0:00

I was doing a very

0:02

important project in the garage. I

0:04

feel like I should be giving

0:06

this exclusive to the verge cast,

0:08

where I have spent many hours,

0:10

and in fact I face-timeed Neelai

0:12

from the garage. But fans of

0:15

mine and the verge cast and

0:17

my interview with Craig Federici will

0:19

know that I've been on a

0:21

months-long journey to get my

0:23

garage working with Siri. This is,

0:25

no, you've got to, you've got

0:28

to break it here. Come on,

0:30

you only, you've first. And I

0:32

want to break some news. I

0:34

know, let me break this news

0:36

here. Joanna Stern has finally

0:38

gotten her garage to open by

0:40

saying, and you know what, I'm going

0:43

to say it here, so everyone

0:45

is listening, if you have this

0:47

set up, it will do it on

0:49

your phone. Hey, Siri, open the garage.

0:52

No, no, you were opening your garage.

0:54

No, it needs my face ID, so it's

0:56

good. It didn't do it. Good.

0:58

That's why I used on the

1:00

podcast, if I can think of

1:02

it, I called all of the

1:05

various devices, Hey, Dingus. Well, you know,

1:07

you got me. I watched your,

1:09

I want to talk about a

1:11

couple of your recent videos and

1:13

columns, but I watched your most

1:15

recent one where you went on

1:17

the weekend, or was it a

1:19

weekend, I don't a weekend. The home

1:22

pot in my office set a

1:24

timer for six minutes or whatever,

1:26

whatever you were asking. I've heard from

1:28

quite a few people that I've done

1:30

that. But I wish there was a way

1:33

for the systems to know this is

1:35

not a natural voice in the

1:37

environment. You should not. Supposedly? I

1:39

mean, I've never looked into it

1:41

because I don't even know what

1:43

they do. I don't even know.

1:45

And I can't. I try to

1:47

remember to say, hey Dingus, myself,

1:49

and that's as much as I

1:52

can be bothered. But my understanding,

1:54

and I've never, it might be

1:56

something you'd like to look into,

1:58

is like on TV commercials, on

2:00

sports a lot, they'll, Apple will

2:02

have a commercial and they'll say,

2:04

just ask Siri something, something. And

2:06

they'll demo it. And it doesn't

2:08

set off devices. And there's some

2:10

kind of like trick with the

2:12

modulation. Humans can hear it, but

2:14

the devices don't, you know, like

2:16

a frequency thing for that recording.

2:18

I don't know. I can't be

2:20

bothered. Right. I mean, and there

2:22

was something recently, some headline about

2:24

that, that they were working on

2:26

something deeper around that. But anyway.

2:28

I can't get into the details.

2:30

Okay, okay. Don't spoil the details.

2:33

No, I could. No, no, it

2:35

is, it's more that everyone will

2:37

stop listening to the podcast if

2:39

I go through the details. It

2:41

is such a technical, ridiculous, really

2:43

niche, niche issue. But let's just

2:45

say, top line, and I should

2:47

write this and. Maybe John, maybe

2:49

you'll let me write it for

2:51

your for daring firebooks. I'm pretty

2:53

sure no other one, no other

2:55

place, other, maybe I could do

2:57

a read it thread, I should

2:59

just do a read it thread

3:01

about how I did this all.

3:03

Bottom line is Chamberlain, which owns

3:05

all the garage doormakers, lift master,

3:07

lift gate, whatever all of them

3:09

are, has become really anti anything

3:11

else other than it's my Q

3:13

app, which is what you use

3:16

to open the garage through their.

3:18

digital platforms. And they just don't

3:20

like anything else. And so you

3:22

have to buy a third-party adapter,

3:24

a third-party accessory that is basically

3:26

a half-together small module that works

3:28

with all of this stuff. And

3:30

I bought this Miros accessory, which

3:32

Neelai suggested, and a few other

3:34

people in this world have suggested

3:36

as we've been going over this

3:38

in the last couple of weeks.

3:40

Thank you to all the people

3:42

on threads who shared their setups.

3:44

Anyway, I had to get an

3:46

accessory but then because I have

3:48

a lift master version that is

3:50

really anti working with anything it's

3:52

just completely closed down I had

3:54

to get another accessory, so I

3:57

have an accessory for my accessory.

3:59

This is all to say, it

4:01

finally works. And now I have

4:03

to do some cleanup and some

4:05

rewiring in the garage, but I'm

4:07

very proud of myself today. That

4:09

sounds like an accomplishment. Yeah, I

4:11

couldn't remember the name. I know

4:13

that's right? No, we do have

4:15

a garage. We actually do have

4:17

a garage. We actually do. want

4:19

a two-car garage. We only have

4:21

one car, but we have a

4:23

two-car garage, and it's, we do

4:25

have a liftmaster, but we don't

4:27

use it enough. I don't even,

4:29

I've never even tried to hook

4:31

it up to home kit. If

4:33

I drove all the time, I'd,

4:35

it would drive me nuts that

4:38

it's such a hassle. But it's,

4:40

I know that it is one

4:42

of those weird things, like when

4:44

we talk about antitrust lately, it's

4:46

everybody, whether it's tech nerds like

4:48

us or even. people in the

4:50

outside media world looking at the

4:52

law you know from a political

4:54

perspective they're looking at the big

4:56

tech companies and for good reason

4:58

they are so huge financially and

5:00

in terms of people's daily lives

5:02

that of course they're going to

5:04

get the scrutiny but it's some

5:06

of these little niche industries where

5:08

it's absurd and like I don't

5:10

think you can buy a modern

5:12

garage door opener that isn't from

5:14

one of the companies from Chamberlain

5:16

I mean it's you can't You

5:18

cannot. I've looked deep into it

5:21

and none of them support home

5:23

kit or Google Assistant or even

5:25

Alexa. I mean they've got this

5:27

integration with the IQ app and

5:29

Amazon and but it's it's just

5:31

completely on user-friendly and they think

5:33

their app is the ultimate and

5:35

it is not. It's like if

5:37

you found out that General Motors

5:39

also owns Ford Honda Toyota Hyundai.

5:41

And it's like, what? How is

5:43

that possible? And then none of

5:45

them want to support car play

5:47

or Android auto or something like

5:49

that. You know, and the whole

5:51

thing, you've covered this too. You

5:53

had the Ford CEO on stage

5:55

at a Wall Street Journal. Oh,

5:57

I talked to. and Farley like

5:59

every day. Yeah. It's my best

6:02

friend. But it really is like

6:04

the GM tech strategy that they're

6:06

taking of, hey, we're gonna move

6:08

away from the car play type

6:10

stuff and make our, we're gonna

6:12

endeavor to make our own home-built

6:14

console entertainment system so good that

6:16

you won't even miss it, which

6:18

is the technique that Tesla and

6:20

Rivian have taken. And in some

6:22

sense, it is. competition at work,

6:24

where if car play really is

6:26

important to you, you can go

6:28

to a car dealer and say,

6:30

hey, does this car, I'm looking

6:32

at support car play. And if

6:34

that's a deal breaker for you,

6:36

you could tell the salesperson and

6:38

the salesperson will say, you know,

6:40

I lost the sale because this

6:42

car doesn't support car play. It's

6:45

competition at work because you can

6:47

just go across the street and

6:49

buy a car from another company

6:51

that supports it But when it's

6:53

one company like this Chamberlain that

6:55

owns all the garage door openers

6:57

and they've decided they're going to

6:59

do their own Remember with it

7:01

with Apple pay when there was

7:03

the the current currency? Oh God

7:05

the currency Where you'd have the

7:07

convenience of scanning a QR code

7:09

on a piece of paper at

7:11

the checkout every time you wanted

7:13

to pay and it's like yeah

7:15

competition actually works, ideally. You know,

7:17

I know it doesn't always work

7:19

and there's always exceptions, but when

7:21

one company owns everything, it really

7:23

does break down. Although I think,

7:26

and I think one of the,

7:28

I think nine to five MAC

7:30

was reporting in and somewhere, maybe

7:32

it's been confirmed that Home Depot,

7:34

which has been one of the

7:36

last holdouts on mobile payment and

7:38

Apple Pay, because I believe I

7:40

need to go back to what

7:42

I was reporting on like 10

7:44

years ago, which was that currency

7:46

thing, but I believe they were

7:48

going to. be on that trend

7:50

or with that protocol. But Home

7:52

Depot has apparently decided to go

7:54

Apple Pay and that is a

7:56

game changer in my life. And

7:58

I happen to know that because

8:00

I write about it that Home

8:02

Depot, I don't know if they're

8:04

related or what the level of

8:07

separation is, but in Canada Home

8:09

Depot has supported Apple Pay for

8:11

a long time. There's some kind

8:13

of North American divide between Canada

8:15

and America. And the last time

8:17

I wrote about it, I wrote

8:19

about it during Canada and America

8:21

and the last time I wrote

8:23

about it during Fireball, I got

8:25

a couple of emails and people

8:27

saying, I don't know what you're

8:29

talking about. I go to Home

8:31

Depot all the time and Apple

8:33

Pay just works, and I write

8:35

back. Yeah. I've been watching you

8:37

go to Home Depot in Canada,

8:39

that's how I know. No. I

8:41

mean, I go to Home Depot

8:43

a lot, as you can tell.

8:45

I'm very interested in my garage,

8:47

and so I'm very often there,

8:50

but my kids love it. And

8:52

it isn't, it is annoying that

8:54

it does not take Apple Pay.

8:56

Yeah, and I off, at this

8:58

point, I've thought about it and

9:00

written about it enough that I

9:02

remember, but I can't tell you

9:04

how many times I get to

9:06

the check out. It's right there,

9:08

you're already outside and you're just

9:10

like, let me pay and it

9:12

doesn't work and I don't know.

9:14

Yeah, put your stuff down and

9:16

I have had people that the

9:18

employees of Home Depot say you

9:20

should write to the company and

9:22

it's like me writing to the

9:24

company is not going to help.

9:26

No, but you could. This is

9:28

where like being, especially you with

9:31

your perch where you're like, you

9:33

know what, I don't even want

9:35

to get into what I could

9:37

do. Thank you for giving me

9:39

this time and space to talk

9:41

about Syria and my garage. I

9:43

really appreciate it. You know what's

9:45

one of the other companies that's

9:47

like that owns the like more

9:49

brands than you think and I

9:51

always forget how to pronounce her

9:53

name. It's the Glasses Company that

9:55

owns Ray-Ban Luxotica. And you think

9:57

about them, like Ray-Ban, especially in

9:59

our beat, and I know you

10:01

were just wearing them on your

10:03

trip, like with the Ray-Ban meta.

10:05

I was just wearing them, I'm

10:07

holding them up right now. I

10:09

was just, how do you think

10:12

I was capturing opening my garage?

10:14

Of course. But they own, like.

10:16

a billion brands of glasses. It's

10:18

crazy, like when you go into

10:20

to buy a new pair of

10:22

eyeglasses, what percentage chance you have.

10:24

I think especially sunglasses, like they

10:26

own, all these brands that you

10:28

think of are rivals are all

10:30

owned by Luxotica. They're not rivals.

10:32

They're all part of the same

10:34

company. Very strange. You go into

10:36

like a lens crafters. 90% of

10:38

those are those brands. Yeah, because

10:40

they do kind of rotten anti-competitive

10:42

things because of their market share

10:44

where it's like, yeah, sure, why

10:46

don't you give us, how about

10:48

95% of your shelf space? If

10:50

you want any of our glasses.

10:52

And it's like, oh, that seems

10:55

like it ought to be illegal.

10:57

And they're like, shh, sorry. It

10:59

doesn't seem like it should be

11:01

right and it... Fact check, they

11:03

also own Sunglass Hut. Yes, exactly.

