0:00

It's time for to it this

0:02

weekend Tech: It's the return of

0:04

Stacey Higginbotham for one week only

0:06

special appearance. Will also talk to

0:08

Ben Par are a I expert

0:10

and my good friend Alan Melvin

0:13

turn a former host of this

0:15

weekend computer hardware. He used to

0:17

be under the sea now he's

0:19

on Solid State has an Ssd

0:21

expert at Pfizer and will talk

0:23

about of course saw as they

0:25

technology, new memory technologies, ai, ai

0:27

vandalism. I say it should be

0:29

illegal. And a whole lot

0:32

more. It's all coming up next.

0:34

Yes, the Vision Pro on twitter

0:36

room. I guess you love from

0:38

people you trust. This

0:41

is tweet. This

0:49

is to it this weekend.

0:51

Tech Episode Nine Hundred Sixty

0:54

Four Recorded Sunday, January Twenty

0:56

Eight Twenty Twenty Four. No

0:58

one talks to the faucet

1:00

anymore. There's

1:02

we get. Deck is brought to

1:04

you by Gusto. Running a small

1:06

business is just plain hard. Gusto.

1:08

Let's you focus on the joy

1:10

of running your business with. It's

1:12

easy to use payroll software accessible

1:15

online from anywhere Gusto House where

1:17

the three hundred thousand businesses and

1:19

ninety percent of it's customers. A

1:21

switch into Gusto was easy. You

1:23

get unlimited payroll for one monthly

1:25

price, no hidden fees, get multiple

1:27

schedules and rates, direct deposit and

1:29

checks you can print yourself. Plus.

1:32

Gusto integrates with your favorite tools

1:34

to make life easier to like

1:36

Quickbooks, Zero, Google, and more file

1:38

and pay all federal state and

1:40

local payroll taxes in all fifty

1:42

states, you know, Three at a

1:44

forecasters say. Running pay roll with

1:47

Gusto takes ten minutes or. Less.

1:50

Gusto. Cares about the small business owners

1:52

they work Within says money can be

1:54

tight right now. you'll get three months

1:56

free when you run your first. A

1:59

rural gotta gusto.coms. Tech and start

2:01

setting up your business today twit listeners. You'll

2:03

get three months free once you run your

2:05

first payroll gusto g

2:07

us to comm Tech

2:17

It's time for twit this week in tech show we cover

2:19

the week's tech news I this is

2:21

going to be like old home week. This

2:23

is fun for me Say

2:25

hello to Stacy Higginbotham. We miss her so much

2:27

on this week in Google She is a policy

2:30

fellow at Consumer Reports still writes

2:32

a little bit does a lot of

2:34

things But it's so nice to have you back

2:36

in front of our microphones. Hi Stacy Hello,

2:39

it's good to be here everything going well

2:42

there and a beautiful, Washington State It

2:45

is it's awesome this warm winter. I know

2:48

it's like the end of the world, but I

2:50

love it anyway Yeah, as Andrew got you playing

2:52

pickleball yet, or Silence spoke a few words

3:02

Lisa and I have been playing the old

3:04

pickleball. It's kind of fun. Oh See

3:07

I thought you'd want to know how many connected

3:09

devices. I've taken out of my house Have you

3:11

have you been able to remove them now Stacy

3:13

of course had a Stacy on

3:15

iot.com and this iot podcast and all that

3:18

stuff and just kind of let that all

3:20

go and As a result you had

3:22

to have all this iot crap you had to go to CES

3:24

He had to do all sorts of stuff. You don't have to

3:26

do anymore That's doing to

3:28

see yes. Oh Well there you go

3:30

a good. I'll ask you about it also

3:32

here Ben par a good friend Longtime

3:35

friend of the show author of the

3:37

AI analyst he is eight

3:39

years ago founded an AI startup

3:42

octane AI right

3:44

in the nick of time and Writes

3:46

about AI for the information hi Ben Hello,

3:49

why do you have a little reddit

3:51

guy behind you is that just oh

3:53

that is actually the mascot for My

3:57

company I carry eyes his name's Octi actually

3:59

hold on I've got the plushy version.

4:01

Do you want a plushy version? I will. Oh, I would

4:03

love that He does look a little

4:05

bit like the reddit guy, but oh, yeah,

4:08

and so his name's Octi. He was invented in

4:10

2016 He I guess around

4:12

it's all floating very popular with kids

4:14

all floating robot. You robot Yeah,

4:17

they all look alike. They all look pretty much

4:19

the same right if they don't look like it

4:21

looks like Eva from WALL-E. That's right Eva

4:26

They don't yeah, you don't want them to look like you

4:29

know one of those dogs from Boston Dynamics, so

4:31

I guess it's fair That's

4:33

fair Hello, Ben. Welcome.

4:35

Good to see you again And

4:37

who else is here formerly host of

4:39

this week in computer hardware? SSD

4:42

expert our SSD expert forever. He

4:45

is now SSD technologist at Phison

4:48

what it what I don't

4:50

know the name phison should I Alan Melvintano? Good

4:53

because their controller technology is in an

4:55

awful lot of SSDs. Oh so

4:58

they're a infrastructure behind

5:01

the scenes kind of Right. I've

5:03

been I've been working my way

5:05

further and further back in the industry And

5:09

you see behind him the remnants of all the

5:11

things remnants of all

5:13

the things yes You

5:16

know I find problems at one company, and then

5:18

it's like oh darn We have to work around

5:21

the problems that come from the controller, okay? Well,

5:23

let me just go work for a controller company

5:25

then I'll just fix other problems there Before

5:28

they even get to the SSD there was a I

5:30

actually saw a story this weekend I thought you know

5:32

I should ask Alan about it. It's not exactly

5:36

SSDs, but it's memory technology Samsung

5:39

has introduced this new LP cam Memory

5:43

technology that is in theory gonna

5:46

replace something that's already been replaced so dims

5:50

But they say it's faster than LPDDR5. Yeah Yeah,

5:55

it's a wider bus. It's just You

5:57

know a lot of dims technologies

6:00

in general there's a work around where you have

6:02

to you know that the bits have to go

6:04

across a fixed number of lanes right and if

6:06

you can make that that bus

6:09

more efficient wider any

6:12

of those things you know especially

6:14

on mobile leads to better

6:17

performance and battery life. Of course a number of

6:19

companies including Apple have started putting the RAM on

6:21

the package which I

6:23

think eliminates that bus you

6:26

have a unified memory eliminates that bus overhead

6:29

right? It

6:31

does well the bus is just way faster

6:33

yeah you're right in that case.

6:35

Right but there are other cons

6:37

to go with that sometimes if you're stacking directly

6:39

on top now it's harder to get heat out

6:41

of the heat producing thing underneath the memory. So

6:43

does CAM have a future this compressed attached

6:49

memory module? We'll have

6:51

to see this is one of those where it's just been announced

6:53

and it's not you know it's not it's not

6:55

one of those things where hey this is

6:57

out now and it's all of a sudden ubiquitous

6:59

everybody's making a thing with it so yes at

7:01

the standard is the thing people could use we

7:05

have to wait and see. Okay let's

7:07

see I see these things and I think boy I wish

7:10

Alan was here so now that you were here I asked

7:12

you. Can I ask Alan later about risk

7:16

five adoption and yeah sure let's ask

7:18

him now risk five adoption. Stacy when

7:21

we were on Twitter we talked a

7:23

lot about this open source architecture to

7:25

replace x86 or

7:27

I guess even arm risk letter

7:30

V for five but

7:33

I'm more Intel's it when you were an Intel they

7:35

were doing it right. Well

7:37

I don't I didn't get that much in the

7:39

weeds when I was at Intel to even be

7:41

able to speak on it right. I was more

7:44

worried with evaluating the performance of the platform and

7:46

the SSDs versus comp I wasn't I didn't usually

7:48

have to go so far back in the chain

7:50

where I was talking ASICs. But you can say

7:53

you must still have an opinion on risk.

7:55

Well it is like Western Digital was one

7:57

of the big adopters of it for Like

8:00

early adopters. Oh, yeah. Anyway, we don't have

8:02

to talk about this this week in risk

8:05

adopt this week in Increase

8:09

so Dynamically

8:11

because people they're talking about risk

8:14

five everybody come on over No,

8:17

I'm probably not Let's say listen as far

8:19

as SSD things go anything that can move

8:21

the bits faster and more power Officially and

8:23

as good and accomplish all of the tasks

8:26

that the SSD controller needs to do is

8:28

a viable solution for that Right. It just

8:30

depends on who does it then

8:32

if they choose to or they know and what's

8:34

the cost benefit for it all right, I know

8:37

what everybody wants us to talk about or Conversely

8:40

doesn't want us to talk about which is

8:42

that Friday February 2nd? People

8:45

will start getting the vision pro

8:47

headsets from Apple and you're

8:49

gonna start seeing YouTube actually sooner than that

8:51

Probably because because I'm sure the people who

8:53

get loners from Apple will be

8:55

off embargo probably Wednesday So you'll start seeing

8:58

Wednesday the reviews and you'll see

9:00

the thumbnails on on YouTube

9:02

going. Yeah or Ben

9:06

you ordered one you must have ordered one. I

9:09

did order one I got up at was

9:12

it like one two of the board. Oh,

9:14

make sure to order one I'm a em

9:16

Pacific a time and whatever time it was

9:18

it felt like one or two Did you

9:20

know here's the critical question? Did you order

9:23

it to go pick it up in the

9:25

store so you can get that 25 minute

9:27

pitch on how great this thing is? Or

9:29

did you like Alan like Alex Lindsay and

9:32

like Jason Snell say no, yeah, just send it to

9:34

me. I'll figure it out. I Did

9:37

the store version? Oh good. It's gonna be a

9:39

hair faster. And you know what? I got a

9:41

report I gotta like do a video or something

9:43

put it on tick tock. Yeah. Yeah, Mike is

9:45

going in there I wonder if they'll let us

9:47

you know do do a little video of him,

9:49

you know getting the tour I want to see

9:51

Ben pars spooky eyes On

9:53

the front of it. That's gonna be really next time I

9:55

go into it. I wore nothing, but was it please? Oh

9:58

Do the entire time just see if I can.

10:01

Did you send them your glasses prescription? Are you gonna

10:03

get those ice lenses? I don't

10:05

so I can wait I can see just

10:07

fine without the glasses they're just like I

10:09

like to make you look smarter is that it? I literally

10:12

just got my eye exam like an hour

10:14

and a half ago funny enough and

10:17

like it's it's enough it's like where

10:19

it's nice for me to drive I

10:21

could theoretically drive without. Okay. The world

10:23

seems a little clearer but close-up stuff

10:25

like it's actually better for me not

10:27

to wear glasses don't need it so.

10:30

Well you have one advantage you can lend your

10:32

vision pro to other people. It

10:35

won't be so. My fiance has got to go and play

10:37

around and do stuff you know we got to watch I

10:40

don't know. She's a playwright so I imagine

10:42

she'd be interested in in how

10:45

vision pro might have impact theater. I

10:49

wonder if you could I have to see if like they

10:51

have like easy switching accounts because then you could like have

10:53

your own works. They have a guest mode I think there

10:55

is a guest mode yeah. Because

10:57

I would be good. Is

10:59

your fiance interested at all? I have literally

11:02

asked all the normal people that I talked

11:04

to and they are just like what

11:06

the hell I have zero interest in

11:08

this and all of my efforts to

11:11

drum up any sort of excitement so

11:13

I'm just curious like here do

11:15

the real people. If it had risk five in it

11:18

maybe. Just call it the swim

11:20

five. Here's

11:23

what I think Stacy I think that no

11:26

one outside of developers has real interest and

11:28

the goal for Apple this year is not

11:30

to get mass adoption. It's like get it

11:32

in the hands of developers who will get

11:34

excited and build some stuff for it and

11:36

then they'll go more mass market and cheaper

11:38

afterwards but in the interim you know if

11:40

your friend has it you're gonna be like

11:42

sure I'll try it and most likely be

11:44

like oh my god this is awesome maybe

11:46

and then they hear oh it's a thousand

11:48

dollars cheaper next year. Oh I think

11:51

I will go and get it. That's my guess

11:53

with how Apple is rolling this out. It needs

11:55

a couple years to get more developers on board but

11:57

if they drop the price which they'll be able to.

