

01:11:20   Like in a whole different world.

01:11:22   - In a way they will never be this smart, you know?

01:11:25   - I'm happy with my talents and my skills in life.

01:11:29   Like I'm confident in those and I'm happy with where I am.

01:11:32   I will never be this smart.

01:11:34   And I don't think there is any course of my life

01:11:37   that could have led me to be able to write

01:11:40   specialized language models with cheap inference

01:11:42   from limited domain data.

01:11:44   I don't think I could have ever done this.

01:11:46   - It was never in the cards for any of us.

01:11:47   - It was not gonna happen for me, it just wasn't.

01:11:49   And you know what?

01:11:50   I'm fine with that.

01:11:51   I'm fine with that.

01:11:52   - The highest I can go is probably buy an actual ferret.

01:11:56   I think that's as far as I can put it.

01:11:58   - That's as far as your brain will allow.

01:11:59   - As far as I, look, I have a ferret.

01:12:01   I'd like an actual ferret.

01:12:02   - Ferret ownership.

01:12:03   - Yeah.

01:12:05   Steven, thank you for this research.

01:12:08   All the links, thank you.

01:12:11   I think hopefully we have a better understanding

01:12:13   of at least some of the things that are expecting us in June.

01:12:18   I'm sure more reports will come out

01:12:19   over the next couple of months, but this is great.

01:12:22   Thank you.

01:12:23   - Yeah, it was really fascinating to see what they're doing.

01:12:26   And like I said a second ago,

01:12:28   software and hardware, we get rumors, right?

01:12:31   But this is actual work that Apple is doing in the field.

01:12:35   And that's just so unlike what we normally get to talk about.

01:12:39   So it was a lot of fun to dive into.

01:12:42   - I think what this shows is that

01:12:44   if you wanna do this stuff, you will only get good hires

01:12:48   if you allow them to publish.

01:12:50   - Yes.

01:12:51   - And it is so on Apple,

01:12:53   but I think is a good sign that they're serious, right?

01:12:58   Like for example, we spoke about it before,

01:13:02   like if you write, people that write stuff on the app store,

01:13:05   they don't put their names on it.

01:13:06   Like they're not allowed to put their names on it.

01:13:08   Like you can't sign like, oh, I was this person

01:13:11   and I wrote this story on the app store, right?

01:13:13   Like that's not a thing that the writers

01:13:16   and the app store teams get to do.

01:13:19   And that maybe limits the type of people

01:13:21   that they would be able to employ,

01:13:22   people that want to put their name on their work.

01:13:25   But like this is an example of that,

01:13:27   where if you are a researcher, you want to be published

01:13:31   because that's what it's all about.

01:13:34   And they are allowing these people to do that

01:13:37   because I guess, because they are aware of the fact

01:13:39   that they're only gonna get the brightest minds

01:13:42   if they allow them to continue publishing.

01:13:45   And I expect that there's some stuff

01:13:47   that they can't publish yet,

01:13:48   but we'll be able to publish later

01:13:49   once the products have been announced as well.

01:13:52   And that you kind of get it that way.

01:13:54   - Yeah, I think there'll be a whole flood

01:13:57   of these white papers and research documents.

01:14:00   Again, not that we know what they say,

01:14:01   like seriously, go look at the PDF that's in the show notes.

01:14:05   It's next level.

01:14:06   It's very easy to get through the first like three sentences

01:14:11   and then it falls apart real fast, real fast.

01:14:14   - Yeah, and then you get into like some of the later pages,

01:14:16   there's like mathematical equations.

01:14:19   - There's a lot of charts and there's a lot of lines.

01:14:21   And I guess I should know what the generic pre-training cost

01:14:25   versus generic complexity graph is,

01:14:28   but I actually don't know what that means.

01:14:30   - Well, have you considered the conclusion

01:14:32   of the hyper network research on small documents?

01:14:36   - You just opened it up for me, now I understand.

01:14:38   Thank you.

01:14:39   - Good. - Thank you.

01:14:40   - Good.

01:14:41   I think that does it for this week.

01:14:44   Thanks for hanging out with us.

01:14:46   We love doing this show and we love doing it for you.

01:14:48   - Hope you found it pleasurable.

01:14:49   - We did, the most pleasurable podcast.

01:14:52   If you want more of us, we're online.

01:14:55   Federico is the editor-in-chief of maxstories.net.

01:14:58   Continues to be the home of amazing Vision OS app coverage.

01:15:01   Like I feel like every time I open max stories,

01:15:03   you're costing me money.

01:15:05   I've downloaded so many new apps.

01:15:07   - Great.

01:15:07   - If only there was an app affiliate program

01:15:09   you could take advantage of.

01:15:10   - I know, right?

01:15:11   (laughing)

01:15:14   - Too soon?

01:15:14   Federico is on threads as Vatici

01:15:17   and he's on Mastodon as vatici@maxstories.net.

01:15:22   You can find Mike's work on a bunch of other shows

01:15:25   here on Relay FM.

01:15:26   Episode 500 of Upgrade, congratulations.

01:15:29   - Thank you.

01:15:30   - You started after us, but you passed us in numbers.

01:15:32   - Yep, it's just a very strange thing to have occurred.

01:15:35   - It is very, very strange.

01:15:38   A lovely episode, y'all talked about the report card.

01:15:41   You do this cool thing where you do,

01:15:43   basically it's like a lame version of the Ricky's

01:15:46   for every hundred episodes.

01:15:48   - I wouldn't call it that, it's a draft,

01:15:50   which I think drafts predate the Ricky's.

01:15:53   - That doesn't sound right to me.

01:15:55   - Right, okay, okay.

01:15:57   - I'll come back with a white paper.

01:15:59   You can follow Mike on threads as imike

01:16:01   and he is on Mastodon as imike@mike.social.

01:16:05   You can find my writing on 512pixels.net

01:16:08   and I co-host Mac Power Users.

01:16:10   It comes out each and every Sunday here on Relay FM.

01:16:13   I am ismh@eworld.social on Mastodon

01:16:16   and ismh86 over on threads.

01:16:20   I'd like to thank our sponsors this week,

01:16:22   NetSuite, Nom Nom, and Squarespace.

01:16:24   And until next time, say goodbye.

01:16:27   - I'll leave it there, cheerio.

01:16:29   - Bye, y'all.