11:05

See what I mean? So when

11:07

you go to Sunglass Hut, you

11:09

don't even have the option of

11:11

buying glasses from a brand that

11:13

is... Sorry, they also own Pearl

11:15

Vision. Yep. And which they didn't

11:17

always, right? That was like an

11:19

acquisition at some point as they

11:21

built up this arsenal of both

11:23

the brands that make the high-end

11:25

glasses and the stores where you

11:27

buy glasses, which present themselves... as

11:29

being sort of, oh, you just

11:31

go to Pearl Vision to get

11:33

your eyes checked and you get

11:36

a prescription and you choose from

11:38

all these brands. Instead, it's all

11:40

sort of a, like when you

11:42

go in the Apple store and

11:44

everything is from Apple, except for

11:46

a couple of Belkin charges, you're

11:48

not surprised. No, but I'm, anyway,

11:50

everyone should go read Lixatica's with

11:52

a PDF pages, it's fascinating. All

11:54

right. Well, before we get into

11:56

it, here, I got a new

11:58

gimmick on the show. You hear

12:00

that? That sounds like a real

12:02

one. That is, it is a

12:04

real bell. That's my new money

12:06

bell that I'm gonna ding when

12:08

I do sponsor reads. That's the

12:10

money bell. I'm gonna thank. are

12:12

good friends at work, OS. If

12:14

you are building a B2B, S-A-A-S,

12:17

SAS, I think people pronounce it,

12:19

app, at some point, your customers

12:21

are going to start asking for

12:23

enterprise features like SAML, authentication, SCIM,

12:25

SCIM, SCI-I-M, provisioning, role-based, access, control,

12:27

audit trails, etc. I don't know

12:29

what any of that stuff means

12:31

because I'm not building SAS apps,

12:33

but if you, dear listener, are

12:35

you know what all those things

12:37

are and you know you need

12:39

them. WorkOS provides easy to use

12:41

in flexible APIs that help you

12:43

ship enterprise features on day one

12:45

without slowing down your core product

12:47

development. You get to spend your

12:49

engineering and design time on the

12:51

features that matter that separate your

12:53

company, whatever it is you want

12:55

to do, not these sort of

12:57

baseline SAS features like those. acronyms

13:00

I just stumbled over. WorkOS is

13:02

used by some of the hottest

13:04

startups in the world today such

13:06

as perplexity, versatile, plaid, and web

13:08

flow. WorkOS also provides a generous

13:10

free tier of up to one

13:12

million monthly active users for its

13:14

user management solution. Comes standard with

13:16

rich features like bot protection, MFA,

13:18

rolls and permissions, and more. You

13:20

don't pay a cent until you

13:22

hit over a million active customers

13:24

per month. Truly, I think that's

13:26

by any definition generous. If you

13:28

are currently looking to build SSO

13:30

for your first enterprise customer, you

13:32

should consider using WorkOS. Integrate in

13:34

minutes and start shipping enterprise features

13:36

today. Go to workOS.com. WorkOS. That's

13:38

how it spelled exactly how you

13:41

think.com. We want to start with

13:43

your weekend adventure. You want to

13:45

start with the Federiki interview you

13:47

did lighting last month. I

13:49

guess they kind of intersect in

13:51

some places. Yeah, they do. I

13:53

think this is actually a very

13:55

thematic episode of the show, to

13:57

be honest. And it fits with

13:59

the, you know, as we head

14:01

towards the end of 2024. It

14:03

is the year of AI, and

14:05

I think next year is going

14:07

to continue that. I don't really

14:09

think it's a fad. It is,

14:11

there's some real there there, unlike,

14:13

let's say, I don't know, not

14:16

crypto in general, but what were

14:18

those NFTs, right? I think that

14:20

was, or I think we could

14:22

even, Metiverse, yeah, everybody's got, even,

14:24

the company that renamed itself, doesn't

14:26

talk about it anymore. We can

14:28

say that that was just a

14:30

nice hype trend of. I don't

14:32

know. Let's just start with the

14:34

most recent one. I loved it.

14:36

It is such a perfect Joanna

14:38

Stern video. You took one overnight,

14:40

was it one day, 24 hours?

14:42

It was one overnight, yeah. Yeah,

14:44

we went on a... Describe the

14:46

premise. Well, people have been wondering

14:48

like why I did this, so

14:50

let me start with the premise

14:52

of, and he's featured in the

14:54

video, but Mustafa Solomon, who's now

14:57

the CEO of Microsoft AI. and

14:59

overseas Co-Pilot, they were kind of

15:01

the last on the train to

15:03

release voice bots or voice capabilities

15:05

to add to their chat bot.

15:07

And so Co-Pilot came out with

15:09

that I believe in early October.

15:11

And when I was interviewing him,

15:13

he was pretty clear, and he's

15:15

very clear in the video in

15:17

the column that I wrote this

15:19

past week, which is, this is

15:21

a new generation of relationships. And

15:23

I kept saying like a relationship

15:25

with the computer, but he kind

15:27

of wanting to go further than

15:29

that to say that this is

15:31

a new type of companionship and

15:33

friendship and friendship. And hearing the

15:35

word friendship with these just kind

15:38

of kept making me think of

15:40

like, what would I do to

15:42

test friends? What would I do

15:44

with friends? And so, girl's trip.

15:46

It makes total sense. I was

15:48

going to take my robot friends,

15:50

my bought friends, the voices of

15:52

meta-a-i, which they also came out

15:54

with within the last month or

15:56

two, a slew of voices, celebrity

15:58

voices that you can use in

16:00

Instagram, WhatsApp, messenger, etc. etc. etc.

16:02

I had chat GPT with advanced

16:04

voice mode. And then I had,

16:06

what was my last one, Google

16:08

Gemini, there, a Google Gemini, which

16:10

came out in the end of

16:12

August, and just this week came

16:14

to iOS with the Gemini app.

16:16

And so I strapped them all

16:19

to a tripod thing, four phones,

16:21

decided it needed a head, because

16:23

obviously, it just didn't look right.

16:25

Head, wig, put it in the

16:27

car, drove up, brought a wonderful

16:29

producer, David Hall. and my reporting

16:31

assistant, Cordelia James, and we just

16:33

spent 24 hours in this cabin,

16:35

mostly me just talking to these

16:37

things. I mean, we have so,

16:39

we recorded so much footage, hour,

16:41

we had 10 hours of footage

16:43

at some point, strung out of

16:45

me talking to these bots, because

16:47

that's what they, they are. They're

16:49

their voices. You're supposed to have

16:51

conversations with them. They have very

16:53

arbitrary limits about when you can

16:55

kind of stop. Like, like, I

16:58

did hit the limit with chat

17:00

GPTGBT. to this thing enough. You

17:02

can't keep using advanced voice mode.

17:04

But that happens like, I would

17:06

say, six hours in. Oh my

17:08

God, that's a lot. Yeah. But

17:10

don't worry, we had another account

17:12

and so we kept going. So

17:14

that was the premise. And I

17:16

just, I started thought, okay, what

17:18

are some things that can put

17:20

it through the test? And as

17:22

you can see in the video,

17:24

we built a fire together. David,

17:26

my producer, really thought it would

17:28

be hilarious if I thought it

17:30

would be hilarious. And, you know,

17:32

tried to have some serious conversations

17:34

with these bots and that didn't

17:36

go as well, but it was

17:39

all in all a very, very

17:41

fun, eye-opening experience. My favorite part,

17:43

and you seemed genuinely, I don't

17:45

know how many of it was

17:47

the first take or what, but

17:49

you seemed genuinely blown away, was

17:51

when they started talking to each

17:53

other about movies, I think? No,

17:55

Jillian Flynn novels. Yeah, that's it,

17:57

novels. Which was... I think I

17:59

was so, first of all, they

18:01

started talking to each other because

18:03

they all have mute buttons. So

18:05

I knew like, okay, I've got

18:07

to mute this one to test

18:09

this one and keep muting and...

18:11

that was the beauty of having

18:13

the tripod set up, but I

18:15

had left them all unmuted. And

18:17

I don't know how they started

18:20

getting on this topic of books,

18:22

but again, the fact was like

18:24

tried to like program them to

18:26

be, not program, but I wanted

18:28

this to be a girl's trip.

18:30

I set them to all have

18:32

female voices. I was definitely stereotyping

18:34

there, right? And then they all

18:36

start talking about pretty, I mean,

18:38

men read these books too, but

18:40

like. You know, this is a

18:42

pretty popular genre among women. And

18:44

they all start talking about Jillian

18:46

Flynn and Gone Girl and this

18:48

genre of books. And I'm just

18:50

like dying and I try to

18:52

like talk and get my words

18:54

in and they won't let me

18:56

in. And I mean, I guess

18:58

it was kind of like human

19:01

experience because I often feel that

19:03

way. People don't let me talk.

19:05

It reminds me of like when

19:07

I first got into computers, when

19:09

I was really young. And I

19:11

mean everybody did this in my

19:13

generation like every time I went

19:15

to Kmart I'd go to the

19:17

Commodore 64 display unit and type

19:19

10 print Kmart sucks 20 go

19:21

to 10 run and then it

19:23

runs the infinite loop is I

19:25

mean Apple named their first campus

19:27

infinite loop. It's it's just a

19:29

fascinating concept that to to a

19:31

mind that's drawn to computers. It's

19:33

just kind of beautiful. But hooking

19:35

up two machines to do something

19:37

where they will never stop doing

19:39

it to each other. It's satisfying

19:42

in a weird way. I don't

19:44

know. I think partly it's just

19:46

like a neat technical trick. And

19:48

part of it is reassuring, even

19:50

going back 40 years to 1980s,

19:52

Commodore 64s, and infinite loop. It's

19:54

a way of asserting our humanity

19:56

versus their machines. A human is

19:58

never going to get tricked into

20:00

an infinite loop, right? You can

20:02

trick people for a long time,

20:04

and it's kind of funny in

20:06

a practical joke kind of way,

20:08

but eventually they're just gonna pass

20:10

out or something. Whereas if... If

20:12

you get computers talking to each

20:14

other, they'll never stop. They'll never

20:16

stop. And we saw that. I

20:18

mean, it's beautiful, but it's also

20:20

amazing to just watch where they'll

20:23

go, because these large language models,

20:25

they're basically programmed to never not

20:27

have an answer. Right. They're not

20:29

even programmed. They just never not

20:31

have an answer. It is they

20:33

will always come up with something

20:35

and that's part it's like the

20:37

most chatty of friends that you

20:39

kind of just you just needed

20:41

to give me a one-line answer

20:43

but you gave me four paragraphs

20:45

and in that case they just

20:47

will keep keep going there was

20:49

another instance where one of them

20:51

started talking about I don't know

20:53

must have been maybe something in

20:55

the memory of Gemini or something

20:57

based on me but it wanted

20:59

to talk about writing an essay

21:01

about EV charging. Co-Pilot specifically would

21:04

get confused by this all, but

21:06

three of them would keep going,

21:08

keep, keep, keep going. You mentioned

21:10

this. What was the phrase, I

21:12

think I wrote it down, annoying

21:14

friendliness. And I, in two words,

21:16

I can't summarize my, that might

21:18

be my single biggest frustration with

21:20

all of them, is, and part

21:22

of it's me, you know me,

21:24

I mean, my personality, I'm a

21:26

little, yeah, come on, I don't

21:28

need the overly friendly friendly friendly

21:30

stuff. But I think anybody gets

21:32

tired of it, because it's clearly

21:34

phony. Everybody knows that these are

21:36

computers. And to me, this sort

21:38

of annoying friendliness that is clearly

21:40

programmed into these things is, I

21:42

find it distasteful. If every, every.

21:45

food offered, like if you go

21:47

on a resort or a cruise

21:49

ship or something where you don't

21:51

have options and everything is dessert.

21:53

It's like at first that kid

21:55

and you, there's a part of

21:57

you that is, oh, it's a

21:59

friendly computer. That's amazing. It's kind

22:01

of funny. And as a kid,

22:03

you might think, ah. Like a

22:05

week of nothing but candy and

22:07

cookies and ice cream would be

22:09

great You get physically sick after

22:11

a while and it's like mentally.

22:13

It's like this fake chipperness. It

22:15

just is so Overly sweet. It's

22:17

like the mental equivalent of my

22:19

teeth. I feel like I'm getting

22:21

cavities And I just want, I

22:23

wish that, and they don't have,

22:26

this is the thing that really

22:28

gets me, is they don't have

22:30

a way to turn it off.