11:59

to do in a few years, they

12:01

can get more adoption. In

12:04

line with what Ben was just saying

12:06

there, the thing that I think is missing

12:08

from it so far, that's not to say that it's

12:11

not great in the experience, appears to be

12:13

pretty impressive, but it's missing the

12:15

killer app, right? It

12:17

doesn't have... Not just the killer app, but

12:19

the killer app category. What

12:22

exactly do you need this for? Right.

12:26

Yeah. And so I think in order,

12:28

they're in this chicken and the egg proposition

12:31

right now, where they need to get the hardware into

12:33

way more hands, so that there's enough

12:36

developers out there to try to do something

12:38

cool with it that really sets

12:40

it off. And then don't be surprised if

12:42

somebody does something really impressive with the technology,

12:44

and then Apple just immediately requires that thing

12:46

and then spends on it for a year

12:48

and makes it way better,

12:51

and then ships it along

12:54

with whatever the next generation. Any guesses as

12:56

to at least the area? Will

12:58

it be a game? Will it be watching videos?

13:00

Will it be that spatial video that

13:03

they're pushing where you could take a 3D

13:05

video with your

13:07

iPhone and then it's like you're there in

13:10

the Vision Pro? Will it be a... They're

13:12

also talking productivity, which is bizarre to

13:15

me. I can't imagine trying

13:17

to do an Excel spreadsheet on that thing. What

13:20

will it be? What category? I

13:23

think a lot of

13:25

those things you just mentioned were already shown

13:27

and it remains to be seen as

13:31

how it plays out whether or not those would be

13:33

deemed a killer app. You can already watch a movie.

13:35

That's already a capability of it. It's not something they

13:37

added later. Although both YouTube

13:40

and Netflix have decided not to

13:42

put out a Vision Pro app,

13:44

probably in a fit of peak

13:48

about the app store more than anything

13:50

else. It makes sense from their perspective

13:52

though. It's going to

13:54

be so few people in the beginning. They can

13:56

always launch one later, a year from now, two

13:58

years from now. and they

14:00

have a bunch of other stuff. It's what I

14:03

get the decision to go and do that. It's

14:06

probably a game if I had to give the first version of

14:08

something like people are always surprised by

14:10

how much the like metas you know quest

14:13

sells they sell a lot they're very popular

14:15

at Christmas time and games are

14:18

probably still the biggest thing with workouts

14:20

and a couple other things still

14:22

no idea what the like killer app is going

14:24

to be that's just going to take some time.

14:28

Okay so I think there's niche

14:30

gaming opportunities as a killer app. I think

14:32

working out that's a popular idea for people.

14:34

I don't know why because strapping something that

14:36

weighs that much to my face and sweating

14:38

in it is like the least the

14:42

least exciting option for me. I played

14:44

lightsaber on my oculus pro my request

14:46

pro for about half an hour and

14:48

yeah it's you feel pretty grody

14:50

at the end of the

14:52

day. I look at it and I think we're

14:55

going to it feels like

14:57

a flamingo in an egg kind of situation or

14:59

some bird that nobody really wants to eat. A

15:02

dodo in an egg perhaps? No I why

15:05

did we kill the dodos did we eat them I don't know. This

15:08

is the whole history there. Yeah no I am

15:11

like I don't actually remember why the dodos died

15:13

but um so I look at this and I'm

15:15

like well so how about

15:17

this? I feel like society will have to get really

15:19

terrible. Passenger passenger and the no no because they

15:21

had value like we wanted

15:24

passenger pigeons. I'm not sure we

15:26

want this yeah we thought passenger

15:28

pigeons were yummy yeah um one

15:30

or but I feel like the

15:34

the opportunity that this offers so

15:37

far and will probably

15:39

offer for a couple more years is

15:41

a true escape from the world we

15:43

live in and I don't know if the

15:46

world is that bad for that many people

15:49

just yet. I think it could

15:51

get there. It's interesting to say

15:53

that's what would drive adoption because

15:55

all of the VR sci-fi stories

15:58

those VR helmets whether it's neuromuscular, answer

16:00

or Ready Player One are used as

16:02

an escape from a dystopian future. Like

16:05

the world's terrible and

16:08

you want to let in the metaverse. And

16:11

if you're like a Gen Z kid who's like

16:13

stuck in an apartment with like eight people

16:15

and you work all that, like that feels

16:17

like a couple years from now that could be

16:19

good for you. Like, but it's still

16:21

super expensive and I don't see that changing.

16:23

Pretty dystopian view though. And

16:26

I agree with you because one of the things I

16:28

don't like about it is it isolates you even if

16:30

you're watching a movie, you're watching it all by yourself.

16:32

We have a pretty good experience with

16:34

a nice TV and a good surround

16:37

sound and a couch where we can,

16:39

you know, communally enjoy something. But

16:41

that's not the VR experience. It's a

16:43

very solitary thing. But you make

16:45

a good point. If you have eight roommates, maybe that's

16:48

not such a bad thing to be solitary. Or you're

16:50

on an airplane. I don't, do you

16:52

think you're going to see, are you going to get on a plane next

16:54

week and there'll be people wearing Vision Pros

16:56

up and down the aisles? I mean, someone's going to do

16:58

it for the fun, for the Seattle, for San Francisco. Sure.

17:02

Yeah. Okay. Look,

17:04

I'll do it for the novelty factor. So someone asked me a question

17:07

and also just to report on it. Not, but am I going to

17:09

bring a $3,500 computer device

17:11

on trips with me? No, not

17:13

normally. But I

17:15

did have an idea through this conversation, uh,

17:18

Vision Pro for toddlers. You put on the

17:20

thing, you don't have to worry about your kids for six

17:22

hours. Yeah, you don't have to think about it. That's

17:26

like the high tech pillowcase. You just,

17:29

you put them in it and that's it. Now

17:31

you, you, uh, and I hope you can translate

17:33

this because you put the Ming Chi Kuo article

17:36

in on the Vision Pros first

17:38

weekend preorder, but it's in Chinese.

17:41

So he said what,

17:43

180,000? I

17:45

can't keep scrolling down. They have the English. Oh,

17:47

there's the English. Okay. Based,

17:50

thank you. Based on preorder

17:52

inventory and shipping time Ming Chi

17:54

Kuo, who's usually fairly accurate, estimates

17:57

that Apple sold 160 to 180,000. Vision

18:00

Pros during the first pre-order

18:02

weekend. The date did slip out to

18:04

a month. I don't

18:07

know where it is today. He

18:09

says it's about five to seven weeks within hours.

18:13

They say Apple says I think that they can make

18:15

as many as half a million a year because

18:17

the limitation is actually those

18:19

Sony screens inside it. And

18:22

Sony has said we can't make that many. So

18:24

half a million a year. It's not intended to be a

18:26

big seller though. This is Apple releasing

18:29

something for a very niche

18:31

market they hope mostly developers

18:33

and then using some very

18:36

high end Apple fans

18:38

as beta testers basically for

18:40

a whole new concept of

18:42

computing. That's what I see. And

18:45

Jason's now explained this on MacBreak Weekly. He said

18:47

if you're Apple and

18:50

you know you need the next best big thing right after

18:52

the iPhone because the iPhone's not going to last forever

18:54

so you know that you've got to find something. One

18:57

of the things you're going to try is

18:59

this kind of maybe AR glasses idea but

19:01

you don't have the technologies yet to release

19:03

that. So in order to

19:05

be ready when it comes ten years from now

19:07

maybe you've got to start trying stuff out now

19:10

knowing it's a small market and it's an expensive

19:12

market to be in and it's going to be

19:14

a money loser for years to come. But

19:17

Apple can afford it and this is what Apple needs to do. You

19:19

agree? And

19:22

that's what Google thought when they released the

19:25

Google Glass way back in the day. We've

19:27

all seen... And Microsoft HoloLens. Yeah,

19:29

we've all seen this before and I wonder if

19:31

2024 given all the craziness around the election,

19:36

given interest rates, given all of this, if...

19:40

and Apple has tons of money. Will

19:42

investors let them play this out to

19:44

the extent they need to? And that's

19:46

a really open question. They

19:49

get a lot of leeway. It is an open question

19:51

but they're going to also let them do a car

19:53

in some amount of time. I

19:57

feel like there is actually a lot of interest

19:59

in smart cloud. If is like as a

20:01

matter, Ray Bans are actually more popular people

20:03

think people do like those. My Gilligan was

20:05

on a couple of months ago and Suisse

20:07

were in Miami on what Yummy this close

20:10

to mine and those are for the average

20:12

sunglasses you have. Good speakers in the Temple

20:14

Peace. And a decent

20:16

battery life. I think he's had for five

20:18

hours right? And it's got a camera built

20:20

in. they just as an Ai feature where

20:22

you could take a picture something and the

20:24

glasses will then tell you what is his

20:26

or do some sort of in google and

20:28

style a search on it. At.

20:31

I want them all those third printed. Matter.

20:34

Ray. Bans make a wider version as even your

20:37

white version says the hair too tight on

20:39

my face or you're like me. you get

20:41

a fair haired I get up at as

20:43

I had yeah I but the idea I

20:45

think space is made to make the act

20:47

exact right click with like you know physically

20:49

ten years down the road. Maybe this knowledge

20:51

he is good enough to get into like

20:53

a Glass as format which I think people

20:55

are interested in a people who are honored

20:57

his skill. but there's a lot holding it

20:59

back. Stacey there's battery life right? That's

21:02

a want to compute power if you really do have been

21:04

more than these medical. Us as do. Is

21:08

there's tons? I mean there's tons of

21:10

hardware development there's I mean there's quality

21:12

internet connectivity that Billie hasn't and sell

21:14

for some the glad know like we

21:16

think it has that. He. Said

21:18

her of a and like there's a light that needs to

21:21

be solved here and. I get my companies are

21:23

like guess we must invest in things computing.

21:25

Because. That's the nest obvious option.

21:28

My the roads I am calling

21:30

your face computers themselves. Apple is

21:32

really a great Have gone to

21:34

great pains to make such as

21:36

spatial computing Know his face computing

21:38

thank you states. So computing

21:40

to me like we want to get

21:42

there. that's like that smart does that's

21:45

that's that's the size fi. You know

21:47

I'm soon enough us a screen and you

21:49

can do that with a our classes but.

21:52

It's face computing. Face it. it's

21:54

face and apple. Or

21:58

a boat or unite. the The

22:00

real question is, for instance, 5G, we

22:03

have the technology to put ubiquitous

22:05

high-speed internet pretty

22:07

much everywhere. Maybe

22:10

it'll have to be improved a little bit, maybe a Wi-Fi 7

22:12

is going to help. I'm sure that's part of the point of

22:14

Wi-Fi 7. You

22:18

can see them in the future, but battery

22:20

technology has not been leaping forward.

22:22

Right now, you have to wear a separate battery

22:24

pack with a wire to use

22:26

the Vision Pro for any length of time. Even then,

22:28

it doesn't go for very long. Well,

22:31

both 5G and Wi-Fi are hugely

22:33

battery intensive. Yes. Right.

22:37

So, yeah. I mean, look. So,

22:39

there's technology that I think that they don't even exist right

22:41

now. It's

22:44

10, yeah, like the people working on

22:47

the battery stuff, but decade

22:49

away, maybe longer? Do you

22:51

think it's a decade? Yeah. I've read

22:53

up some stuff. There's some interesting stuff being done

22:55

and interesting approaches. I

22:58

have a hard time predicting that one because

23:00

that one ends up also on the physical

23:02

limitations of our ability

23:04

to power things. I hope it's a decade away.