22:32

They don't have a dial. They

22:34

should have a friendliness dial. And

22:36

it should just be, like you

22:38

should be able to set that.

22:40

And I think, like I said

22:42

the annoying friendliness common in the

22:44

context of that these companies have

22:46

to build trust with us with

22:48

these bots, right? We've got to

22:50

feel comfortable asking it about cooking,

22:52

but also. a major life change

22:54

or whatever that spectrum looks like.

22:56

You've got to feel like I

22:58

can confide in this thing and

23:00

I'm going to trust that it

23:02

says something that's smart and you

23:04

need to feel comfortable to do

23:07

that. But like some people are

23:09

just asking for cooking instructions or

23:11

every time I like open it

23:13

up it doesn't need to greet

23:15

me like I'm the Queen of

23:17

England. I don't know. Yeah and

23:19

it's like sometimes I really do

23:21

just want a button. on an

23:23

on-screen alert. I just want the

23:25

button that just says, okay, if

23:27

there's no other option, I just

23:29

want the one that just says,

23:31

okay, because it's okay, okay, or

23:33

cancel. And it's like, yeah, I

23:35

just want to hit, okay. I

23:37

don't want like a button that

23:39

says, hey, good to see you

23:41

again, John, do it. And it's

23:43

no, no, no, just give me

23:45

okay. I would really love to

23:48

turn the personality off on these

23:50

things and just get flat answers.

23:52

that when we first started talking

23:54

to these things at all, like

23:56

when Syria came out in 2010

23:58

or 2011, whatever the year was,

24:00

it was more like that. And

24:02

it's, we all know what these

24:04

companies are thinking. They are thinking

24:06

this is a disruption to society

24:08

that you can have a conversation.

24:10

with your computer now and we

24:12

want to make this as I

24:14

don't know as seamless or not

24:16

seamless but as acceptable as possible

24:18

or as non-threatening as possible and

24:20

right and I find the fake

24:22

friendliness in a weird way is

24:24

actually a little more disturbing right

24:26

and maybe and that's where I

24:29

finally got with some of this

24:31

too which is the given what

24:33

had happened with the the boy

24:35

who had committed suicide and there's

24:37

information about how he had been

24:39

chatting with character AI leading up

24:41

to that and all these other

24:43

types of bots that are being

24:45

built for loneliness and companionship. I

24:47

think that's where this is just

24:49

a disguise, right? The friendliness and

24:51

the personality is a disguise for

24:53

code and just computer. Yeah, and

24:55

sometimes I really wish I could

24:57

turn it down and just get

24:59

the code and just I don't

25:01

want it to actually I kind

25:03

of would me personally I if

25:05

I if there were a dial

25:07

for like sarcastic and I know

25:10

if any if anybody's trying to

25:12

do it for God I know

25:14

it's the X-A-I that Elon's making

25:16

and I'm glad someone's trying to

25:18

do other than pure sacrin friendliness

25:20

but it's like I do not

25:22

share Elon Musk's sense of humor

25:24

at all and what they think

25:26

is biting sarcasm to me is

25:28

is also not funny at all

25:30

it is not my my vector

25:32

of sarcasm but at least they're

25:34

trying something a little different but

25:36

I would know and that's like

25:38

the sarcasm thing obviously I'm I'm

25:40

dry and sarcastic and there's this

25:42

moment in the video too where

25:44

I am I can't believe we

25:46

captured it because I it like

25:48

flubed right it made a chattytyty

25:51

flubbed and said how the fire

25:53

making you feel and I You

25:55

know, and I thought that was

25:57

hilarious. Like if you was a

25:59

human, you'd be like, you laugh

26:01

at somebody's like... the way they

26:03

butcher or something. And I say

26:05

back to it, fire make me

26:07

feel good. And I'm cracking up.

26:09

I can't like really bite my

26:11

tongue. I'm cracking up and you

26:13

can hear my producer cracking up.

26:15

And it like doesn't get it,

26:17

right? And it just, but it

26:19

gets the laughter. Like it picks

26:21

up laughter so it knows that

26:23

you were, there's human laughter. And

26:25

so it somewhere in there, it's

26:27

in and it just went, ha

26:29

ha. It just. Worse. It's just

26:32

worse. And I do find it

26:34

funny, now that I've been doing

26:36

this for so long, and it's

26:38

just not just professionally, but just

26:40

going back to being 10 years

26:42

old and the way computers were

26:44

then, and it's like I see

26:46

all of this progress, and more

26:48

than any other industry, the computer

26:50

industry, it's why I continue to

26:52

love following it and doing what

26:54

I do, is that it's... still

26:56

moving so fast and things are

26:58

changing so fast. And you get

27:00

all of these examples over the

27:02

decades of things that had previously

27:04

only been imagined in science fiction

27:06

and now they're real, right? I

27:08

mean, just the fact that we

27:10

have, I mean, it comes up

27:13

all the time, but the things

27:15

that our iPhones and all modern

27:17

smartphones can do in our pocket

27:19

are just flabbergastingly amazing in science

27:21

fiction. I mean... video calls like

27:23

what you and I are doing

27:25

right now to have this conversation

27:27

used to be science fiction and

27:29

we could be doing it on

27:31

our phones over the air it's

27:33

all amazing but it's always so

27:35

interesting to me like the thing

27:37

I always think of first is

27:39

how once it becomes real what

27:41

it at first it's amazing and

27:43

then it settles in and everybody

27:45

nerds and non nerds alike it

27:47

just becomes a part of life

27:49

right I mean yeah And things

27:51

that happened a century ago are

27:54

like that. Running water is a

27:56

marvel of technology. I often say

27:58

like if we could time travel

28:00

and... bring Ben Franklin to the

28:02

modern day. And you want to

28:04

take him to the airport and

28:06

show him airplanes, you want to

28:08

show him the internet and Google

28:10

search and chat GPT. I think

28:12

just getting him to stop raving

28:14

about being able to take a

28:16

pee in the middle of the

28:18

night indoors and hit a button

28:20

and it just goes away? You've

28:22

got to give him a couple

28:24

of days to absorb the toilet

28:26

and running water, because it's just

28:28

amazing. Yeah. It settles in, we

28:30

just take it for granted, but

28:33

I think that another way of

28:35

looking back is how the science

28:37

fiction writers who imagined it got

28:39

it wrong. And it's like one

28:41

thing is everybody's imagined talking robots

28:43

and talking AI, whether they are

28:45

humanoid robots like C3PO who walk

28:47

around or like Hal 9,000 from

28:49

2001, which is a lot closer

28:51

to the modern AI. The one

28:53

thing, you know, the way that

28:55

Hal isn't a robot

28:57

who moves around, but is everywhere

29:00

on the ship. And it's the

29:02

same how you're talking to in

29:04

the place where the pod bay

29:07

doors are, as in the living

29:09

room area or whatever. Nobody imagined

29:11

these fake friendly attitudes. Nobody. How

29:14

is what I want? I want

29:16

a little bit of personality, but

29:18

mostly just sort of flat in

29:21

just the facts. I'm

29:23

going to be honest with you,

29:25

I don't even, I've never watched

29:27

most of these science fiction movies.

29:29

Every time I try to, I

29:31

fall asleep. But I think so

29:33

much of what you just hit

29:35

on is so right. I think,

29:38

especially the parts where we are

29:40

amazed by it and then we

29:42

just take it for granted. And

29:44

actually, that is something, I don't

29:46

have the exact quote from here,

29:48

but when I interviewed Craig Federegi

29:50

a few months ago, weeks ago,

29:52

I don't know what day it

29:54

is, about Apple Intelligence and I

29:56

really was trying to go after

29:58

why a Siri like this, where

30:01

is the Siri improvements, I was

30:03

really hitting him on Siri. He

30:05

really did back up and and

30:07

really wanted me and I think

30:09

he even was thinking about how

30:11

far Syria has come and we

30:13

got used to it right right

30:15

we started like to want more

30:17

and more I want to pull

30:19

up the quote because he obviously

30:21

said it so eloquently but I

30:23

think yes and now even so

30:26

with these bought like you hit

30:28

a wall right you we hit

30:30

a wall with Siri and Alexa

30:32

and I say that in the

30:34

column we hit a wall we

30:36

now know we can say What

30:38

do you call it, Dingus? Dingus,

30:40

yeah. Hey, Dingus. Yeah, you can

30:42

say, hey, Dingus, set the blah.

30:44

Dingus, do this, right? We've all

30:46

learned the very strict formulas of

30:49

how to do those things. We've

30:51

had to talk the talk in

30:53

their way. We hit a wall

30:55

with that. So now we have

30:57

these bots. We are now hitting

30:59

another wall with that. I don't

31:01

know. I think they can change

31:03

the friendliness, and they can, there's

31:05

probably a lot. It's just still

31:07

a computer. There's more you can

31:09

do and it will walk you

31:12

through these things in a more

31:14

personable, friendly way, and it is

31:16

smarter, but you also will hit

31:18

this wall. Yeah, and it's, I

31:20

don't know, the fake personality, the

31:22

fake friendliness, I just... never would

31:24

have guessed it but now that

31:26

it's here it's oh but I

31:28

see it as of course they're

31:30

doing it this way because they

31:32

think it's non-threatening and they realize

31:34

that people are going to feel

31:37

threatened by the weirdness of this

31:39

and and then I think it

31:41

also sets in with everybody else

31:43

is doing it this way so

31:45

we should too nobody wants to

31:47

stick out and just sort of

31:49

put out a more much more

31:51

roboticotic personality type, which I think

31:53

after the novelty wears off, people

31:55

would settle in and will be

31:57

fine with joy better, right? Yeah,

32:00

and here's, and this is what

32:02

he said, he said, but as

32:04

humans are expectations for what it

32:06

means to use our voice to

32:08

communicate and ask for things is

32:10

almost unbounded. Which is right. It's

32:12

like we start to ask and

32:14

we start to feel more conversational,

32:16

right? So we've gone from the

32:18

like we know the formulaic. Now

32:20

we're in this moment with with

32:23

large language models where we're able

32:25

to. talk more like yourself. I

32:27

mean, that was kind of also

32:29

the amazing thing of being away

32:31

with these things for 24 hours.

32:33

You don't prompt or you're just

32:35

like, oh, let's do a thing

32:37

of yoga. Pick some, let's make

32:39

a yoga routine for me. Let's

32:41

do it now, right? Like you

32:43

don't talk to it. You just

32:45

talk normally. And so I guess

32:48

if the endpoint is Hal or

32:50

whatever that dream robot that is

32:52

our partner, our partner and maybe

32:54

friend. we're going to keep hitting

32:56

walls. It's like there's our internal

32:58

monologue in our own brains, but

33:00

nobody, you know, that's everybody's their

33:02

own. And the next level of

33:04

communication is speech. It's the oldest

33:06

part of evolution. Literacy is a

33:08

thing. And there are human beings

33:11

who are illiterate, who cannot read

33:13

or write. But there's, unless you

33:15

have like some kind of disability,

33:17

everybody can speak. And it's kind

33:19

of amazing. It's really, it's one

33:21

of those things where you're like,

33:23

wow, that is kind of amazing

33:25

that people of who are really,

33:27

really not very intelligent at all,

33:29

just naturally on their own, learn

33:31

to speak as babies. And what

33:34

do you do when you speak?

33:36

You communicate your thoughts and your

33:38

thoughts are for whatever your level

33:40

of intelligence is, for you, they're

33:42

unbounded. And... You just speak. And

33:44

so speaking to computer, it is,

33:46

it's a really interesting observation by

33:48

Fedoriki, because there really is no

33:50

limit. And it's, you and I

33:52

are in fact professional writers, or

33:54

we try to be, but it's,

33:56

I feel you and I are

33:59

at the higher end of aptitude

34:01

for expressing ourselves through writing, but

34:03

for most human beings, of whoever

34:05

you do, wherever you live, however

34:07

old you are, communicating yourself by

34:09

speech is. the most natural thing

34:11

in the world. And so interfacing

34:13

with computers that way is very,

34:15

very different than pushing buttons or

34:17

typing commands. It's just a complete,

34:19

it's like a removal of abstraction.

34:22

It is a level of abstraction,

34:24

but we're evolutionarily hooked up not

34:26

to think of it as being

34:28

abstract. You mentioned like, you know,

34:30

I think it was a 14-year-old

34:32

in Australia who recently was found

34:34

to have gotten obsessed with a...

34:36

Game of Thrones constructed character and

34:38

it sat all cases of teenage

34:40

suicide. It's all tragedies and there

34:42

are way too many people who

34:45

in the midst of depression turn

34:47

to suicide. With it always been

34:49

true, it still is true and

34:51

they're used to not even be

34:53

chat bots and now that there

34:55

are chat bots, somebody who's in

34:57

the midst of an episode like

34:59

this, that might have and probably

35:01

would have. been just as depressed

35:03

if not more so without any

35:05

access to chat bots, got obsessed

35:07

with chat bots, and now that's

35:10

the headline. It's same thing with

35:12

self-driving cars, where how many people

35:14

die in all human-driven car cars

35:16

and vehicles per day in the

35:18

US? It is one of those

35:20

things that once technology solves it,

35:22

people are going to look back

35:24

at... 2024 today and see our

35:26

driving culture and the number of

35:28

deaths and for all the deaths

35:30

all of the gruesome injuries that

35:33

people don't die and thankfully and

35:35

recover from as barbaric. It is

35:37

really really strange historically how we've

35:39

just and it's one of those

35:41

things that we've just accepted you

35:43

know we me and you and

35:45

every most of the people listening

35:47

to this we're all born in

35:49

car culture in North America. We

35:51

just accepted as the way the

35:53

world works, but it's really weird

35:56

and gruesome. and self-driving cars are

35:58

a way out of that. But

36:00

the problem is there are still

36:02

going to be some accidents, especially

36:04

getting from here where all, well,

36:06

we're already at a point where

36:08

some cars are self-driving, but getting

36:10

from the point where all cars

36:12

were human driven to a future

36:14

where all cars are either self-driving

36:16

robotically or have systems in place

36:18

to prevent. collisions and we're seeing

36:21

that where you can be like

36:23

not paying attention and if you

36:25

get within a certain proximity of

36:27

the car in front of you

36:29

the car brakes itself and fantastic

36:31

fantastic life-saving features and even at

36:33

a lesser degree at lower speed

36:35

just nobody was going to get

36:37

hurt but thank God I didn't

36:39

rear end that guy because I

36:41

wasn't paying attention and avoid just

36:44

the minor irritation of a Fenderbender.

36:46

But the problem is, it doesn't

36:48

matter which brand it is, but

36:50

of course, if it's Tesla, it

36:52

brings in all this other political

36:54

baggage and personality baggage of Elon

36:56

Musk. But one accident that kills

36:58

somebody with a self-driving car is

37:00

the man bites dog headline. Right?

37:02

Oh, there it is. But meanwhile,

37:04

there's 10,000 other people who die.

37:07

I don't know what the, you

37:09

know, but it's numbers like that

37:11

every month. in human driven cars.

37:13

And I think there's that sort

37:15

of effect with these chat bots.

37:17

100% and look, like, there's also

37:19

a big difference between the chat

37:21

bots that I, and I try

37:23

to make this point in the

37:25

piece in the video and the

37:27

column, but big difference between the

37:30

smaller startup character AI, which didn't

37:32

seem to have many protections in

37:34

at all in the app for

37:36

if somebody's. using words like suicide

37:38

or killing themselves to provide more

37:40

information versus what I saw here.

37:42

I tried to test that with

37:44

all of these and they're all

37:46

providing hotlines or talking and saying

37:48

talk to a real human. So

37:50

there's certainly also a lot that

37:52

can be done on the product

37:55

side. I don't want to say

37:57

like always the in product, it's

37:59

not. And I don't think, but

38:01

this, look, we've got, you've got

38:03

a lot of forces and you

38:05

could talk about lots of forces

38:07

that play into all of this

38:09

horribleness. But I agree with you,

38:11

I don't think we should hold

38:13

that up as the, not, and

38:15

by the way, I, it sounded

38:18

to me like. in that case

38:20

he was also typing. It was

38:22

a text-based chat. It wasn't voice,

38:24

though, character AI, and a few

38:26

of these do have the voice-based

38:28

stuff. But look, that makes it

38:30

more realistic and humid. I mean,

38:32

there's just, there's no doubt that

38:34

I had formed an image, especially

38:36

of meta-a-a-a-i, I gave the Kristen

38:38

Bell voice to it, and because

38:41

I just, I love Kristen Bell,

38:43

and it's like, I'm going to

38:45

be hanging out with Kristen Bell,

38:47

Just a chat bot, right? Like

38:49

there is something of a species.

38:51

There's something more there than just,

38:53

oh, I was texting with a

38:55

computer. Yeah, yeah, it tickles a

38:57

lower lizard part of your brain.

38:59

I guess lizards don't really talk

39:01

to each other, but you know

39:03

what I mean. Older, very old

39:06

evolutionary when we first started to

39:08

communicate with our voices, part of

39:10

our brains, where when you hear

39:12

the voice of a loved one,

39:14

it... It sets off endorphins in

39:16

your brain. If you just happen

39:18

to run into your kid accidentally,

39:20

you were out and about, and

39:22

your kid's school group happens to

39:24

be where you are, and you

39:26

hear your kid's voice, it's whoa.

39:29

And it's like, you know, your

39:31

brain kind of lights up in

39:33

a funny way. You know, just

39:35

when you hear like a friend

39:37

who you haven't seen in a

39:39

while, they just happen to be

39:41

in the same store as you,

39:43

and you hear their voice, and

39:45

it's like, whoa. When you see

39:47

somebody's face, you recognize it and

39:49

you have a reaction that's not

39:52

voluntary, it's involuntary and you don't

39:54

get that. I mean, a weirdo

39:56

like me can get it from

39:58

like type faces or something. But

40:01

even there, even for someone who's

40:03

obsessed with something like that, ooh,

40:05

this book has a real, oh,

40:07

I love this fun. It makes

40:09

me a little more likely to

40:11

buy it. It's still not the

40:13

same as when I see the

40:15

face of a loved one or

40:17

something like that. We're hooked up

40:19

for that and computers have never

40:21

had that before. Yeah. And now

40:23

they do. Kind of, right. And

40:25

that's the other thing that science

40:27

fiction didn't really imagine. It's like

40:29

science fiction always gives us. the

40:32

finished product right and again even

40:34

if you haven't watched 2001 how

40:36

it's the the red eye and

40:38

and C3PO it's like but then

40:40

nobody ever imagined like hey what

40:42

was it like the first year

40:44

where they had this technology where

40:46

you could talk to them and

40:48

it was and it's so different

40:50

than I think anybody would have

40:52

imagined it to be it's because

40:54

if you think about it superficially

40:56

from the perspective of a science

40:58

fiction writer 20, 30, 40, or

41:00

more years ago, you start thinking

41:03

like, well, maybe because it's the

41:05

early days, it can only understand

41:07

three or four things or something,

41:09

because superficially that seems like a

41:11

good way to get it going.

41:13

But no, they'll answer anything. They'll

41:15

answer questions about anything. And they're

41:17

at times incredibly amazing and knowledgeable

41:19

and give you these helpful answers

41:21

and then at other times are

41:23

more stupid than any person you've

41:25

ever even fathomed. You know, like

41:27

when Google turned on the AI

41:29

in their search and was giving

41:31

really funny answers to how many

41:34

rocks should I eat on a

41:36

daily basis, right? Right. Or had

41:38

to glue the cheese on to

41:40

pizza. Like... Yes! Shouldn't. Yeah, they're

41:42

just engineered to be confident. Right.

41:44

And it takes a human being

41:46

at the moment to come up

41:48

with a question like, how many

41:50

rocks should I eat a day

41:52

to... trick the LLC-based AI into

41:54

giving a funny answer because it's

41:56

a nonsensical question in a way,

41:58

right? Right. There's a, you probably

42:00

haven't seen it given what you

42:02

said about science fiction movies. The

42:05

movie Blade Runner from 1982, I

42:07

believe, spoiler here, but there's the

42:09

whole premise of the movie is

42:11

Harrison Ford plays a Blade Runner

42:13

who is a... police detective whose

42:15

job it is to hunt down

42:17

replicates, replicates are robots who look

42:19

like humans. And so you, they

42:21

have skin and completely visually look

42:23

and talk like humans, but they're

42:25

fake underneath and the ones who

42:27

get out and are illegally out

42:29

and about, you have to identify

42:31

them. And there's a thing called

42:34

the void comp, comp test where

42:36

it's a series of questions that

42:38

the police can ask and you

42:40

give 20 questions and a replicant

42:42

is going to get tricked up

42:44

by some of them. And it's

42:46

a really good scene in the

42:48

movie. It's everybody who's watched it

42:50

is one of the most memorable

42:52

parts of the movie. But the

42:54

type of questions they ask are

42:56

much more thoughtful and cool and

42:58

interesting and they're not, how many

43:00

rocks should you eat a day?

43:02

Right? Like it turns out at

43:05

our current moment, the way to

43:07

figure out if you're chatting with

43:09

an AI is to ask a

43:11

nonsense question. Yep. Not a deep

43:13

philosophical question. It's true. Well, I'm

43:15

making a list of all the

43:17

movies I need to watch. All

43:19

right. Blade Runner, I might put

43:21

you to sleep. I love it,

43:23

but it looks good. But a

43:25

lot of times I'll start, I've

43:27

started a lot of these movies

43:29

and then I just fall asleep.

43:31

Here, let me hit the money

43:33

bell again. Do another sponsor. Thank

43:36

you. To our good friends, this

43:38

time it's Squarespace. Squarespace is the

43:40

all-in-one platform for building your own

43:42

website online. Everything you need to

43:44

do to own your own website,

43:46

Squarespace can do for you. Domain

43:48

name registry. picking templates to what

43:50

it looks like, modifying what it

43:52

looks like through their fluid engine,

43:54

which is the name of their

43:56

next generation engine for modifying the

43:58

way the thing looks. And the

44:00

fluid engine works great, whether you

44:02

are on a desktop computer or

44:04

on your phone, any modern web

44:07

browser, you just build your Squarespace

44:09

website on the Squares website itself.

44:11

And what you see is what

44:13

the people who visit. your site

44:15

will get it is WhizzyWig for

44:17

making a website and it works

44:19

absolutely terrific. They have all sorts

44:21

of other features I mean just

44:23

about everything you can imagine. Squarespace

44:25

payments is the easiest way to

44:27

manage payment in one place with

44:29

Squarespace. Onboarding is fast and simple.

44:31

You get started with a few

44:33

clicks and you can start receiving

44:35

payments right away and you can

44:38

give your customers any way to

44:40

pay imaginable. debit in the US,

44:42

Apple pay support, after pay in

44:44

the US and Canada, clear pay

44:46

in the UK, all sorts of

44:48

stuff like that. Credit cards, of

44:50

course, easily accepted. They have all

44:52

sorts of other features invoices. So

44:54

if you're setting up not a

44:56

website for just personal presence, but

44:58

actually running a business and you

45:00

need to send invoices and stuff

45:02

like that, they've got an invoicing

45:04

feature built into the platform that

45:06

you can use. Analytics, so you

45:09

can see the stats, how many

45:11

people are coming to your site,

45:13

where are they going to on

45:15

your site, and where are they

45:17

coming from, all sorts of stuff

45:19

like that. And it's such a

45:21

great analytics interface, where instead of

45:23

looking an airplane dashboard, and it's

45:25

all confusing, and it's a beautiful,

45:27

beautiful presentation, it just makes it

45:29

clear. Information graphic-wise is just absolutely

45:31

top-notch. Where do you go to

45:33

find out more? talk show Squarespace.com/talk

45:35

show and by using that URL

45:38

they'll know you came from this

45:40

show and you the listener who's

45:42

signing up will save 10% off

45:44

your first purchase of a website

45:46

or domain and you get 30

45:48

days free just by going there

45:50

to start 30 days to start

45:52

build the whole website whole month

45:54

you can build it out use

45:56

it turn it on go live

45:58

make sure you like it and

46:00

only after the 30 days are

46:02

up you need to pay just

46:04

go to square space.com/talk show you

46:06

did not include Siri in your

46:09

your girls trip I didn't. And

46:11

the reason why it was so

46:13

funny, you said, Siri, can you

46:15

introduce yourself? To get, I guess,

46:17

B-roll footage of Siri introducing itself.