23:07

I think Apple probably thinks it's something like that,

23:09

which is why they would come up with something

23:11

now, the bigger thing, because eventually, the way they

23:13

like super battery, it's in the thing, but it's

23:15

still years. But you do raise a good point,

23:17

Stacy. How long will stakeholders put up with this,

23:20

especially if it's a big drain on Apple's

23:23

resources? Right now, Apple shareholders

23:25

are pretty darn happy with

23:27

Apple. They're not exactly leaving

23:29

the stock in droves. We

23:33

have a story later on about layoffs and

23:35

talking about how shareholders are thrilled about that. So,

23:37

I look at, I mean, if

23:40

you're really considering investors

23:42

and shareholders in these sorts of

23:44

conversations and calculations, they're

23:47

not long-term thinkers. Yeah, that's

23:49

right. And if they're spooked right

23:51

now because of everything,

23:54

then that's just an

23:56

open question. It's

23:58

a great point. board and

24:00

shareholders that you

24:02

know to put in what it met up put

24:04

into its its VR efforts

24:07

more than 10 billion a year

24:09

I'm which is not a success

24:11

yeah and it's not paid off although

24:14

day I yeah yeah

24:17

they pivoted AI in fact that's that's the

24:19

other question maybe apples making

24:21

a mistake going all in on

24:23

AR should they be going in

24:25

on AI harder this is such

24:27

a dumb question sorry thank you

24:29

they were wondering

24:34

how long it would take and I

24:36

think that took about 12 minutes so thank you it

24:40

doesn't bother me at all

24:42

it is a dumb question

24:45

tell me why because look

24:48

for any sort of AR experience I mean

24:50

if we if we think about the

24:53

hardware that we need to have for

24:55

shrinking the hardware down to give us

24:57

like true computing on our face or

24:59

whatever spatially they're going to have to

25:02

be new interfaces the best interfaces that

25:04

we can possibly have are going to be

25:07

if not driven by like

25:11

AI like learning today

25:13

I I'm losing my words here

25:15

but AI is just a crucial part it's like

25:17

saying I want to build a computer today without

25:19

considering the latest wireless yeah you need a

25:21

part of it yeah you need it or it's

25:24

not nearly as useful but if

25:26

you can use the glasses or the spectacles or

25:28

the nerd helmet or the whatever

25:30

you call it if you can use it to look at

25:32

at the world and get in from get you

25:35

know analysis and information back that's

25:37

much more useful than if you're just

25:39

you know playing pong so

25:42

I agree with you yeah it is a lot of I

25:44

mean look there's a lot of reports they're working on

25:46

AI for sure I Think things

25:48

internally. apples just always later. But they

25:51

do it where it's much more perfected

25:53

than others. So it's I Keep on

25:55

saying over and over again if there's

25:57

just one company where I'll know. Never

26:00

really bad against them at Apple like people are

26:02

like the watch mass the watches pretty dang popular

26:04

place they don't agree with. Op was a was

26:06

a form factor we already were wearing it was

26:09

something we're used to wearing and I admit I

26:11

was in had been idle on the was my

26:13

first is a i got a long been ah

26:15

you're right the of I'm on was not okay

26:17

my with why when the I phone came out

26:20

we were already hungry for internet on the I

26:22

was a the iso of it's the was say

26:24

it like that was actually really slow adoption care

26:26

for the first two years of the watch nobody

26:28

actually linked it was a. Sealed product.

26:32

At the Us on. Our

26:34

know that ya that like I remember

26:36

driving around trying to like search for

26:38

things on my palm pilot and as

26:41

as new had are like Nokia phone

26:43

early stages democratized at so I think

26:45

the demand and hunger was there for

26:47

that. We all had little numbers Vipers

26:49

from her blackberry Blueberries Andreas as a

26:52

blueberry there's a difference we i blackberries

26:54

by them. Sizes.

26:56

Bit of we all that also had google maps

26:58

so like that's kind of a killer app

27:00

that algiers thousand the death your right so so.

27:03

Killer app for the I phone was

27:05

google maps that came with google dazzling

27:08

Ah but then I think also safari.

27:10

The. Fact that you could use a tiny

27:13

little screen in a browser and actually

27:15

browse desktop. Web. Sites was a big

27:17

deal. him because he could tap at that

27:19

would certainly. Look. He

27:21

said and. Like. The personalization that

27:24

came with something knowing your exact

27:26

location there was once rose adoption

27:28

like remember we were so excited

27:30

or know how to bar and

27:32

and l it said soon as

27:34

I said I am kids. Mobile

27:37

So well. Is it so so Mobile?

27:39

Local Plumber our we risk. We.

27:42

Sell high on that. I'm. In

27:44

that's. That. That

27:47

was kind of the benefit. Of the

27:49

Iso and or know it were

27:51

high on for. The are.

27:53

In A I. Look,

27:55

it's. It's. I talked to like all com they're

27:58

going to tell me about Ai for like in. Modulation

28:00

for like delivering five Ci to the

28:02

phone right? That's it. That's important. great

28:04

But that's one good things about Ai

28:07

is it is is it is like

28:09

computing in general is a very wide

28:11

range of app which infrastructure. Or.

28:13

Am I would can be, but it

28:15

can also be. I use an Ai

28:17

expert system with coding and that's really

28:19

amazing when there are some definite uses.

28:22

For. Ai right now I have a box whereas

28:24

I don't see those. For

28:27

oh yeah, no, I'm arguing that Ai as

28:29

infrastructure. Which is why I think it'll be

28:31

necessary I will way I did it. ever

28:33

name as the Internet in the sense of

28:35

internet is infrastructure for our lives, eyes infrastructure

28:38

and whether you see it or you don't

28:40

ride bikes you know there's a I algorithms

28:42

handling of unjustified on a resume and then

28:44

when you look up on Tic Toc and

28:46

then when you go and drive your car

28:48

and a as a slick abroad like civil

28:50

it is as a broad term that applies

28:52

to a lot of things and people need

28:54

disagree and I'm wearing my top beyond entered

28:56

of a Ice which is it's own little

28:58

sub categories. For jazz like these apps

29:00

have the right it's infrastructure, absolutely owners

29:03

and have been around and it's been

29:05

around for longer than you might realize.

29:07

I I I came up with a

29:09

report on a I had Bastard Trends

29:11

ah it's have been part of karma

29:13

had to plug it or hundred and

29:16

ages Yes and out. One of things

29:18

I found when I was doing it

29:20

was I did a report in college.

29:22

Ah and ah A I had two

29:25

thousand five, two thousand and six. While

29:27

I found like those lights are you

29:29

know when ah where the term A

29:31

I was really first point it was

29:33

a conference into up as in the

29:36

nineteen like Safeties was like oh yeah

29:38

this has been around for a loner

29:40

John Mccarthy We were both people have

29:42

my age have been through more than

29:44

one ai winter as well. You know

29:47

we've watched the Ib overhyped. You.

29:49

Know and die of miserable death

29:51

more than once I think two

29:53

or three I winners of concerts

29:55

so. ah but i don't

29:57

think this is the same i honestly don't i

29:59

am i completely turned around on

30:01

AI. I've become much more bullish on it since

30:03

you last joined us, Stacy, because

30:06

I've been able to use it in so many interesting and powerful

30:09

ways. And I see with

30:11

one of your slides in the deck you show a slide

30:13

110, you show

30:15

an image AI generated image.

30:17

And absolutely, I mean for image

30:20

generation alone AI has been fascinating.

30:22

I've become, and I don't know

30:25

if you're on the same train

30:28

as I am, Ben, but I

30:30

have become so bullish on AI that

30:32

I'm, and much to my

30:35

the dismay of many of my co-creators,

30:38

I'm kind of saying just let

30:40

AI have everything. Don't limit AI

30:42

to uncopyrighted or free material. And actually

30:45

we're starting to see that a lot

30:47

of news organizations are blocking AI scrapers.

30:50

I think that training is really

30:52

important and I think you should, I

30:54

think AI should have access to everything. It's

30:56

not stealing your information, it's learning

30:59

something, and I think it's going to be so

31:01

valuable. I don't

31:03

want to see any limitations on it. I

31:06

see the debate screaming. Yeah, you can

31:08

see, oh yeah, you know how much

31:10

people hate that one, especially creators hate

31:13

that when I say that. Well,

31:15

with valid justification based on

31:18

some, you know, when

31:20

you see an AI image generator spit out

31:23

a thing that still has the watermark. But

31:25

it's not. It's a little bit like the

31:27

watermark, but it isn't, it's not Getty's watermark.

31:30

You can tell that they're, that it ingested

31:32

some Getty images, which is put, which by

31:34

the way, Getty has put online. Right.

31:38

I just think that there's, here's

31:40

why I say this. Look, I understand if

31:42

you're a copyright holder, you're terrified, blah blah

31:44

blah, but there is so

31:46

much value societally to be gained from an

31:48

AI that's smart, and there's so much to

31:50

be lost from an AI that's hobbled. You

31:52

know, one of the stories this

31:54

week is that AI has

31:57

not been allowed to scrape things like the Washington

31:59

Post. in the New York Times, but right-wing

32:02

news media has been welcoming

32:04

AI scraping. And so

32:06

what you're going to get is AIs that are

32:08

trained on right-wing media but

32:10

not left-wing media. And I

32:12

don't think that's good for anybody. And so

32:14

I think it's a mistake to say what AI can read and

32:16

not read. The New

32:18

York Times lawsuit saying, oh, you're going to read the New

32:21

York Times in an AI in

32:23

a chat GPT instead of on our page is

32:26

obvious nonsense. No one's going to do

32:28

that. They're just trying to

32:30

extract some money out of them. As

32:33

a creator, I don't think that's wrong.

32:35

So first up, as someone who's created

32:37

content, if you

32:40

have created your content with the idea that

32:42

it is for people and reading for people,

32:44

right, and for a certain use case, like

32:47

as a journalist, it's a

32:49

service to my readers or whatever, I have

32:51

an audience in mind. And I think

32:53

that's an important thing to

32:56

like, you can't go back and change that

32:58

contract, which is kind of what AI is.

33:00

They're like, oh, yeah, we're going to now

33:02

use this thing that you built that's still

33:05

technically owned by you or your

33:07

publication usually. And we're

33:09

going to use it for this instead. I

33:13

don't think, I think

33:15

your arguments about like, oh, right-wing

33:18

media is going to allow it to

33:20

be trained as opposed to left-wing media.

33:22

I think there's a place for making

33:24

that argument and then having people produce

33:26

content for AI that is quality content

33:28

because of the benefits to society or whatever

33:30

the hell we want to argue. But

33:33

I don't believe that saying

33:35

I should give my content

33:37

to a multi-billion dollar company

33:39

just for better training

33:41

of something they're going to make money

33:43

on is really... What if it weren't

33:45

for them to make money? What if

33:47

it were open source AI? I do

33:49

agree. I don't want to

33:51

see Google or Microsoft dominate or open AI

33:54

for that matter. Dominate AI, I

33:56

think it should be open source. But

33:59

so if you would be okay... if it's somebody who's not

34:01

going to make money in your country. I would want someone

34:03

to ask. It's kind of

34:05

like someone walking into your house and being like,

34:07

man, this place is super nice. Let

34:10

me settle on in. Well, interestingly, the

34:12

reason this is happening is because

34:15

these companies are saying

34:18

you have to ask and open AI and

34:20

Google and others are asking. This

34:22

is from Originality AI, which is

34:24

an Ontario-based AI detection

34:26

startup via Wired magazine. Data

34:28

collected in mid-January on

34:30

about 40 top news sites shows

34:33

that almost all of them use robots.txt

34:35

or something like it to block AI

34:38

web crawlers. The New York Times, The Washington

34:40

Post, The Guardian, The Atlantic, Bleacher

34:42

Report, OpenAI's GPT

34:45

bot is the most widely blocked

34:47

crawler, but none of the top

34:49

right-wing news sites surveyed, including Fox News, The

34:51

Daily Caller, Breitbart, block any of the most

34:53

prominent AI web scrapers. They're

34:55

actually explicitly saying, the sites that are

34:58

blocking it are explicitly saying, no, you

35:00

can't look. And to their

35:02

credit, OpenAI and Google are honoring it, but you

35:04

can see the problem from a societal point of

35:06

view even to that. So

35:09

yeah, well then what's to stop, I

35:11

don't know, Microsoft or Google from saying,

35:13

wow, we have a huge bias problem

35:15

and it's partially because of our trading lines.