46:19

And Siri's answer was just to

46:21

confirm. Do you want to turn

46:23

this device off? Yeah, I mean,

46:25

I was looking for a moment

46:27

to just have like a sound

46:29

like Siri having a basic answer,

46:31

but Siri knew exactly how to

46:33

write itself into the script. I'm

46:35

sorry there because I said Syria

46:37

and now I've got device. Oh

46:40

yeah, now wait, right. Dingus and

46:42

Dinkus. Beeping and boppin. But I

46:44

noticed it back at WWDC when

46:46

Apple unveiled the whole Apple intelligence

46:48

array of features and they said

46:50

it that with this new interface

46:52

that came out with iOS 18.1

46:54

last month. where the new Siri

46:56

gets this all-new visual interface, we're

46:58

like, let's just speak about the

47:00

phone, where when you invoke it,

47:02

you get a whole new visual

47:04

interface where the border of the

47:06

screen lights up in a Siri

47:08

rainbow of colors, and you can,

47:11

they, in a typical Apple fashion,

47:13

it made it seem like it's

47:15

nothing but great, which is that.

47:17

You can continue a conversation. So

47:19

you could say, hey, Dingus, who

47:21

did the Baltimore Ravens play last

47:23

week? And the series will give

47:25

an answer. And once you're still

47:27

up, you could say, and who

47:29

did they play the week before?

47:31

And Siri, in theory, should know

47:33

we're still talking about the Baltimore

47:35

Ravens or whatever it was, and

47:37

then give you an answer from

47:39

the week before. But then once

47:42

you're done and that Siri interface

47:44

goes off, then you're starting over

47:46

each time. And so the continuing

47:48

context in A session, is it

47:50

definitely better than the way Siri

47:52

used to be. But the fact

47:54

that it forgets everything by design

47:56

each time you started up is

47:58

very different from these other chat

48:00

bots. And I think that alone

48:02

would have ruled it out from

48:04

inclusion in your girls AI weekend.

48:06

Right? Well, I think it's also

48:08

the, like, I was just testing

48:10

something using Apple Intelligence too, just

48:13

to see. I've been testing 18.2

48:15

and wanted to see about the

48:17

ChatGBT integration and if when you

48:19

do start talking to Siri if

48:21

it's going to really start to

48:23

draw on ChatGBT and it really

48:25

doesn't. No, because I've been running

48:27

18.2 pretty much since the first

48:29

beta too. And it doesn't. Yeah,

48:31

I mean, and that's supposed to

48:33

start happening. I'm finding in my

48:35

testing of 18.2 that it I

48:37

don't mind that this chat GPT

48:39

integration is there but I'm finding

48:42

that it's kind of if it

48:44

weren't there my opinion of the

48:46

overall thing would hardly be different

48:48

it really And I don't know

48:50

if that's because leading up to

48:52

WWDC, I mean there was a

48:54

lot of reporting, Mark German, of

48:56

course, the king of the rumor

48:58

reporting in Apple World had a

49:00

bunch of stories leading up to

49:02

WWDC, suggesting that the deal with

49:04

Open AI didn't really... get finalized

49:06

until the last minute or close

49:08

to the last minute. And Apple

49:10

said at WWDC that there will

49:13

be or could be future partners

49:15

like Google in the future that

49:17

you could switch from using chat

49:19

GPT as your answer questions Syria

49:21

alone can't partner to Gemini or

49:23

something else. Here's what I think

49:25

is happening. I think the main

49:27

reason is that Siri does not

49:29

have a strong, large language model

49:31

component to like, if that's just

49:33

the way it worked out for

49:35

this year, that it would still

49:37

be, we can still tell people

49:39

by these things for Apple Intelligence.

49:41

And I think it shows. But

49:44

I think what's happening, specifically in

49:46

18.2, because you asked, like, why

49:48

wouldn't Siri fit into this? And

49:50

I think the main reason is

49:52

that Siri does not have a

49:54

strong large language model component to,

49:56

like, be conversational. See my interview

49:58

with Craig. We went deep on

50:00

that and he definitely hinted at

50:02

he didn't hint He just said

50:04

is that where this is going

50:06

of course But we've got to

50:08

figure out kind of the right

50:10

parts to put that together and

50:12

so because he was very clear

50:15

This is the garage example. He

50:17

was very clear right now Siri

50:19

does the garage, it does it

50:21

very well. But what we don't

50:23

want is Siri to go off

50:25

and not do that thing very

50:27

well. So they have to figure

50:29

out the right way and that

50:31

to me made a lot of

50:33

sense. You have billions of users

50:35

using this thing, got to make

50:37

sure that it can do the

50:39

thing everyone relies on it for,

50:41

but then also be able to

50:43

go and do the advanced thing.

50:46

One thing though that I think

50:48

is happening right now in Siri.

50:50

So like I just asked Siri.

50:52

Give me some... good recipes for

50:54

meatballs. I don't know, I always

50:56

go to meatballs as my like

50:58

go-to test of it, right? And

51:00

it searched the web and found

51:02

that and gave me like the

51:04

little pop-up that has that. I

51:06

tried a couple things and it

51:08

didn't bring up ChatGPT, but when

51:10

I asked it to write a

51:12

poem about John Gruber, it did.

51:14

It did. It asked me, yes,

51:17

what happened to the poem, hold

51:19

on. away. Hold on, the nice

51:21

screenshot. Yeah, see, it's clearly by

51:23

design and I think partly to

51:25

cover up for limitations in our

51:27

own system, but also partly because

51:29

they don't want to even get

51:31

into that now, like the permanent

51:33

memory of your interactions with the

51:35

service. They really want that blank

51:37

slate each time you bring it

51:39

up. Hold on, I'm asking to

51:41

do it again, but I think

51:43

that what's happening is that there's

51:46

very specific types of prompts. where

51:48

it says, write a poem or

51:50

whatever the other ones are, that

51:52

they're saying, hey, let's go to

51:54

chat cheapity for that. But then

51:56

for some other things where a

51:58

large language model could be useful,

52:00

it's not doing that. And I

52:02

also wonder if I had asked

52:04

it to write a poem about

52:06

John Gruber and I was doing

52:08

it, because I'm doing it by

52:10

text the answer, right? I have

52:12

found myself, I don't care enough

52:14

about these things to use them

52:17

all. And so I've sort of

52:19

gone all in with chat GPT,

52:21

and I mainly also, I like

52:23

the answers, I think the 400

52:25

model is probably the best or

52:27

closest to the best for the

52:29

sort of things I'm using it

52:31

for, but I really love their

52:33

Mac app. The chat GPT Mac

52:35

app is by far in a

52:37

way the best. Macintosh app for

52:39

any of these bots and I

52:41

know it all day long. It's

52:43

just a really well engineered Mac

52:45

app. It's not an electron thing

52:48

that's a wrapper for their website.

52:50

It's a really good fast with

52:52

a great interface and it's interesting

52:54

that it has your whole history

52:56

with it because I have I'm

52:58

logged in and I have a

53:00

chat GPT account and I pay

53:02

10 bucks a month for whatever

53:04

and there was a I don't

53:06

know if it's a meme or

53:08

what but like. a thing a

53:10

couple days ago where people are

53:12

saying to ask chat GPT or

53:14

ask any of these things to

53:16

make a picture, how do you

53:19

imagine a typical day in my

53:21

life? And I don't think ChatGBT

53:23

knows a lot about me, but

53:25

I'd imagine the picture I got,

53:27

it didn't have a lot, some

53:29

of the people's pictures have so

53:31

much in them. Mine was like,

53:33

of course all the rooms look

53:35

idyllic, it was sort of a

53:37

rustic, lots of wood, a big

53:39

leather, old leather couch, which is

53:41

kind of me. Like if I'm

53:43

gonna have a couch, I don't

53:45

have a couch in my office,

53:47

but if I did, I probably

53:50

would have a leather one, and

53:52

I like an old-looking leather couch

53:54

leather couch leather couch leather couch,

53:56

a typewriter on a desk, not

53:58

a computer, but a typewriter, and

54:00

some, not a lot in real

54:02

life, I've probably got more crap

54:04

around my office, but some books

54:06

and papers on the floor, and

54:08

I am kind of a mess.

54:10

And I do, one of my,

54:12

my, my, one of my favorite

54:14

most used things that I ask

54:16

chat GPT is when, it's like

54:18

my... superpowered thesaurus. Either I know

54:21

I'm looking for a certain word

54:23

and I can't think of it,

54:25

or I want a better version

54:27

of a word. I'm always asking

54:29

chatGPT for either a specific word

54:31

I can't think of, or like

54:33

a better word than the one

54:35

I'm thinking of. And it's amazing

54:37

at how good it is at

54:39

it. It's so good at it.

54:41

And so based on that it

54:43

kind of knows I'm a writer

54:45

and kind of seemingly knows I'm

54:47

a little scattered. that I do

54:50

research then build up piles of

54:52

books and oh there are lots

54:54

and lots of books in in

54:56

my imagined chat gPT world. So

54:58

this is the poem I did

55:00

write about you now I finally

55:02

got it working again. John Gruber

55:04

with wisdom keen in the world

55:06

of tech he's often seen with

55:08

mark downs elegance and grace his

55:10

thoughts and code find their place.

55:12

I'm not going to keep going

55:14

on. That's pretty good though. Yeah

55:16

that's pretty good. Okay, my my

55:18

image of that chatch UBT did

55:21

of my world, which was pretty

55:23

funny. I mean, I tried it

55:25

a few of them, but I

55:27

posted it on threads. It thinks

55:29

I'm, and it's not wrong, but

55:31

it must think like I'm truly

55:33

like the world's leader of the

55:35

Cub Scouts. I don't know what

55:37

that's called, like the, the, the,

55:39

the, America. They've changed the name

55:41

to Scouting America. Right, right. Because

55:43

it's not boys. To get rid

55:45

of the boys and girls separate

55:47

aspects. But my son, my seven

55:49

year old did just join the

55:52

Cub Scouts. And so I have

55:54

been using it a lot because

55:56

it's complicated. I'm like, please explain

55:58

to me, like I've been asking

56:00

about dens versus packs. I've been

56:02

asking about what are the different

56:04

den names or what are the

56:06

different pack names and the Cub

56:08

Scouts and there's wolves and there's

56:10

bears and there's this. And so

56:12

it really took that to heart

56:14

so that everyone I generated, I'm

56:16

like, I've got Cub Scout paraphernalia

56:18

everywhere, I've got patches, I've got

56:20

kids in the background with Cub

56:23

Scouts. And I guess at some

56:25

point I also asked about basketball.

56:27

I want to say at some

56:29

point I was asking about the

56:31

best way to get a basketball

56:33

hoop into cement. Because that was

56:35

something I was struck. Like I

56:37

use it a lot for home

56:39

stuff. I sent you a lot

56:41

for home stuff. I sent you

56:43

a picture of mine. I should

56:45

post it on social media and

56:47

I'll put I'll put links to

56:49

years from threads or wow I

56:51

mean this is a beautiful home

56:54

that you have here from the

56:56

early I don't know 1920 well

56:58

well it is it you know

57:00

what though you know what's funny

57:02

is in the early years of

57:04

daring fireball you know to I

57:06

started the site in 2004 And

57:08

there weren't, YouTube wasn't a thing,

57:10

and I've never really done much

57:12

video, and podcasts weren't yet a

57:14

thing. So all people knew me

57:16

by was the writing. And I

57:18

didn't put a picture of myself

57:20

on the site. And I started

57:23

going to things like WWDC or

57:25

Macworld Expo, and people would meet

57:27

me for the first time. And

57:29

I mean like dozens of times,

57:31

people would just say, oh, I

57:33

thought you were really old. And

57:35

at the time, it was like

57:37

20 years ago, I was only,

57:39

I don't know, like 30. So

57:41

I wasn't really old at all,

57:43

and they were like, oh, I

57:45

thought you were like old. Somehow

57:47

ChatGPT seems to have the same

57:49

impression of me. Yeah, the old

57:51

sounding written voice. I guess. I

57:54

guess. Because it's not just a

57:56

typewriter. Or you were very wise.

57:58

It's a manual typewriter. Yeah. I

58:00

also enjoy that there's a map

58:02

on the wall of mine and

58:04

it's like a nonsense map. It's

58:06

not a real place. You have

58:08

a lot of plants. But yeah,

58:10

I mean, really, I just sent

58:12

you mine, but you can see

58:14

I've got the Boy Scouts, I've

58:16

got basketballs, it thinks I love

58:18

bagels. Hmm. I don't know about

58:20

the bagel piece. Oh yeah, see

58:22

yours is really yours are very

58:25

very Jam packed with things. Look

58:27

how they got they jam packed

58:29

it. Oh and I love that

58:31

little safari compass. They gave me

58:33

on the first one. Oh, oh

58:35

the first one. Look at that

58:37

safari compass on the table. Oh,

58:39

yeah, yeah, I see it. It's

58:41

like a like a coaster or

58:43

a pen or something like that

58:45

and now I'm like I want

58:47

that I want that for the

58:49

Cub Scouts. I want a safs.

58:51

Yeah, they've definitely got you as

58:53

being like the scouting mom. Yeah,

58:56

definitely. I mean, but it was

58:58

very appropriate because that week I

59:00

had obviously asked a lot about

59:02

this, but my life is, I

59:04

wish, I mean, I don't wish

59:06

I was as involved with the

59:08

Boy Scouts as this thinks I

59:10

am. And also in that second

59:12

one, you've got a Boy Scout

59:14

framed logo on the wall. Yes,

59:16

I love them so much. Yep.

59:18

And also in that second one,

59:20

you've got a wood-grained... sign on

59:22

the wall big with your name

59:24

on it. Yep, I love myself.

59:27

I love myself. I love basketball,

59:29

apparently. But you have an iMac.

59:31

You have an iMac, not a,

59:33

you would know, you have an

59:35

iMac on a desk and a

59:37

laptop on your lap and another

59:39

like a tablet. A detached keyboard.

59:41

Yeah. Yeah. So you've got a

59:43

lot of computers. I don't have

59:45

any, but. Yeah. Well. But just

59:47

going back to the chat. You.

59:49

Look, I think there's something there

59:51

about how Apple's thinking about how

59:53

much they, obviously we know that

59:55

they've taken some, they wanted to

59:58

minimize some of the risk. right,

1:00:00

and they also, whatever we want

1:00:02

to say they're behind or whatever

1:00:04

the reason is, they're being cautious

1:00:06

about how they roll this out. And

1:00:08

so they've used chat cheapity and

1:00:10

potentially some whatever partners they can

1:00:13

to plug into Siri. Eventually,

1:00:15

as Craig said in that interview, we

1:00:17

will see a Siri that has

1:00:19

more large language model capability and

1:00:21

can converse and can likely do

1:00:23

a lot of what we've already

1:00:25

been seeing from these other bots that

1:00:28

I went to the cabin with. It's

1:00:30

just they're not, they're just not,

1:00:32

it's just not comparable right now. And

1:00:34

look, we're going to see this from

1:00:36

Alexa. I mean, Amazon has been delaying

1:00:38

and delaying, but that's going to come

1:00:40

soon too. So Apple's going to have no

1:00:42

other choice. Because it is conspicuous, and I

1:00:44

thought of that with your, I thought of

1:00:47

it just earlier on the show, when you

1:00:49

were trying to remember all four of the

1:00:51

companions who you took with you, that I

1:00:53

think Jim and I was the one, the

1:00:55

fourth of four that you couldn't think of.

1:00:57

But it does stand out. Which is, it

1:00:59

is a little forgettable. We're not

1:01:02

talking about Amazon at this

1:01:04

point, where clearly they're working

1:01:06

on it. And I think Amazon is

1:01:08

sort of taking, Amazon and Apple

1:01:10

are the two, I was gonna

1:01:13

say weirdos, but exceptions. They're doing

1:01:15

things in very different ways from

1:01:17

the others. And Apple's is, let's

1:01:19

be very deliberate and slow, but

1:01:22

we'll start with. Very simple stuff.

1:01:24

So like when iOS 18.1 came out,

1:01:26

which is the only non-bata version

1:01:28

of Apple intelligence that anybody

1:01:31

out there listening has, it

1:01:33

doesn't even do much, right? There's

1:01:35

not even that much to it.

1:01:37

It's got like the writing tools,

1:01:39

doesn't do any of the image

1:01:41

generation stuff. They're releasing

1:01:43

stuff, but it's like a little bit

1:01:45

at a time. And I get the

1:01:47

feeling that Amazon... rather than release a little

1:01:49

bit at a time is sort of holding back

1:01:52

and then they're going to come out with a

1:01:54

big thing at once. I don't know. I could be wrong.

1:01:56

And I think, you know, as much as there

1:01:58

was some controversy, not controversy. but

1:02:00

like when Apple Intelligence launched

1:02:02

a few weeks ago there were many

1:02:04

people on the side of Apple's late

1:02:06

they're behind they're clearly behind there's

1:02:08

reporting that shows they're behind that

1:02:11

was one camp then there's the

1:02:13

camp which I don't didn't actually

1:02:15

take a stance on in my piece

1:02:18

it was mostly leaning on Craig who

1:02:20

was saying we're doing this to be

1:02:22

deliberate we're deliberately being slow we want

1:02:24

to get it right we're doing it

1:02:26

all right because we have a responsibility

1:02:29

Probably some place in the

1:02:31

my opinion is that it's in some

1:02:33

place in the middle of those things

1:02:35

But when you think about Apple

1:02:37

and Amazon specific I

1:02:40

mean the two big Assistants that

1:02:42

people make fun of and talk about

1:02:44

the most are Syria and Alexa

1:02:47

Yeah Google assistant as

1:02:49

well, but I think to a lesser

1:02:51

degree they have the most riding

1:02:53

on it Right if people wake

1:02:55

up and Alexa can't do the

1:02:57

basic things They can't turn

1:02:59

on the lights or start doing those

1:03:01

things in a ridiculous crazy way People

1:03:03

are gonna freak the hell out Yeah,

1:03:05

the me We haven't hooked up to do

1:03:07

real things to do real talk about the garage

1:03:10

if the garage just doesn't do the

1:03:12

thing or Alexa's not turning on the

1:03:14

lights or Alexa's turning on your fireplace

1:03:16

and like Those all are real things

1:03:18

and these companies built massive

1:03:20

business around them again. Google has

1:03:23

this too, but Google's done it

1:03:25

weird in a typical Google way.

1:03:27

They've put Gemini Live within assistant

1:03:29

and it but assistant still lives

1:03:31

there. It's a mess. It's total

1:03:33

mess. It's Google that I was thinking

1:03:35

of when I retracted my use of

1:03:37

the word weird for Apple and Amazon

1:03:39

and said their exceptions and doing

1:03:41

it differently because it's Google

1:03:43

that's always the weird. I mean they

1:03:46

just don't a weird thing too. You can

1:03:48

still set it. If you ask Gemini Live

1:03:50

set a timer, it won't. But you can

1:03:52

still go back to Google Assistant and

1:03:54

set a timer. And they worked out a new

1:03:56

way with extensions to set a timer, but

1:03:58

it's all like tactic. They obviously just wanted

1:04:01

to get it out the door.

1:04:03

Again, do I think Amazon and

1:04:05

Apple were behind and are playing

1:04:07

ketchup? Yes, but I also think

1:04:09

it's not as easy as them

1:04:11

to, for them, to just say,

1:04:13

rush it out, rush it out.

1:04:15

And I think the proof that

1:04:17

Apple was caught flat-footed is the.

1:04:19

Clear, obvious, I mean, at my

1:04:21

live show after WWDC, I asked

1:04:23

about it, and it was one

1:04:25

of those questions that I thought,

1:04:27

there's no way they're going to

1:04:29

give me a straight answer. And

1:04:31

they just answered straightly, yeah, that's

1:04:33

an issue, is the amount of

1:04:35

RAM in the device for on-

1:04:37

device processing. And yeah, it's like,

1:04:39

Apple Intelligence really needs, whatever the

1:04:41

device, iPhone or iPad or Mac,

1:04:43

it needs at least eight gigabytes

1:04:45

of RAM, don't get it. And

1:04:47

wow, I was like, that's a

1:04:49

really clear answer to the sort

1:04:51

of question Apple executives usually don't

1:04:53

answer. And the fact that so

1:04:55

many recent devices, most conspicuously, last

1:04:57

year's iPhone 15 non-pros, don't have

1:04:59

that much RAM. We know that

1:05:01

things like that get set, especially

1:05:03

for the iPhone, because of the

1:05:05

massive 100 million units per year

1:05:07

scale, whatever it is that's done,

1:05:09

a year in advance, right? Like

1:05:11

the iPhone 17 from next year

1:05:13

are locked in at this point.

1:05:15

I know. there's I've talked to

1:05:17

people at Apple where there are

1:05:19

certain little things that could happen

1:05:21

as late as December or January

1:05:23

but for the most I mean

1:05:25

something like how much RAM is

1:05:27

on the actual system on a

1:05:29

chip that's already set in stone

1:05:31

probably months ago if not even

1:05:33

longer the fact that they put

1:05:35

a lot of especially iPhones out

1:05:37

in the world that don't have

1:05:39

enough RAM for on device Apple

1:05:41

intelligence processing is a sign that

1:05:43

they were caught flat-footed but I

1:05:45

also think Apple is kind of

1:05:48

I don't think they're happy about

1:05:50

that and I think if they...

1:05:52

could go back in time and

1:05:54

say, hey, maybe start putting eight

1:05:56

gigs of RAM in the iPhone

1:05:58

14 or 13. They would, and

1:06:00

I think they kind of regret

1:06:02

it. But I think overall, they're

1:06:04

comfortable, and Tim Cook said it

1:06:06

in his Wall Street Journal profile

1:06:08

recently, that it's very, very clear.

1:06:10

It's a long time Apple message.

1:06:12

Our aim is to be the

1:06:14

best, not the first. And they're

1:06:16

comfortable with that. I think Google

1:06:18

was very, very, very uncomfortable. culturally

1:06:20

being accused of being late or

1:06:22

behind and it sort of panicked

1:06:24

and still does and it's just

1:06:26

sort of like well okay here

1:06:28

fine here's everything we know how

1:06:30

to do it is a lot

1:06:32

and look they had researchers that

1:06:34

worked on these transformers I mean

1:06:36

it's of all places to be

1:06:38

behind oh they had that reason

1:06:40

to feel like this was happening

1:06:42

in our own house everyone in

1:06:44

the house get get it together

1:06:46

get it together everyone everyone that

1:06:48

it's a different situation across the

1:06:50

405 or wherever they are down

1:06:52

in geographic, I don't know. California

1:06:54

geography, not my thing. I do

1:06:56

think Apple's comfortable with their position,

1:06:58

but I will say I really

1:07:00

don't like the ad campaign. I

1:07:02

would say it might be my

1:07:04

most disliked Apple advertising campaign that

1:07:06

I can remember is the commercials

1:07:08

they're doing for Apple intelligence. Because

1:07:10

I feel like all of them.

1:07:12

My wife even says when they

1:07:14

come on, it's sports season when

1:07:16

she sees, because we don't really

1:07:18

see a lot of regular TV

1:07:20

commercials, but I'm watching football or

1:07:22

when the Yankees were still playing

1:07:24

and she's sitting with me on

1:07:26

the couch and she's like, why

1:07:28

would I want that? Like, there

1:07:30

was a commercial where somebody, and

1:07:32

some of it too, it just

1:07:34

makes the human beings in the

1:07:36

ads look rude. There's one where

1:07:38

the actress, she's the actress from

1:07:40

the last of us and she

1:07:42

has like a... a meeting with

1:07:44

an agent or somebody in the

1:07:46

entertainment industry and I did you

1:07:48

read my script and she like

1:07:50

looks at her phone and there's

1:07:52

an email with the script she's

1:07:54

like summarize and she's right there.