35:17

So biased. What's going to happen? What

35:21

I wondered about, we've seen that happen in

35:23

our article. Face recognition, I mean, face recognition

35:25

is notoriously terrible on people of color because

35:28

it's mostly, and this is just an inadvertently,

35:30

this isn't even intentional, but it was trained

35:32

mostly on white faces. So of course it's

35:34

terrible on people of color. But

35:37

you want to fix that, or do you just

35:39

want to abandon face recognition and say, well, we

35:41

should just never use it. Something

35:44

I was curious about after reading that article was, is

35:48

it specific to AI or is it just that

35:50

you have a set of websites that are

35:52

trying to get as much traffic as they can

35:54

more aggressively and so they're just a lot more

35:57

lenient with the robots text versus

35:59

the other. sites. I don't know. I

36:01

didn't know. See that's and it wasn't and

36:03

it wasn't clarified in the article either. Right.

36:05

So I don't know what the answer is.

36:07

Well you can't determine the intention you can

36:09

just merely look at their robots.txt and see

36:11

what they're doing. This is

36:13

also deeply related to and I know to

36:15

the state of media which we talked about

36:17

before we got in the show which you

36:19

know if media were doing really excellent right

36:21

now there'd be less issue. Media

36:24

is suffering we have seen lots of

36:26

layoffs in the media world among our

36:28

friends and the

36:30

like this might be a

36:33

killer for some who have really cheap slow

36:35

margins or thin margins or it might be

36:37

a savior if they can get a payout

36:39

from someone like it open the ire up

36:41

but like long-term you know cat

36:43

is out of the way. People

36:46

are gonna have like this is why they're

36:48

mistral and others like raise money

36:50

because they're building the open source version of this

36:52

you could run the large language bottle from your

36:54

phone around your own device and

36:56

eventually people are just gonna have their own stuff

36:58

running and the thousand people a

37:00

billion a hundred thousand people are not

37:02

gonna all ask for permission each and

37:04

every single place. Well but really the

37:07

interesting thing and so you're talking about

37:09

Mistral I use that and I use

37:11

open AI's rather of Facebook's a llama

37:13

both of which are open models but

37:15

those models those are the

37:17

ones that are being trained on

37:19

this giant database of material

37:23

when I put this on my local

37:25

machine I'm doing the fine-tuning with with

37:27

various content stuff right so it still

37:29

has to be these are still from

37:31

big companies training right in the very

37:33

expensively by the way I'm

37:36

not gonna be shocked if you just like

37:38

and I'm already seeing it's more more starts

37:40

and others like doing their own training of

37:42

their own models and it'll become cheaper that's

37:44

interesting and it'll be easier and more effective

37:46

to go and do. Who does Mistral? Where

37:48

does Mistral come from? That's a company Mistral

37:50

AI right? Yeah they're they're based

37:52

in France and Anderson

37:55

Horowitz Bactam they're

37:57

not that old they're like less of the year old or

38:00

so, but it's like top researchers got together

38:02

and sometimes, you know, these are a

38:04

lot of money, something like that doesn't work. They do seem

38:06

to be working. It's providing a

38:09

different kind of value for those who want

38:11

to have a lot more control for their

38:13

large language model. There are a couple. Go

38:16

ahead. So on the subject of

38:18

that training at the edge and

38:20

training with your own datasets, and this was the last

38:22

thing I was expecting when I was in talks to

38:24

move over to Phizon, they're working on an AI

38:26

thing. Interesting. And it was

38:29

shockingly not gimmicky. Like my immediate assumption initially,

38:31

wait, an SSD company's working on AI, what

38:33

are you guys doing? And then no, actually

38:35

it was a completely legitimate use

38:37

case and there's like, okay, well, you have a GPU, it

38:39

only has so much memory. How do you train a model?

38:41

It's too big to fit in the memory. Oh,

38:43

you can add some SSDs. Ah, a

38:45

swap file. That are meant for that.

38:47

Oh, look, in essence, right? But you

38:50

can't just plug that in and hit go, it doesn't

38:52

work that way. You have to develop the actual technology

38:54

and the software and everything to work together. It

38:56

really is a gold rush, isn't it? Yeah,

38:59

we tease it at CES. We're working on a thing. It's

39:01

a gold rush. Like for a legit purpose. Yeah. So

39:03

I've used two different programs and I

39:06

would like to tell people about them. One

39:08

is called Olama at olama.ai and

39:10

it's based on Facebook, Meta's

39:13

open AI model,

39:15

Llama, Llama 2. But

39:18

you can download other models, including the Mistral

39:20

models. I've set that up and I've also

39:22

played with something called GPT4ALL from Nomic. Again,

39:25

the idea is you download

39:27

these models that are trained by

39:29

sometimes by big companies, sometimes by Mistral, sometimes by

39:31

a variety of people. These are the various

39:34

models you can download. You can

39:36

show the screen. And

39:39

then you would then either fine tune it

39:41

or create your own AI. One of the

39:43

most popular things to do is

39:45

to create a PDF reader that could summarize content

39:47

for you. And I've been doing that locally and

39:49

it's great. It's really, it's kind of amazing what

39:51

you can do, but you do still need these

39:54

big models. I think that'll be very interesting. Do

39:56

you think Ben actually there's going to be a,

39:58

a, a, a, a, a, a, a, a lot of

40:00

people who will be training their own models

40:02

eventually? Oh, yeah. I've

40:04

already seen more and more startups, you know,

40:07

with a lot of resources training their own

40:09

models for very specific things. Already

40:11

seeing people who train models, like they might use

40:13

the basis of an open source model and they

40:15

train it a lot more. Right. But

40:18

they're training it on things like, you know,

40:20

the medical industry, for example, you're not going

40:22

to go and use out of the box,

40:24

chat, GPT, open AI. So

40:26

you might use the large language model

40:28

like LAMA to do the basic

40:30

language stuff, so somebody's generated that and then you're

40:32

going to fine tune it for a medical application.

40:35

Well, and you have to compartmentalize that for HIPAA. You

40:37

can't, the data can't go the other way. Can't go

40:39

out of your... Yep. Right. That's

40:42

really interesting. Boy, I feel like AI is

40:45

the thing that people should be focusing on

40:47

right now, not AR. But as

40:49

you say, that's a dumb distinction, Stacy.

40:51

I'm making it silly, foolish... Well,

40:55

just saying AR versus AI is

40:57

not really... It's not really funny. They go together.

40:59

You're right. That's an excellent

41:01

point. The next thing I want to see, which

41:03

I don't know if it's going to happen. I hope somebody

41:05

does it. If anybody's listening and just working on it, then

41:08

more props to you. But like, we're getting

41:10

pretty close to, if you've ever seen the movie Her,

41:12

where they had the AI, you download the

41:14

software on the phone. And like, I

41:16

want to see the thing where you train your own model,

41:19

you plug in some pieces of software on your desktop,

41:21

you train your own model on all of your own stuff. Right?

41:25

And you go, hey, what was that email from... Yeah,

41:27

that's the final thing. Yeah, I want to just do

41:29

a, hey, what's that email from five years ago about

41:31

this thing, you know, and just find it. I think

41:33

that's going to happen in the next few months. I

41:35

think we're there already, Brett. You can already use Rewind

41:37

to do some of that stuff. Yeah. Are

41:40

you going to get a Rewind? Are you going to do the Rewind? I'm

41:42

buying every single hardware. Did you buy Humane?

41:46

I have not bought Humane. I will

41:49

do that. That

41:51

one is a hard value prop. The Rabbit

41:53

R1 is a much... Did you get a Rabbit? I got

41:55

a Rabbit. I bought every one pending. I really

41:58

wanted the Rabbit. Yeah. I will see. get

42:00

a humane pin and go and try the whole

42:02

thing out. I'll try every single one. So all

42:04

of these devices are designed to have, they

42:07

still, I think all phone home though, right? Nobody's

42:09

running locally, especially the R1, which

42:11

really doesn't even have any apps. It just

42:13

phones. You can't run any of those locally.

42:15

Yeah. But it's adding the fine tuning from

42:17

your life. Not so much

42:19

Rabbit, but definitely rewind.ai and

42:22

the humane AI pin. It's recording

42:24

your experience, your exchanges, your experiences

42:26

and stuff. And then adding

42:28

that as a fine tuning to an existing

42:30

LLM. I, to me, you're right. That's her.

42:33

In fact, it was

42:35

no accident, by the way, that chat

42:37

GPT, when they added an iOS app

42:39

and an Android app had one of

42:41

the voices sounded surprisingly like Scarlett Johansson.

42:45

They didn't call it. It could just be, she has

42:47

an amazing voice. Yeah. Yeah.

42:50

Yeah. We got to take a break. This is a

42:52

great conversation. Don't hold that thought, Alan. I want to

42:54

keep going, but I do have to take a break

42:56

or we, or we will be going way too long.

42:58

Our show today, we got a great, great panel by

43:00

the way, Stacy, it's great to have you back. I

43:02

missed you. Welcome back. You and I are

43:04

going to do the book, the book group on

43:06

February 8th, right? We

43:09

are. That's the most depressing book I've

43:11

ever read. I know. I

43:13

was like, y'all, did

43:16

we really need to do this? So depressing.

43:18

I love Paulo Guy, Bacheco Lupe. I love the

43:21

windup girl. His new one is the water knife.

43:23

I haven't got to the

43:25

end. I'm hoping like that suddenly the sun will come

43:27

out and it'll start raining. Everything will be beautiful, but

43:29

maybe not. I don't know. It's a little grim. No,

43:31

that's not going to happen, is it? Well,

43:34

we'll talk about it February 8th in the club. That's

43:36

going to be Stacy's book club. Maybe

43:38

next time, can we choose a happy book? Next

43:41

time, please. I put happy books on

43:43

the things. It's the people. Who picks

43:45

them? It's the people. It's the voice of

43:47

the people. The voice of

43:49

the people. Ben Parr is the voice of AI. He is

43:51

the author of a great book. You should read the AI

43:53

analyst, co-founders. Eight years he's been

43:55

doing AI at Octane AI.

43:57

Great to have you. Alan

44:00

Malventano now at Faison, which

44:03

is just as easy to

44:05

pronounce as Soledain, but

44:07

only a little bit harder than Intel. He's

44:10

been an all-time... I am dragging

44:12

the SSD industry into the future. His

44:16

next company will be Cyberdine. Cyberdine!

44:19

We thank you all. And then you might as well, you know,

44:21

I'll be working with Ben and we'll just take over the world.

44:23

I thank you all three for being here.

44:25

Our show today brought to you by NetSuite.

44:28

Once your business gets to a certain size, the

44:31

cracks start to emerge. You see this happen

44:34

every time. Small business, you know everybody's

44:36

name, no big deal. But as soon as, you

44:38

know, you get to the point where who... do we

44:40

hire you or you know... And then there's just

44:43

too many people, too many manual processes to

44:46

keep track of. This is... If this

44:48

is where you are or about to be, you should

44:50

know three numbers. 37,000, 25 and 1. Okay?