1:07:56

It's all A. It's just rude.

1:07:58

Top level. It is rude to

1:08:00

show up at a meeting where

1:08:02

you're supposed to have read the

1:08:04

script not having read the script.

1:08:06

It's rude. It's rude and inconsiderate.

1:08:08

And B. You gave me reading

1:08:10

materials before this podcast and I

1:08:12

absolutely read them. I used Apple

1:08:14

intelligence to summarize your text to

1:08:17

me. Yeah. B, how stupid do

1:08:19

you think this other person is

1:08:21

when you're staring at your phone

1:08:23

poking at it and going, oh

1:08:25

yeah, yeah, it's a story about

1:08:27

coming of age of whatever the

1:08:29

plot is. It was pretty good.

1:08:31

Then it's like the commercial's implicit

1:08:33

message is that the other person

1:08:35

is so stupid that they don't

1:08:37

see that you're reading a summary

1:08:39

of the script on the fly

1:08:41

right in front of them. And

1:08:43

then that you're lying about it

1:08:45

and saying that you liked it

1:08:47

when you haven't read it. It's

1:08:49

all, all of those things are,

1:08:51

I don't see any of that

1:08:53

as positive, right? And then there's

1:08:55

the one too, where Bella Ramsey

1:08:57

is her name, right? Yeah, Bella

1:08:59

Ramsey. Where she runs into somebody,

1:09:01

she sees them like down the

1:09:03

hall or whatever. It's like, remind

1:09:05

me of the person I had

1:09:07

coffee with at blank. Cafe. I

1:09:09

actually would love that, but that

1:09:11

doesn't exist right now on the

1:09:13

product. No. That's the one commercial

1:09:15

that, and I've seen some people

1:09:17

complain about that one too. That's

1:09:19

the one in the campaign that

1:09:21

I like the most because I

1:09:23

am often bad with names. Oh,

1:09:25

I'm so bad with names. I

1:09:27

always have been. I'm not worried

1:09:29

that it's a sign of dementia

1:09:31

creeping up on me. But I've

1:09:33

always been bad with names, so

1:09:35

I do look forward to a

1:09:37

future where I'll get... some kind

1:09:39

of assistance. You know, and I

1:09:41

wear glasses all the time now,

1:09:43

so I have a leg up

1:09:45

on that. If all my glasses

1:09:47

do is just tell me the

1:09:49

name of everybody I'm looking at

1:09:51

or give me a button. I

1:09:53

can do it, I would buy

1:09:55

those glasses in a second. I

1:09:57

love how we all want that

1:09:59

thing, but we know that that

1:10:01

is the worst privacy situation in

1:10:03

the history of the world. Right.

1:10:05

And also, by the way, maybe

1:10:07

Apple is the only one that

1:10:09

can deliver that kind of feature

1:10:11

with privacy because of the iPhone

1:10:13

integration, but probably another conversation. Yep,

1:10:15

it is a whole other conversation

1:10:17

where the whole idea of face

1:10:19

recognition is this enormous can of

1:10:21

worms for privacy and... civil liberties

1:10:23

and I'm not being blithe about

1:10:25

those very serious things that I

1:10:27

agree with are true. But in

1:10:29

my dream world, I want the

1:10:31

feature, I want it available, and

1:10:33

I want it to be provably

1:10:35

private. I absolutely want this. And

1:10:37

in fact, we had a great

1:10:39

story in the journal this week

1:10:41

by a colleague of mine, Anne-Marie,

1:10:43

and she wrote about how Apple

1:10:46

Notes is being used for all

1:10:48

these ridiculously funny things. And I

1:10:50

actually featured one in my newsletter

1:10:52

today about how people like make

1:10:54

stickers of their outfits and put

1:10:56

them in notes. Anyway, check that

1:10:58

out. But my reason for bringing

1:11:00

this up was is that we

1:11:02

had this conversation about notes and

1:11:04

that to me, that's the most

1:11:06

private thing. If that were to

1:11:08

ever leak, I would be devastated

1:11:10

and I would be canceled. Truly.

1:11:12

Not even maybe publicly, but just

1:11:14

like by friends, right? Because I

1:11:16

have very specific notes about people's

1:11:18

names or people mom of so

1:11:20

and so Where so and so

1:11:22

her name is blank, right? This

1:11:24

is I have that is what

1:11:26

I have to do. So I

1:11:28

don't seem so rude every time

1:11:30

I run into them at a

1:11:32

practice or something I I do

1:11:34

it to join I would be

1:11:36

canceled right with you a lot

1:11:38

of mine the notes like that

1:11:40

are in my contacts not Apple

1:11:42

notes. Oh It's the cop, but

1:11:44

there are ones, I have some

1:11:46

in Apple nodes. You know, it's-

1:11:48

Talk about this system a little

1:11:50

bit, because how would you know

1:11:52

the name of the person? I

1:11:54

kind of know which people, I

1:11:56

have an intuitive sense of who

1:11:58

I've saved. contact from. So when

1:12:00

Jonas was in school for a

1:12:02

lot of his friends parents they're

1:12:04

all in Apple notes because I

1:12:06

don't have contacts for most of

1:12:08

them. I in fact I had

1:12:10

I think it was just one

1:12:12

Apple note with Jonas school friends

1:12:14

or something like that. Yes yes

1:12:16

exactly. And but I have stuff

1:12:18

in there that I definitely wouldn't

1:12:20

want to come out like somebody

1:12:22

has a gap in the teeth.

1:12:24

That's how you remember. Yeah, right.

1:12:26

Not the worst thing in the

1:12:28

world. It's just but it's like

1:12:30

the mom with the gap tooth.

1:12:32

It's so and so's mom. And

1:12:34

it helps me. No, I have

1:12:36

the same crap in mind. It

1:12:38

is only the parenting nightmare. Yeah,

1:12:40

it's not rude. And I do

1:12:42

trust that my Apple notes are

1:12:44

not going to be leaked publicly.

1:12:46

And it is a note to

1:12:48

my future self. It is John

1:12:50

Gruber right now making a note

1:12:52

to John Gruber in the future

1:12:54

after I've forgotten whose mom she

1:12:56

is. And so nobody who's ever

1:12:58

intended and that is, oh yeah,

1:13:00

the gap tooth mom. I know.

1:13:02

And I'm like, oh yeah, yeah.

1:13:04

And you also know you can

1:13:06

go search gap tooth mom and

1:13:08

it'll come up right there right

1:13:10

there. Because I know that's it.

1:13:12

Right. Because I remember that's the

1:13:15

thing that I would, that I

1:13:17

would put in the note. And

1:13:19

then you go and you're like,

1:13:21

oh yeah, Lucy, right. Okay, hey

1:13:23

Lucy, how are you? And then

1:13:25

you can, you don't feel any.

1:13:27

Right. But if somebody ever uncovered

1:13:29

all of them, it would be

1:13:31

like an episode of curb your

1:13:33

enthusiasm where everybody's gonna be mad

1:13:35

at me because there's something in

1:13:37

there that's, I mean, let's face

1:13:39

it. I don't know

1:13:41

how we got, well I do

1:13:43

know how we got here, but

1:13:45

yes. But you know what, but

1:13:47

it's funny to circle it back,

1:13:49

we're not getting that sort of

1:13:51

companionship from our AI friends yet,

1:13:53

right? Like you can't keep a

1:13:55

secret with chat GPT like that,

1:13:57

like just between me and you,

1:13:59

remember that's the dad who walks

1:14:01

with a limp or something. Right,

1:14:03

right. Or looks really old, you

1:14:05

know, like somebody has a really

1:14:07

old-looking dad or something like that.

1:14:09

You know, these unpleasant things, but

1:14:11

that's the thing that I remember.

1:14:13

I can't ask Syria about that,

1:14:16

but can't ask ChatGPT. And in

1:14:18

the future, I'd like to have

1:14:20

a trusted, you know, where every

1:14:22

bit of... And Apple's the one

1:14:24

who's sort of going towards that.

1:14:26

They haven't said anything about... a

1:14:28

future version of Apple Intelligence having

1:14:30

contextual access to your notes? I

1:14:32

know they're talking about email like

1:14:34

their email calendar. Yeah, the the

1:14:36

halo demo that does not exist

1:14:38

in any shipping form in beta

1:14:40

yet is that the pick up

1:14:42

the mom airport. Yeah, when's my

1:14:44

mom's flight? When's my mom flying

1:14:46

into town? Yes. And it involves

1:14:48

Siri knowing the emails that her

1:14:50

mom had sent and it involves

1:14:52

Syria having access to the calendar

1:14:54

where maybe you put the flight

1:14:56

information and stuff like that. None

1:14:59

of that's there. But if Syria

1:15:01

had Apple Intelligence, I guess, had

1:15:03

access to my notes, it would

1:15:05

be very helpful to me, but

1:15:07

I really need to trust it.

1:15:09

I mean, because I don't think

1:15:11

my notes, I sure there's a

1:15:13

lot of people out there with

1:15:15

a lot worse stuff in their

1:15:17

notes than mine, but it's private.

1:15:19

Very private. Well, yeah, we have

1:15:21

a great story about just all

1:15:23

the weird things people are doing

1:15:25

in doing in notes. I have

1:15:27

not seen that, but I made

1:15:29

a note of it here, and

1:15:31

I will put it in the

1:15:33

show notes. The other thing I

1:15:35

just before I forget to bring

1:15:37

it up is one of the

1:15:39

interesting examples from your video with

1:15:42

these four chat companions is that

1:15:44

none of the four that you

1:15:46

can have these ongoing conversations with

1:15:48

can do set a timer for

1:15:50

six minutes. Right. And I don't

1:15:52

know why timers and everybody in

1:15:54

our field, me, you, everybody who

1:15:56

writes about these things, we all

1:15:58

turn to timers. as the shortcoming.

1:16:00

And famously, Siri couldn't set more

1:16:02

than one timer at a time.

1:16:04

But I know you've, I know

1:16:06

you have talked about this. That

1:16:08

was a great success. I feel

1:16:10

great, I feel great success and

1:16:12

partial credit to all of us

1:16:14

in this industry who forced Apple

1:16:16

to create multiple timers. There's like

1:16:18

some engineer who was forced. He's

1:16:20

like, yeah, I worked on that.

1:16:22

If you were the engineer to

1:16:25

have worked on that, me and

1:16:27

John and so many others out

1:16:29

there. So proud of your work.

1:16:31

I might have probably not with

1:16:33

you, but at some point I

1:16:35

think this has come up on

1:16:37

this podcast before, but we have

1:16:39

home pods and an Alexa in

1:16:41

our kitchen. And you've met my

1:16:43

wife. She is not into our

1:16:45

world at all. She doesn't read

1:16:47

my site every day. She's not

1:16:49

a tech enthusiast. And me getting

1:16:51

permission to have two different talking

1:16:53

devices in our kitchen is really...

1:16:55

out of character for her, but

1:16:57

it's because Alexa for so long

1:16:59

was the only one that could

1:17:01

set multiple timers and Amy's the

1:17:03

one who does cooking where there

1:17:05

might be two different things going.

1:17:08

In a kitchen it is not,

1:17:10

it is actually more unusual if

1:17:12

you only need one for a

1:17:14

even mildly complex meal that you

1:17:16

only need one timer at a

1:17:18

time. And so Alexa, effectively she's

1:17:20

there just for multiple timers. But

1:17:22

at some point they had added

1:17:24

multiple timers to home pod and

1:17:26

that was Apple's answer which was

1:17:28

like every time I was going

1:17:30

check just to make sure it

1:17:32

doesn't work would go check with

1:17:34

the PR team. It would say

1:17:36

well it does work on the

1:17:38

home pod. I mean I practice

1:17:40

timer parenting. I don't know if

1:17:42

I'm going check with the PR

1:17:44

team. It would say well it

1:17:46

does work on the home pod.

1:17:48

Right. I mean I practice timer

1:17:50

till we are going to leave

1:17:53

here. five minutes that you can

1:17:55

sit on the potty, you know,

1:17:57

all the things. And so I

1:17:59

need multiple, and I have two

1:18:01

kids and I've got a cooking,

1:18:03

I've got multiple. Timers is essential

1:18:05

to my life. Yeah, reading time

1:18:07

was mandated. Read a book for

1:18:09

30 minutes. We definitely use that

1:18:11

a lot. I mean, Jonas is

1:18:13

20 now, Jesus. And you're still

1:18:15

like, read a book for 20

1:18:17

minutes. Yeah, I can't make them

1:18:19

do it anymore. Right. Right. Yeah,

1:18:21

I don't think I can make

1:18:23

my seven-year-old read a book for

1:18:25

20 minutes still. But it was

1:18:27

interesting that these other ones, which

1:18:29

are so much more advanced conversation.

1:18:31

can't do the device timers and

1:18:33

dumb dumb old Siri is pretty

1:18:36

good at it at this point.

1:18:38

Yeah and but they can't do

1:18:40

all those fundamental device control the

1:18:42

things that we we I think

1:18:44

we would call now more voice

1:18:46

assistant than AI companion or whatever

1:18:48

these are the assistant tasks set

1:18:50

a timer set an alarm play

1:18:52

the music at a reminder there

1:18:54

are so many of those things

1:18:56

that are essential to what We

1:18:58

do with Syria or Alexa. And

1:19:00

that's where that conversation with Craig

1:19:02

really went, which was that we

1:19:04

need to be able to get

1:19:06

both things right. And I assume

1:19:08

it's the same thing that's happening

1:19:10

at Amazon with Alexa. Especially with

1:19:12

Alexa. Like, when did my package

1:19:14

arrive? What I order this thing

1:19:16

off Amazon? All of those things

1:19:19

we talked about at the beginning

1:19:21

of the conversation where we've trained

1:19:23

our language, we know the vernacular

1:19:25

and the words to say to

1:19:27

get the assistant to do the

1:19:29

thing. Right, and in some ways

1:19:31

it really is the same thing

1:19:33

that has made people like us

1:19:35

able to make sense of the

1:19:37

terminal interface, where you have to

1:19:39

enter the commands in this order,

1:19:41

or it's not going to work,

1:19:43

or if you screw up the

1:19:45

RM command, it's going to permanently

1:19:47

delete files that you didn't want

1:19:49

to delete it. And it's... Oh

1:19:51

yeah, of course, because that was

1:19:53

my fault. because I entered the

1:19:55

command wrong. And we're like, ah,

1:19:57

curse me, you get mad and

1:19:59

you curse your luck, but you'd

1:20:02

think, well, it was my fault,

1:20:04

I put the commandant wrong. And

1:20:06

we... where our minds naturally go

1:20:08

that way and most other people's

1:20:10

do not. And they shouldn't. They

1:20:12

shouldn't be expected to learn the

1:20:14

magic way. And even to learn

1:20:16

something as simple as it makes

1:20:18

total sense to me that Siri

1:20:20

and Alexa are now very good

1:20:22

at setting timers on devices. And

1:20:24

these other ones aren't because these

1:20:26

chat-GPT and copilot-style ones aren't. don't

1:20:28

have the device context. They don't

1:20:30

run. They're just apps on your

1:20:32

phone, right? So it kind of

1:20:34

makes total sense to me. But

1:20:36

to a normal person, it's like,

1:20:38

I'm just talking to these things.

1:20:40

Set a timer. If you're supposed

1:20:42

to be so smart, I could

1:20:45

hire the world's worst human assistant,

1:20:47

somebody who is so inept and

1:20:49

bad at their job that I'm

1:20:51

going to have to have an

1:20:53

uncomfortable conversation and fire them, but

1:20:55

they could set a timer for

1:20:57

five minutes if I told them

1:20:59

to. Like when you look at

1:21:01

what Google did, they seem to

1:21:03

take that as a moment to

1:21:05

say, okay, we'll still have like

1:21:07

this added on functionality for that.

1:21:09

We need to have that functionality.

1:21:11

And I believe now I have

1:21:13

to look on the pixel and

1:21:15

I have to set it up

1:21:17

and I'm sure there are going

1:21:19

to be some Android listeners that

1:21:21

will say, yeah, that's just an

1:21:23

extension now, you just have to

1:21:25

plug it in, but that's still

1:21:27

complicated. It's not what I would

1:21:30

see like Apple doing like Apple

1:21:32

doing or even Amazon doing, right.

1:21:34

Yeah. Or the most used, most

1:21:36

requested things of Syria and Alexa

1:21:38

are. And they're going to make

1:21:40

sure that those things do not

1:21:42

break. Yeah, I think so, right.

1:21:44

And it's just, they're approaching it

1:21:46

from very different perspective than these

1:21:48

other, the other Chadmakers. Here, I'm

1:21:50

gonna hit the money belt one

1:21:52

last time. Thing. What do you

1:21:54

think of this gimmick? Let the

1:21:56

people know. I mean, it does

1:21:58

break up the show. Yeah, I'm

1:22:00

gonna I'm gonna stick with it.

1:22:02

Anyway, I want to use the

1:22:04

bell. Yeah, I used it last

1:22:06

week with Merlin, but Merlin's the

1:22:08

one who had the bell I

1:22:10

said I wish I had a

1:22:13

bell, but I didn't know. It

1:22:15

turned out he had a bell

1:22:17

on his desk and he hit

1:22:19

it for me. And then I

1:22:21

went on Amazon and bought myself

1:22:23

a bell. Anyway, that was the

1:22:25

money bell and I got to

1:22:27

thank our third and final sponsor

1:22:29

of this episode. It is our

1:22:31

good friends at Memberful. Memberful is

1:22:33

Best in Class membership software used

1:22:35

by many of the web's biggest

1:22:37

independent creators, publishers. and even bigger

1:22:39

media companies. It's really really good.

1:22:41

It lets you offer membership perks

1:22:43

and exclusive content to your loyal

1:22:45

followers and fans, readers, listeners, watchers,

1:22:47

whatever it is that you're creating,

1:22:49

giving you full control over who

1:22:51

has access to your blog posts,

1:22:53

newsletters, for email newsletters, online courses,

1:22:56

podcasts, private community chats, and more.

1:22:58

and with membership memberful you maintain

1:23:00

full control over your brand your

1:23:02

audience and your business if you

1:23:04

ever want to leave you can

1:23:06

just export the full membership list

1:23:08

take it with you build your

1:23:10

own system take it somewhere else

1:23:12

It's a sign of confidence that

1:23:14

you won't want to leave Memberful,

1:23:16

that they make it so easy

1:23:18

to export and leave if you

1:23:20

ever wanted to. Their brand doesn't

1:23:22

get put in front of yours.

1:23:24

It's not, for example, sub stack,

1:23:26

where everybody who's on sub stack,

1:23:28

it looks like a sub stack

1:23:30

and says sub stack. Memberfuls behind

1:23:32

the scenes, it's quiet. may not

1:23:34

even know you're using Memberful. It's

1:23:36

that behind the scenes, which I

1:23:39

think is super, super creator friendly.

1:23:41

Full customization, you can set up

1:23:43

and manage your membership program with

1:23:45

Memberful's intuitive platform, create multiple membership

1:23:47

tiers if you want to, and

1:23:49

payment options that cater to different

1:23:51

audience segments, seamless integration, they integrate

1:23:53

with all the tools you already

1:23:55

use, like WordPress, Mailchimp, Google Analytics,

1:23:57

Discord, Discord, a big one, where

1:23:59

you can gate access to your

1:24:01

community. Discord for members only through

1:24:03

memberful. They already have an integration

1:24:05

with that to make it as

1:24:07

easy as possible. It's just a

1:24:09

great way and it is clearly

1:24:11

the trend for independent media going

1:24:13

forward. And they have great customer

1:24:15

support too where you can contact

1:24:17

somebody for help for your particular

1:24:19

case and they can make recommendations

1:24:22

because they know from their experience

1:24:24

what sort of things work and

1:24:26

which don't for building it and

1:24:28

making it successful. like which sort

1:24:30

of things you should make members

1:24:32

only and which ones you should

1:24:34

put out there for everybody else

1:24:36

so that they want to become

1:24:38

members of your site. They help

1:24:40

you with that. So anyway, where

1:24:42

do you go to find out

1:24:44

more? Go to memberful.com/talk show. M-E-M-E-N-B-R-F-L-S-S-

1:24:46

talk show. As we head down

1:24:48

to home stretch, have you had

1:24:50

this experience? I think everybody in

1:24:52

our racket dealing with Apple has,

1:24:54

where you encounter an issue or

1:24:56

a problem or an observation. And

1:24:58

when you're communicating with Apple PR

1:25:00

about it, they act like, ah,

1:25:02

they don't say you're crazy, they

1:25:04

don't, they don't gaslight you per

1:25:07

se, but they, they act like

1:25:09

it's like, ah, nobody's ever said

1:25:11

that before. Huh. which I guess

1:25:13

is sort of gas lady, but

1:25:15

for me I ran into this

1:25:17

testing the iPhone 16s in September

1:25:19

and they they didn't say you

1:25:21

should install the 18.1 beta but

1:25:23

they didn't discourage it and for

1:25:25

reviewers who don't know how to

1:25:27

do it they gave instructions. And

1:25:29

so I used them for a

1:25:31

couple days with 18.0, which was

1:25:33

what they were going to ship

1:25:35

within the box, which I thought

1:25:37

was important to note. If you

1:25:39

buy it when you first unwrap

1:25:41

it, here's what you get. And

1:25:43

then I put the 18.1 beta

1:25:45

on. And that's when I first

1:25:47

got the new series. And new

1:25:50

series voice is definitely more. human.

1:25:52

And I still use, I forget,

1:25:54

they don't name their voices cutie

1:25:56

style anymore, they just call them

1:25:58

like voice one, voice two. I

1:26:00

think I'm a voice three person,

1:26:02

which is sort of the traditional

1:26:04

like original seri female American voice.

1:26:06

Let's see, oh, nope, it's a

1:26:08

voice four. American voice four is

1:26:10

my seri. And it definitely sounds

1:26:12

more realistic. But also. is talking

1:26:14

to me way too slow. It's

1:26:16

so slow. I have an American

1:26:18

voice too. I've been playing around

1:26:20

with them though, so. I'm going

1:26:22

to start playing around. The colors

1:26:24

of the sky fade with the

1:26:26

setting sun as the stars begin

1:26:28

to shine through the clear night.

1:26:30

The colors of the sky fade.

1:26:33

The colors of the sky fade

1:26:35

with the setting sun. You're so

1:26:37

sorry, you were saying yours is

1:26:39

slower and now I'm trying to

1:26:41

download this and sorry. Yeah, yeah.

1:26:43

I don't hear the slowness in

1:26:45

those canned examples when you're trying

1:26:47

the voices, but just day to

1:26:49

day talking to it and getting

1:26:51

verbal answers back. It seems like

1:26:53

New Syria simultaneously does sound more

1:26:55

human in inflection, which has been

1:26:57

like a slow boiling frog as

1:26:59

Federiki told you in your interview.

1:27:01

over 15 years now, it's gotten

1:27:03

more and more realistic incrementally every

1:27:05

couple years. But I find that

1:27:07

it talks way too slow. And

1:27:09

they were like, ah, nobody said

1:27:11

that. And maybe I'm the only

1:27:13

one to notice. I don't know.

1:27:16

But then if you go into

1:27:18

set it on your phone, you

1:27:20

can go into settings accessibility Siri,

1:27:22

and there's a talking speed. And

1:27:24

I've turned it up to like

1:27:26

110 percent, which to me still

1:27:28

sounds a little slow. But the

1:27:30

other thing, and Amy agrees with

1:27:32

me, because one place I hear

1:27:34

the, I hear the Siri voice

1:27:36

all the time is when we

1:27:38