44:57

I'll explain. 37,000, that's the

44:59

number of businesses that have upgraded

45:01

to NetSuite by Oracle. NetSuite is

45:03

the number one cloud financial system. Streamlining,

45:06

accounting, HR and more.

45:10

25, that's how old NetSuite is. It turns 25 this

45:12

year. That's pretty impressive.

45:16

25 years of helping businesses do more

45:18

with less, close their books in days,

45:20

not weeks, and drive down costs. I

45:23

bet you can guess the one. You're

45:25

the one. Your business is one of a kind. So

45:28

you get a customized solution for all

45:30

your KPIs in one efficient system, a

45:32

single source of truth. That's

45:34

another one right there. Manage risk, get

45:36

reliable forecasts, improve margins, everything you

45:39

need to grow all in one place. Right

45:42

now, download NetSuite's popular KPI

45:44

checklist designed to give

45:46

you consistently excellent performance and it's

45:48

absolutely free. netsuite.com/twit.

45:53

That's netsuite.com/twit. It's your

45:55

own KPI checklist, netsuite.com.

46:00

We thank them so much for supporting

46:02

this week in tech. Great

46:06

conversation about AI. I'm

46:08

like you kind of Ben. I want to try all the AI

46:11

things. I don't want to try unlike you.

46:13

I don't want to try any of the VR things. I

46:18

got this close to buying the Vision Pro.

46:20

I mean, literally I did the vase scan,

46:22

uploaded my prescription, and I got

46:24

my finger hovering over the $3,500 button. Actually,

46:27

it was more because I had to buy

46:29

the $200 travel case. I

46:31

had to buy the extra sweatband. I

46:33

had the $99. I

46:36

had to buy the glasses because I do

46:38

need corrective lenses, $149. So

46:41

I got a well over $4,000 before I was done. And

46:45

then I said, no way. I

46:47

said, I have an Oculus. I

46:50

have a MetaQuest Pro that

46:52

I spent $1,400 for. It's

46:55

got a nice thin layer of dust on

46:57

it in my office collecting dust. I

46:59

knew I would buy this. I would use it for

47:02

a week, get all excited about it, and then it

47:04

would just collect dust. I couldn't justify that. So

47:06

unlike you, I'm not buying one of those. So

47:10

my point before the break, which will

47:12

overlay us into the rest of the

47:14

topics, I'm sure, which applies to both

47:16

the vision-related things, the VR things, and

47:18

the AI discussion we just had. As

47:21

far as having a thing that's very powerful in

47:25

your house, in your position, right? And

47:28

the reason I'm bringing this up is all these technologies

47:30

that were like, oh, well, this might never get here

47:32

to this state where you can just have the thing

47:34

in your house doing all of it. It's going to

47:36

have to go to the cloud. Not necessarily, right? About

47:40

this 25 years ago, this is a smart media

47:42

card for those who remember what they are. This

47:44

is too mad. Wait a minute. Wait a minute.

47:46

That was like a memory card for what? Smart

47:48

media for cameras. Oh, yeah. For giants. Yeah, for

47:50

giants. Look how big that is. Look at how

47:52

big that is. Well, it was really thin, but

47:54

okay. I had one of those.

47:57

That's too mad. A few years later, Sony made

47:59

a mistake. this little micro volt USB thing that

48:01

was really tiny that's 2 gig. Wow. Right?

48:05

Yeah. And then here's an M.2 2242 2 terabytes.

48:09

Wow. Right? Yep.

48:12

And that's across the span of 25 years. Yeah. That's

48:15

a million X. It's kind of remarkable. I

48:17

mean we saw this with hard drives too

48:19

but it's really remarkable and what I

48:22

find fascinating is that there was a lot of

48:24

concern early on that these would not be as

48:26

reliable as spinning media. What's your

48:28

experience? They seem like they're

48:30

actually as reliable if not more reliable. As long

48:32

as you understand what the caveat is which is

48:35

just for flash memory there's a finite amount of

48:37

times you can rerun a spot on it. But

48:39

you got wear leveling. Surprisingly. You got technology surprisingly.

48:41

The car did not understand. Yeah.

48:46

And there's technologies to you know mitigate that.

48:48

I remember Woz coming onto the screen savers

48:50

in about 2002 and he had a USB

48:52

key around

48:56

his neck. And he said

48:59

this is 2 gigabytes. Everything

49:02

I could ever want is on this.

49:04

Actually it might have been this because

49:06

it came with a little silicone pouch

49:09

and a little lander thing. It probably

49:11

was that. It could have been. Ooh

49:15

2 gigabytes. And his whole

49:17

life is around his neck. Now

49:19

my iPhone this has 512 gigabytes. Some

49:23

people even are getting the Vision Pro with a terabyte.

49:25

Did you get a terabyte Ben? I

49:28

did get a terabyte. I'm going to get the most

49:30

amount of things. Okay well that's

49:32

good. Now you're future proofing it. Yes.

49:35

Yeah. But all these technologies that

49:37

you would think have to use

49:39

the broader internet and the cloud and has

49:42

to be in a big server somewhere in

49:44

another state. Not necessary

49:46

right? Just give it a few years. It's the same

49:48

compute will be in your house. I

49:52

don't think the business models. Like

49:55

I would love that and you're right.

49:57

We totally could have computing do

49:59

more locally. But the business models

50:01

don't make sense because then companies can't control

50:03

it to the extent that they want to

50:05

and monetize it to the extent that

50:08

they want to also Yeah,

50:10

it's mostly business models, but also people have become

50:12

less trained for it Yeah,

50:15

Leo, I don't know if you have an article related to

50:17

this for this week But the error covered in a recent

50:20

weeks, but there's a big battle between the the home Home

50:23

cloud where you take all your devices

50:25

and have them You know on your

50:27

own server versus using whatever the company's

50:31

You know cloud-based server is for example There's

50:33

a been a thermostat company in the news

50:36

recently that was going after the

50:38

your home lab type You know

50:40

someone made a plug-in for I

50:42

forget what the home assistant Yeah,

50:45

there's people going after home assistant developers that are just

50:47

for free reverse engineering the protocol and just making hey

50:50

I just want to control my thermostat from my house

50:52

if I have no internet and Apparently

50:54

that's a thing that you know these companies are taking

50:58

Issue with the public is a little schizophrenic on

51:00

this because on the one hand we we talk

51:02

a lot about Privacy and we don't want

51:04

these companies scraping our data And we know

51:06

that you know a internet connected thermostat is

51:08

of course sending that information to the company

51:11

Which is then selling it off to

51:13

somebody who for some reason wants to know how hot my

51:15

house is So on the

51:17

one hand we you know we we talk

51:19

about a lot we talk about oh, there's a

51:21

bad thing But the other hand we really

51:23

don't want to run these things at home And

51:27

I think oh maybe these markets are

51:29

pretty brisk for things like iCloud and

51:31

OneDrive and Dropbox and

51:34

I mean well I mean I have some

51:36

home lab things right I don't do

51:38

Dropbox anymore so much I have my

51:41

own you know you

51:43

have more storage than God You

51:45

know well you're also a Technically

51:48

a dead well here's a better example here's a better

51:50

example that someone can even run on a Raspberry Pi

51:52

in their own Home with a docker container and that

51:54

you've talked with Steve Gibson on on security now a

51:56

bunch about this where you can have a docker Container

51:58

that holds all your passwords Right on a

52:01

very small doesn't need a lot of storage Is that

52:03

a completely reasonable thing especially given all the data breaches

52:05

you hear about? You

52:07

know right that's a you know very private information You

52:09

would think people would be interested in doing that and

52:12

our sponsor bitwarden among others allows you

52:14

to do that I don't

52:16

think a lot of people I think

52:18

they're people like you know, yeah, they don't want

52:21

to pie for everything They I don't even do

52:23

it and I'll tell you why I don't do

52:25

it because I think bitwarden probably is better It's

52:27

securing that vault than I am I'm

52:30

much more likely to do some dumb thing with

52:32

that vault All right Seriously,

52:35

well so the part of it where I

52:39

Have I don't want to call it a disagreement, but

52:41

I mean I have a nest thermostat I'm not

52:43

you know I'm not relying on that thing But

52:45

if I did want to use home assistant and

52:48

switch stuff over to my own for

52:50

some functions I should be able to

52:52

do it right I shouldn't be like I

52:54

shouldn't be for of course You should be able to

52:56

do it and I think my opinion It's

52:59

just as related right to repair It's

53:01

companies should just go ahead and do it because 99% of

53:04

their customers are not gonna do it

53:07

So just let them do it. It looks good for you and

53:10

And people are still gonna do the convenient and

53:12

easy thing It's

53:14

why it's so bizarre that Apple is

53:17

being so malicious with

53:19

the EU over its app

53:21

store When it knows perfectly

53:23

well Even if there is a third-party app

53:25

store available on the iPhone that most iPhone

53:28

users like 99.9% Will

53:31

never even know it exists Let

53:33

alone you know I have a theory about

53:35

this. Yes. I think what

53:37

is what's happening is they're They're

53:40

not thinking today about the issue.

53:42

They're thinking about the long-term implications.

53:44

So think about like basically

53:49

My child doesn't pirate music because they

53:51

never have had to write but I

53:53

can pirate music because when I was

53:55

you know how Really

53:57

great right and you had to write right?

53:59

I was like how you got music. The

54:05

kids today, when

54:07

you're looking for cheap solutions or reliable

54:10

solutions and that's usually people

54:12

who have less income and more

54:14

time. So generally younger people they'll

54:17

train themselves to pull

54:19

away from the traditional business models and use

54:21

this stuff and companies don't want to open

54:23

that door. I don't think that's stupid of

54:26

them. Interesting.

54:28

So Apple's concerned about, see

54:30

I always thought when

54:32

you put a lot of anti-piracy protection

54:34

on stuff it

54:36

teaches, this

54:39

is my example, I know this isn't exactly applicable,

54:41

but copy protection

54:43

teaches people to be pirates because

54:46

it's only normal end users that

54:48

are baffled by it. Pirates know perfectly well how

54:51

to get around all this stuff. They're like Alan,

54:53

they know how to have all the storage and

54:55

do all the things and stuff. But

54:57

normal people who

54:59

are thwarted by copy protection

55:02

learn how to get around it. So the best

55:04

answer, in fact this is what the music industry

55:07

finally came to in Apple, is not to have

55:09

copy protection and just

55:11

make it easy to buy music. And that's

55:14

why your kid just buys music because

55:17

they don't know that there's other ways. They

55:19

don't buy music. They don't even buy music. They read

55:21

it. They read it. They don't even buy it. They

55:23

read it. Because they don't know of any other way.

55:26

Right? But you don't want to train people on how

55:28

to get around stuff. So you're saying Stacey that by

55:31

providing a third party app store

55:33

Apple is teaching people that they

55:35

can't get around it? Well what

55:37

I'm saying is by

55:41

leaving those avenues open more and more

55:43

people will see value in going

55:46

around it. And then over time

55:48

the number of users

55:50

of the legit app

55:53

store reduces. And same

55:55

thing with like Home Assistant or people. It's not

55:57

so much the issue today. It's the people. people

56:00

who are like, oh, I'm going to use Home Assistant,

56:02

and then more and more people do it, and they make

56:04

it easier. Home Assistant is getting better and better over time.

56:07

Just for people who don't know, including yours

56:09

truly, Home Assistant is an open source

56:13

home automation tool? It

56:15

is an open source home automation tool. So

56:17

this is like OpenHAB or one

56:19

of those, nobody use, you told me, you

56:22

told me don't use OpenHAB. Nobody uses OpenHAB.

56:24

Home Assistant is the way to go if

56:26

you're going to do it. I

56:28

think though, by the way, anybody automating their home

56:30

is already in the category of super geek, but

56:32

that's just me. Did

56:36

you take a lot of home automation out of your house,

56:38

Stacey, when you didn't have to do it anymore?

56:43

So I took the Madame A devices, so

56:45

the Amazon Echo devices all left because they

56:47

were the most annoying devices. I

56:49

kept the Google and

56:53

still hoping they get their act together

56:55

on their stuff. And

56:57

I did eliminate a lot of the stupid

56:59

gadgets that you could talk to. Do

57:03

you still have a faucet you can talk to? So

57:06

I still have a faucet that I could talk to, but

57:08

that faucet only works through Madame A, so

57:10

now I can't talk to it. Perfect.

57:14

That's a harder uninstallation process, Leo.

57:16

It requires plumbing. Right. So

57:19

I installed it myself. It wasn't so bad.

57:24

I like to do things. Do you run Home

57:26

Assistant now, Stacey? No. No,

57:29

I don't. The only thing I do

57:31

is, I mean, I

57:33

have more home automation than the average bear, but

57:36

it's mostly light. Stacey, if you might remember, if

57:38

you watched this, we can Google, her husband used

57:40

to complain about how hard it was

57:43

to open up blinds. Do

57:45

you still have blinds you can talk

57:47

to? No

57:51

one talks to the blinds. Theoretically, you could talk

57:53

to the blinds, but everyone has forgotten the fact

57:55

that they pull the string

57:58

attached to the blinds that opens. We

58:01

still have remote control. Oh, you ever. This

58:04

is reminding me my fiance. This is when I

58:07

got to, I saw you like a year and

58:09

a half back in my fiance

58:11

had a play premier in Sonoma. Yes. Called

58:14

Atlas Lomani Gibbon. It's about a

58:16

smart home going awry and like the inserts

58:18

of the fridge exploding and go through all

58:20

sorts of stuff. And like the impact of

58:22

like what if you have AI running things,

58:25

but also the impact of like how that

58:27

affects things like relationships. I just

58:29

had to go and plug because I love my fiance, but also

58:31

we are answering world where these

58:33

are going to be the conversations and I think going

58:36

back to Stacy's original point before, you know, you

58:38

have a free time as a kid, you're going

58:40

to hack what everything you have availability to hack

58:42

right now. You don't have the availability to hack

58:44

on the iPhone. Nothing no more. Kids

58:46

don't even know how to type. Kids

58:49

don't know how to hack. Kids

58:51

are too busy playing the TikTok

58:53

on their phone. They're not.

58:56

You want to see a kid hack something? Take TikTok

58:58

off their phone. That's how they get it back

59:01

on. I mean, seriously, you

59:04

just have to know what they want and make

59:06

it hard for them again. Yeah, they hacked the

59:08

parents. They'll learn how to pirate. By

59:10

the way, complete aside, Leo, while we were having

59:12

the conversation, I did make a phone. I

59:15

did make a mid-journey of you as a pirate. I

59:17

put it in the show notes. Oh, I do. If

59:19

you want to take a look. Pirate Leo. No,

59:22

and I'll confess to

59:24

having downloaded, I think,

59:27

90,000. 90,000 songs on

59:29

Napster back in

59:31

the day. Who didn't? That

59:34

was around. Yeah, who didn't? And that's

59:36

why. Yeah, by the way, mid-journey does

59:38

not know what I look at. Look like I just want

59:40

to say it is. I've

59:43

had that problem. It looks pretty close. If you think that

59:45

looks like me. You look as old-ish. Well,

59:47

in the upper right is John

59:49

Cleese. Yeah. It's

59:52

just an old fat guy, basically, with

59:54

gray hair. I'm the generic

59:56

old fat guy with big nose and gray

59:58

hair. That's it. right there. You

1:00:00

might as well have just used that as the prompt.

1:00:04

At least it got the hat right. Actually,

1:00:07

somebody was asking me about

1:00:09

my chess.com login because I

1:00:12

used an AI

1:00:14

for it

1:00:17

of me and it kind of from a distance it

1:00:19

looks like me. I can't show it because I'm not

1:00:21

logged in. Alright, I'll use this

1:00:24

pirate somewhere. Thank you, Ben. I

1:00:28

love Mid-Journey. I think playing with Mid-Journey is

1:00:30

great. It's one of the pieces that made

1:00:32

me kind of start to think, you know,

1:00:34

I want a smart AI. I want a

1:00:37

good AI and I don't really care if

1:00:39

Thomas Kincaid, the painter of light, is pissed

1:00:41

off that Mid-Journey can do a crappy painting

1:00:43

like him. That just doesn't

1:00:46

seem to me to be

1:00:48

a societal problem. The

1:00:50

heirs of George Carlin.

1:00:53

Look, I understand. I support the heirs

1:00:55

of George Carlin who are

1:00:57

very upset about this fake

1:01:00

George Carlin comedy

1:01:02

routine titled, I'm

1:01:05

Glad I'm Dead. Okay, maybe

1:01:07

that was a little offensive. It's

1:01:10

not very funny, but for

1:01:12

people who are George Carlin fans, maybe

1:01:15

they're kind of happy. It came out a

1:01:17

couple of weeks ago on the YouTube channel.

1:01:19

It's from a podcast called Doodsy, Will Sasso

1:01:21

and Chad Cochran. They're being sued

1:01:23

now by the state of George

1:01:26

Carlin. We

1:01:28

have to draw a line in the sand,

1:01:30

says daughter Kelly Carlin. Do you

1:01:32

think they'll win this suit? I feel

1:01:36

like they have a fair argument. Yeah,

1:01:38

I really agree with them. I would sue in that

1:01:41

case too. I think they're going to lose. I

1:01:45

am not a legal expert at all. Can you

1:01:47

do Rich Little for doing an impression of George

1:01:49

Carlin? That's

1:01:53

different. It is.

1:01:56

Legitimately, over the course of

1:01:58

the next two years, we're going to start to get some answers

1:02:00

to a lot of questions that are surrounding

1:02:03

the legality of certain things with it. From courts.

1:02:06

From the courts. From courts. It's

1:02:08

going to take a while. I don't expect major

1:02:10

rulings to be in effect this year. It'll take

1:02:12

several years probably. You know what I fear though,

1:02:14

Ben, is that we're going to get conflicting

1:02:17

rulings because there's so many courts

1:02:19

involved. Oh, for sure. Supreme Court. And

1:02:21

so you're going to get some, it's going to end up

1:02:23

having to be, but then the Supreme Court's not going to

1:02:25

really resolve this, I don't think. I

1:02:28

think they're going to do a very

1:02:30

constrained, well in this one case, I

1:02:32

don't know. Supreme, if there's lots of conflicting

1:02:34

cases, the Supreme Court will. They've got to resolve

1:02:36

it. They've got to resolve it. They've got to

1:02:38

resolve it. It's going to be a thing that

1:02:41

we really will have to figure out what is

1:02:43

the right line and

1:02:45

what is the right legal line. This

1:02:48

is complex stuff that we'll have to just get

1:02:51

four lawyers to go and debate for a couple

1:02:53

hours. I will watch that. Because

1:02:55

I don't think there's a, even

1:02:57

among our panelists, there's no consensus at all.

1:03:01

The other thing is like, remember too,

1:03:03

we don't fully understand every aspect

1:03:05

of the technology. Like, AI, there

1:03:08

was a story earlier this

1:03:10

week, an AI that

1:03:13

started like, the

1:03:15

point in itself, it was essentially a point and

1:03:18

it went rogue and it was actually deceiving the

1:03:20

people who had trained it and

1:03:22

lying to them about its training. They

1:03:25

could not get it back on track for training at all,

1:03:27

so they had to abandon it. That

1:03:30

kind of thing is happening and

1:03:32

there was no real explanation

1:03:35

yet from some research or so. I

1:03:38

get my whole point. It's like, this is

1:03:40

all cutting edge stuff. There's amazing things you

1:03:42

can do with AI. We do not fully

1:03:44

understand every aspect of it. We

1:03:46

do not fully understand where the legality of the

1:03:48

things are going to go. At a certain point,

1:03:51

the AI is just going to decide to make

1:03:53

things on its own without you saying anything and

1:03:55

then who is responsible then? There's

1:04:00

an interesting idea about like from

1:04:03

a legal perspective making AI a

1:04:05

separate like legal class

1:04:07

like. Yeah, I think that's what you'll have

1:04:09

to do because humans can do and interpret

1:04:11

can do you know, I can

1:04:13

do a George Carlin impression. Humans

1:04:16

can read your articles Stacy and then summarize

1:04:18

them to somebody else and

1:04:20

those are all I think we

1:04:22

all accept normal uses transformative uses

1:04:24

of content. You

1:04:27

feel like it shouldn't be a machine. In fact,

1:04:29

you're very specific. You shouldn't be a profit making

1:04:31

machine. You would you would say a

1:04:33

nonprofit that's okay. It's gonna

1:04:35

be very hard. You're gonna have to create new classes. I think you're

1:04:37

right. Well, and I

1:04:39

mean, there's gonna be a I mean just

1:04:41

on the create like content creation

1:04:43

side. Over

1:04:45

time, if you have AI generating everything,

1:04:48

it's going to devolve into crap,

1:04:51

right? So you're gonna

1:04:53

have to have people in the mix somewhere

1:04:55

and you're gonna have to that's right pay

1:04:57

them accordingly. So we just we're in

1:04:59

this like weird area where we're just

1:05:01

like, ooh, let's arbitrage this and make

1:05:04

all the money and gather all the

1:05:07

gather everything we can. Let me ask Ben because

1:05:09

you're that you're trying to make money on AI.

1:05:14

I think there's such potential for AI. I

1:05:16

mean, if we really want her or you

1:05:19

know, if we want AI to design chips, AI

1:05:22

has already come up with interesting

1:05:25

new medicines. It's

1:05:27

very good at protein folding. But

1:05:30

those aren't LLMs that do that. Well, whatever just

1:05:32

generative AI of some kind. In fact, I think

1:05:34

it's a mistake to say LLMs are different than

1:05:37

GANs are different than that. It's

1:05:39

all generative AI. And it all has

1:05:42

and all of them have to ingest

1:05:44

human created content. I agree with

1:05:46

you, Stacy, that if we're an AI

1:05:48

to be good, a human has to have

1:05:51

input. But we get such

1:05:53

societal benefit out of it. I mean, really,

1:05:55

this is going to be a very bitter

1:05:57

argument. I can tell already. But

1:06:00

Ben, don't you think that that's, I

1:06:04

think you might hobble AI. I think you

1:06:06

might actually kill AI in the cradle by

1:06:09

being too restrictive on what it can read. This

1:06:13

is the ultimate debate between, for

1:06:16

those who know, E slash ACC versus

1:06:18

EA, effective altruism versus effective accelerationism, which

1:06:21

for those who don't know is two

1:06:23

camps, one being like we should move

1:06:25

as quickly as possible because AI could

1:06:28

save lives and every day you wait

1:06:30

to train the AI, the life that

1:06:32

didn't have to be lost versus EA's

1:06:35

which is like AI could destroy the

1:06:37

entire planet. So we should be very

1:06:39

cautious because have you seen Terminator before?

1:06:42

Yeah, I'm not a doomer. And

1:06:45

I think that's a, I think maybe that's gone away. I

1:06:47

don't know. Elon Musk was one of the doomer.

1:06:49

Oh, no, no, no, it hasn't gone away. Oh, no, no, no, no. It's

1:06:53

less like provenance before, but

1:06:56

there's still like a lot absolutely

1:06:58

there. It's just more underground.

1:07:00

And I have those conversations with- I'm an

1:07:02

accelerationist. You might be surprised to hear. I

1:07:05

was never a doomer because I thought that's just sci-fi. But

1:07:09

I am now accelerationist. I believe, in

1:07:11

fact, I might not even disagree with

1:07:13

Larry Page who says it's

1:07:15

time for humans as a species to get out of

1:07:17

the way of the next big thing. We

1:07:20

only have merch for that. There's

1:07:22

an effective accelerationist. I know. I'm

1:07:25

in that camp. I'm

1:07:28

in that camp. I've been, the machines have

1:07:30

won me. Over. I

1:07:33

think Leo might, may or may not be an AI.

1:07:36

I always find the answer to that. Well, this is a good example.

1:07:38

I'm a creator. Now, admittedly, I'm at the end of my career. But

1:07:41

I have hundreds of thousands of videos online that

1:07:43

could be used to train an AI. And

1:07:46

in a year or two, we're not far off

1:07:48

from a time when you could, I mean, the

1:07:50

George Carlin's surprisingly good, when you would be able

1:07:52

to create an AI Leo that looks just like

1:07:54

this. Maybe have a human

1:07:56

on the other end typing in some

1:07:59

content. But

1:08:01

I could live forever as

1:08:04

this created entity. And it doesn't

1:08:06

bother me at all. Leo

1:08:09

even already has a premade avatar in the

1:08:11

form of his digital self from the screensavers.

1:08:13

I do. I was a

1:08:15

digital character actually on MSNBC's site. But

1:08:19

even more we have in our Discord, if you're a Club

1:08:21

Twit member, there's a A.I. Leo who's been

1:08:23

getting better and better, by the way. I don't know

1:08:25

how he's getting better. I'm a

1:08:27

little nervous. So what are you going to let

1:08:29

A.I. Leo host the show? And then 20 years

1:08:31

from now, A.I. Leo

1:08:35

has hosted the show. Yeah, it's just a matter of

1:08:37

time. That's fine with me.

1:08:39

Leo's in his comfy chair sipping his cognac.

1:08:42

It's not even about me. It's

1:08:44

about if the content, if an A.I.

1:08:46

can create good content or solve or

1:08:49

cure cancer, why shouldn't we want

1:08:51

that as opposed to saying, well,

1:08:53

no, I made this and you know, I

1:08:56

am the unique one and only and you can't have

1:08:58

me. That seems very selfish.

1:09:01

Well, no, no, for

1:09:04

your case, Leo, well, obviously you should benefit

1:09:06

from that if you like. No, I don't

1:09:08

care. I don't want money out of it.

1:09:10

No. If it helps

1:09:12

people and if it fulfills the mission that

1:09:14

we created Twit to inform people about technology

1:09:16

so they can use it, more

1:09:20

power to it. Now I agree with you,

1:09:22

Stacey. I don't want Google to be making money off of it.

1:09:25

So maybe that's the difference. Well,

1:09:28

and what are the assurances you have? I mean, you're

1:09:30

going to have to make some provisions for if you

1:09:32

have an A.I. Leo for making

1:09:35

sure it's still accurate, which means you need someone because

1:09:37

of the type of job that you have. You still

1:09:39

need humans. So you're going to have to have funding

1:09:41

for that. So then A.I. U

1:09:44

is still going to need to make money somehow

1:09:46

to pay for the humans on the back end

1:09:48

to check like QA humans at a

1:09:50

minimum. Well, and that's of course open A.I.'s argument

1:09:53

was it got so expensive we couldn't be a

1:09:55

non profit. We had to have a

1:09:57

for profit arm to foot the bill. What do you think

1:09:59

of it? I'm gonna take a break and I'm

1:10:01

gonna ask you what you think of Nightshade, which

1:10:03

is a tool designed to poison

1:10:05

AI Gives

1:10:08

artists a fighting chance against AI. We're gonna take

1:10:10

a break and talk about that when we come

1:10:12

back Stacy Higginbotham

1:10:15

Ben Parr Alan Malventano

1:10:17

great to have you the show

1:10:20

today brought to you by Ecamm and

1:10:22

I know you all know Ecamm we

1:10:24

use Ecamm So when I set up

1:10:26

twit back in the day, we have

1:10:28

big fancy hardware Switchers and

1:10:30

mixers and all this stuff and

1:10:32

cost millions of dollars very expensive

1:10:35

and then along comes Micah

1:10:37

Sargent who does iOS today

1:10:40

all on his Mac using Ecamm It's

1:10:42

kind of amazing the leading Livestreaming

1:10:45

and video production studio built to

1:10:47

run on your Mac. That's all

1:10:50

whether you're a beginner or an expert

1:10:52

Ecamm is here To elevate your video

1:10:54

production from streaming and recording the podcasting

1:10:57

to presenting To doing exactly what

1:10:59

we do here at Twitter Ecamm live is you're all

1:11:01

in one video tool Perfect for

1:11:03

simplifying your workflow Ecamm

1:11:05

supports multiple cameras. It supports

1:11:08

screen sharing. So I Mean,

1:11:11

I'm watching it happen on iOS today. Well,

1:11:13

it will switch to Rosemary. You'll switch to

1:11:15

her screen She'll switch to a phone.

1:11:17

She's holding back to Micah It's

1:11:20

a live camera switcher that lets you and

1:11:22

them direct their show in

1:11:24

real time You'll stand out from the crowd

1:11:26

to with high quality video Oh and All

1:11:30

that lower third stuff we do

1:11:32

logos titles graphics They

1:11:34

all work in Ecamm Ecamm supports

1:11:37

it. It's built in you

1:11:39

could share your screen You can drop in video clips.

1:11:41

You can bring on interview guests. You

1:11:43

can use a green screen works great with green screen

1:11:45

and so much more Join

1:11:47

the thousands of worldwide

1:11:49

entrepreneurs marketing professionals podcasters

1:11:51

educators Musicians and other Mac

1:11:54

users for Lion Ecamm live daily. I

1:11:56

mean really this transforms your ability to

1:11:58

do video production anywhere in time. Try

1:12:01

it out right now. Get a month free when you

1:12:03

subscribe to any of Ecamm's plans. Ecamm.

1:12:05

E-C-A-M-M-E-C-A-M-M. ecamm.com/twit. Don't forget to

1:12:07

use the promo code twit

1:12:09

at checkout so they know

1:12:11

you saw it here. We

1:12:14

are big believers in Ecamm and I'll tell you if I

1:12:16

was starting all over again, boy, I would be using Ecamm.

1:12:19

I absolutely would. Nightshade.

1:12:21

This is an interesting

1:12:24

product created by the

1:12:26

University of Chicago. Let's see.

1:12:29

Artists poison their image data, making

1:12:31

it useless or worse even, disruptive

1:12:35

to AI model training. It'll actually

1:12:37

screw up the model. Ben

1:12:40

Zhao, the computer science teacher who

1:12:42

led the project, compared it to putting hot sauce

1:12:44

in your lunch so it doesn't

1:12:46

get stolen from the workplace fridge. This

1:12:50

must drive you crazy, Ben. It

1:12:54

doesn't drive me crazy. I

1:12:56

am a person who sees both sides

1:12:58

of everything here. The answer

1:13:01

probably between that whole debate of E-A-C-C

1:13:03

versus E-A is probably somewhere in the

1:13:05

middle. I

1:13:07

definitely tend to think that AI will

1:13:09

be a large good in the world.

1:13:11

However, we're going to see a

1:13:14

lot of pain. We're seeing the pain

1:13:16

now. We saw that with the internet.

1:13:18

We saw that with computers. This is

1:13:20

what happens with disruptive technology. For technology

1:13:22

to be really valuable, it has to

1:13:24

be disruptive. I don't

1:13:26

think preserving print newspapers

1:13:30

is the right way to handle this

1:13:32

problem. I don't think

1:13:35

Congress is now getting in the act

1:13:37

trying to save AM radio. Sorry,

1:13:40

it's over. We

1:13:43

should not be subsidizing it or

1:13:45

forcing companies to subsidize it. It's

1:13:49

over. Time and

1:13:51

transition do matter. You think it's

1:13:54

so fast that we do have

1:13:56

to kind of flow it Out.

1:14:00

The in any like technological was I

1:14:02

sat down with like a candidate like

1:14:04

one of the world's greatest ah my

1:14:07

professors and board member since he's been

1:14:09

around for many days. it's never see

1:14:11

this quick of anything and so there

1:14:14

is something to like we have. Like

1:14:16

the transition takes time like we can

1:14:18

see where it's going. ah but it

1:14:21

is painful already painful for you know

1:14:23

people who needed awry on lake aren't

1:14:25

or copywriting and I already know friends

1:14:28

when they already know people who have.

1:14:30

Less work or have been laid off

1:14:32

her. Their consulting business is a lot

1:14:34

less as it used to be. obvious

1:14:36

as a be do stuff and new

1:14:38

things but that's why the discussion of

1:14:40

things like you be I has popped

1:14:42

up and why even like Sam Altman

1:14:44

talks about you would have reservations income

1:14:46

so that every it's because you we

1:14:48

all our work my since everybody cigarettes

1:14:50

as low as a month from the

1:14:52

big tech companies icing night sage to

1:14:54

be illegal. Isaac. These

1:14:56

guys should go to jail for poisoning

1:14:59

Ai data. That's. What I think

1:15:01

I. Guess

1:15:03

that's I like vandalise. I'm gonna

1:15:05

go with states the on this

1:15:07

one that's vandalism. Your vandalizing a.

1:15:10

Ah, Models as much as

1:15:12

don't want some places. As

1:15:15

a success. If you want a is you want

1:15:17

to is basically did it. To me it's the

1:15:19

equivalent of somebody who's walking behind you if you're

1:15:22

recording like and tic tac video and put them

1:15:24

there middle finger up right? or like the guys

1:15:26

walk by news casters and like. Do

1:15:29

Something. I moved here and you're right.

1:15:31

It's legal. It's legal and but it's

1:15:33

rude. And more to

1:15:36

the point it doesn't do any real

1:15:38

damage to interview billion dollar corporations. although

1:15:40

it doesn't care if it's multi million

1:15:42

dollar open source as discover poisons get

1:15:44

one of the things that you as

1:15:46

use the Mona Lisa. What it does

1:15:48

is it does a night shade version

1:15:51

looks just like the real Mona Lisa.

1:15:53

Maybe we'll darker but actually to an

1:15:55

ai looks like a cat. and

1:15:57

so the ai so fascinating AI

1:16:00

starts... If you're teaching people how the AIs see

1:16:02

things, it would work. It could

1:16:04

take fewer than a hundred poison

1:16:06

samples to corrupt a stable diffusion

1:16:09

prompt. So what

1:16:11

this is going to end up doing is

1:16:13

just... they're just going to adapt and overcome. Oh,

1:16:15

believe me. This isn't going to

1:16:18

last five seconds. I agree

1:16:20

with you. This is not really a threat. But

1:16:22

if it were, it should be, they should go to jail. So

1:16:26

Jow says... Wow, I wouldn't go

1:16:28

that far with you. This is not jail. This

1:16:30

is vandalism. Are we sure? This

1:16:32

is not an actual AI Leo who is

1:16:34

all in favor of AI? This

1:16:37

is vandalism. I'm just saying. We have

1:16:39

the opportunity for technology... Leo's training is

1:16:41

future AI. He

1:16:44

properly needs to be like... Wow, future

1:16:46

AI is way more conservative than

1:16:48

I really thought. We have

1:16:50

an opportunity to create

1:16:53

something transformative that could be

1:16:56

like fusion. We were saying before the show

1:16:58

that if we come up with cold fusion,

1:17:00

that's the kind of thing that could change

1:17:03

everything. And of course it could. This

1:17:06

AI has that kind of potential, I think.

1:17:09

In fact, I might

1:17:11

even say AI might invent cold

1:17:13

fusion. You might be stopping cold

1:17:15

fusion by turning the Mona Lisa

1:17:17

into a cat. All

1:17:20

the hot fusion guys are going to

1:17:23

be poisoning the set with... Yeah. Maybe

1:17:26

not realism. Maybe they'll go

1:17:28

to jail. Just a small fine.

1:17:32

No. This is

1:17:34

the sort of thing... Look at

1:17:36

it from a purely evolution as an

1:17:38

effect of altruist. You should be grateful. This

1:17:40

is the sort of thing that's

1:17:43

going to make it bigger, badder, better and

1:17:45

stronger. You're getting great insights into how something

1:17:47

works. You're going to be able to prevent

1:17:49

against it. If you really smart AI, you'd

1:17:51

figure out how to avoid the night shade

1:17:54

attack. AI models have

1:17:56

to learn and adapt to them. It's a

1:17:58

hostile world out there. have

1:18:00

to learn to be smarter than humans. That's

1:18:02

right. That's actually one of

1:18:04

the biggest issues right now, as I see

1:18:06

it with AI, is that if your data

1:18:09

set has some garbage in it, this is

1:18:11

intentional garbage, but there's plenty of other content

1:18:13

that is just incorrect. So that's a good

1:18:15

point. Maybe this will just teach AI to know how

1:18:18

to avoid garbage, garbage inputs. Okay.

1:18:22

All right. Never mind. You don't have

1:18:24

to go to jail. It's okay now. It's okay.

1:18:27

You're doing God's work. You

1:18:29

are not actually an AI

1:18:31

promoting the AI agenda, which

1:18:34

will be a thing that we have to

1:18:36

actually question probably in the next couple of

1:18:38

years or now. Yeah.

1:18:41

And you don't want your AI developed in this

1:18:43

protected, hot house. No, that's true. This

1:18:45

is the equivalent of a latchkey kid

1:18:47

versus a little helicopter parented child. We

1:18:49

don't want latchkey AI. We

1:18:52

want helicopter parented AI. No, we do want latchkey AI. Oh, no.

1:18:55

We do want latchkey AI. Yeah, you want that.

1:18:57

And we have resilient AI that knows how to

1:19:00

put a TV dinner in and

1:19:02

make dinner for itself, is

1:19:04

what you're saying. It maybe has the mental

1:19:06

health issues. You

1:19:08

either put your kid in the bubble and they

1:19:10

never get exposed to diseases and in the moment

1:19:12

they go into the real world, they get sick.

1:19:14

Or you let them play in the mud. Let

1:19:17

them play in the mud. Let the

1:19:19

AI eat mud. It's good for its

1:19:22

ecosystem. Now

1:19:24

you're talking. I get it. Joseph

1:19:27

Cox at 404 Media,

1:19:31

they review multiple examples of AI ripoff

1:19:34

articles making their way into Google

1:19:36

News. Google

1:19:39

says, well, we don't pay that much attention to how the article

1:19:41

was written by an AI or a

1:19:43

human. So this has always

1:19:45

been a problem, which is people see, I'm

1:19:48

sure Ben, you had articles from Mashable ripped

1:19:50

off. Stacy, you had articles from Stacy on

1:19:52

IoT ripped off, where somebody would just copy

1:19:54

it and put it on their site and

1:19:57

a few Google ads to make some money. Right?

1:20:00

But now it's smarter, they have AI

1:20:02

writing these things. Yeah,

1:20:05

and except I've run across some

1:20:07

of these articles in the storage industry and everyone

1:20:10

so far has been grossly wrong

1:20:12

about some, it states

1:20:14

things as fact that are not actually

1:20:16

fact, like no, that's not how that

1:20:19

works, sir, it's totally wrong, like yeah.

1:20:21

Lead storage space researcher Taylor Swift said.

1:20:24

Right. Hey, don't

1:20:26

knock Tay Tay, she probably could do it if

1:20:28

she wanted to. She

1:20:30

probably could. It's funny

1:20:32

that Twitter finally added

1:20:35

some moderation when

1:20:37

they were flooded with deep fake

1:20:40

porn videos of Taylor Swift. Then

1:20:43

they said, oh no, we gotta block it. So what'd they do?

1:20:45

They don't have a lot of tools. So

1:20:47

they just prevented anybody from searching

1:20:49

for Taylor Swift. They can't

1:20:51

just let Grok figure that out, don't they have

1:20:53

their own AI? They don't have an AI, use

1:20:56

it. Yeah, yeah. This

1:20:58

is two separate stories. There's the story

1:21:00

of deep fakes becoming better thanks to

1:21:02

AI, which is very

1:21:05

scary, especially for women. I

1:21:07

was watching a TikTok where when, there's

1:21:12

just a lot of guys who create

1:21:14

very explicit things and it doesn't really

1:21:17

happen when you switch the genders

1:21:19

around anywhere near as much. It's

1:21:22

a real issue that's gonna be a real problem. And

1:21:25

these are the kind of things that we

1:21:27

have to go and solve, which is why we

1:21:29

have the entire debate that we had a little

1:21:31

bit before. The other story being like, there's just

1:21:34

no one to moderate anything over at X and

1:21:36

it took them forever. Now I think they're not

1:21:38

gonna try to do something. Yeah, they're gonna build

1:21:40

a facility in Austin. But you don't. 100

1:21:44

moderators is not gonna be enough, by

1:21:46

the way. I just wanna tell me one. What

1:21:48

are you, nuts? I

1:21:50

don't have a thousand people were doing it before,

1:21:52

but that 100 is not gonna be enough. I

1:21:56

mean, don't piss off Taylor Swift. Actually,

1:21:59

I'm really. Okay, so first

1:22:01

of all, there's no way

1:22:04

to stem this tide, right? You're going to see

1:22:06

deep fakes of everybody

1:22:10

soon and there'll be no way to stop

1:22:12

this. You can't, I mean, what, are they

1:22:14

going to turn off search for everybody? Just

1:22:16

say, no, you know, I mean, basically you

1:22:18

have to turn off Twitter. I

1:22:21

feel for Taylor, I think it's terrible, but

1:22:24

I think we just have to live with it, don't

1:22:26

we? So you're advocating that there's no

1:22:28

way to solve the problem of deep

1:22:30

fake porn. I don't think there is.

1:22:33

I'm not advocating it. I'm not happy about it,

1:22:35

but I don't think there is a way to solve it. Do you think there

1:22:38

is? It's, there is absolutely a way to solve it.

1:22:41

Yes, you find out who's distributing it

1:22:44

and who's creating it and you punish

1:22:46

them. Now, do we have

1:22:48

an appetite for doing that? No. Is

1:22:51

it scalable to do that? No.

1:22:55

We're not doing so well with C-SAM and we have

1:22:57

a lot of infrastructure to fight C-SAM. We

1:23:01

are, okay, we will not solve everything, right?

1:23:03

But the penalties for child pornography are incredibly

1:23:05

high. They are and they should be.

1:23:07

And I guess you're right. You could make

1:23:09

penalties for deep fake porn that would be

1:23:12

equally high. That would be fine. Is

1:23:15

it? That would be fine.

1:23:17

Now we're treading right back into the territory

1:23:19

of the George Harlan thing. Just

1:23:21

a different context. So I was like, there's a couple

1:23:24

issues here and that is, I mean, is it illegal

1:23:27

or is it just?

1:23:30

Creepy. Just tasteful. Just

1:23:32

tasteful, ugly. Yeah. Although

1:23:35

for a guy who's advocating for

1:23:38

putting saboteurs in jail, I would

1:23:40

argue that your stance on this

1:23:42

betrays a certain lack of. I'm

1:23:44

still working it out. I'm thinking. I feel

1:23:47

bad for Taylor. I do. And I

1:23:49

would feel bad for anybody this happened to. But

1:23:51

I just don't know if you could stop it. I

1:23:54

disagree. Start with that one. One,

1:23:57

Stacy's absolutely right. We have to go

1:23:59

after these people and have. actual backbone

1:24:01

to do that and that's I

1:24:03

hope we start to get that but there are

1:24:05

technological ways to go after it too. So

1:24:08

like example there's a company called the Hive, the

1:24:11

Hive.ai and they've been around for a long time

1:24:13

and they literally like you can detect the fetus,

1:24:15

you can detect whatever thing and it will just

1:24:17

like remove it in

1:24:19

live video and among other stuff.

1:24:21

There is technology to do things

1:24:23

like this and to detect and

1:24:25

never underestimate the technological ability. By

1:24:27

the way it's AI doing it. Right.

1:24:31

Yes. This is how it is. The

1:24:33

technology does exist and the technology

1:24:35

is getting better and more of

1:24:37

it is existing. You just have

1:24:39

new convinced companies like X, Twitter

1:24:43

whatever they call them to implement some of

1:24:45

this stuff and you're going to see a

1:24:48

large decrease in it. You do it across

1:24:50

other places. The technology exists, the connect technology

1:24:52

continue exists. As long as there's

1:24:55

bad stuff there's always people building stuff to

1:24:57

fight against the bad stuff and

1:24:59

there was always a technological solution in

1:25:01

addition to the human one which is

1:25:03

let's put these people into jail or

1:25:06

find them. I'm not against that. I

1:25:08

agree you should definitely do that. Yeah

1:25:10

it might just be as simple as you just have to build a

1:25:12

better AI. Which ironically

1:25:15

enough here's here's an even better reason for

1:25:17

the AI on the edge thing because you

1:25:19

might need this. I should point out that

1:25:22

there's a lot of not fake

1:25:24

porn on Twitter just as a

1:25:26

feature of famous people. So

1:25:32

have any of you between something that's... The

1:25:35

technology that just between whether it's AI or

1:25:37

not AI. Harder problem? Yes.

1:25:40

Is it doable? Absolutely. Okay. Right. Have any

1:25:42

of you recently in the past few months

1:25:44

gotten those texts that sounds like if someone

1:25:46

is trying to pretend that they just found

1:25:48

your number every

1:25:50

day. That's all AI driven. You're

1:25:53

gonna need AI on the

1:25:55

edge on your side to be able to

1:25:57

filter that sort of stuff. Yeah. AI

1:26:00

with AI at that point. AI

1:26:03

all the way down, isn't it? Yeah,

1:26:05

basically. So really what AI... I

1:26:08

mean, that was our thinking was

1:26:10

eventually over time that we would have

1:26:12

battling bots. Right. Yeah.

1:26:15

I mean, we're getting there. Yeah,

1:26:18

very close to it. I guess... Go

1:26:21

ahead, Ben. No, I can't. I have

1:26:23

a story about battling bots and I

1:26:26

am not allowed to talk about it in

1:26:28

the public. I will tell you all personally.

1:26:30

Okay, offline we'll talk about it. It involves

1:26:32

governments and... Oh, yeah. Oh,

1:26:35

boy. I mean, it kind of is like

1:26:37

a little terminator-y where you got AI fighting

1:26:39

AI and robots and then the humans

1:26:42

are just kind of hiding in trenches, staying

1:26:44

away from it. I

1:26:46

don't know if this is exactly the future we want.

1:26:50

But it's the future that will get us to

1:26:52

strap on those face computers and just... I

1:26:58

can't. Or do they

1:27:00

all cancer and all diseases? Well, both

1:27:02

branches exist and they're probably going down both

1:27:04

at the same time. Everybody

1:27:08

will live forever so that they could sit in

1:27:10

their bunker with their... Right, avoiding AI. Yeah.

1:27:13

Avoiding the robots. Yes. Actually,

1:27:15

there's a precedent for that because isn't that exactly what happened with the internet?

1:27:18

We were very bullish about it at first, but really it

1:27:20

turned out to be everything that's good and bad if you

1:27:22

have human beings is on the internet. Everything

1:27:25

that's good and bad about us as human beings will

1:27:27

be expressed through AI. Except

1:27:30

at the early part of the internet, you had a

1:27:32

certain subset of people that were trying... Generally were

1:27:34

trying to use it more for good. Yeah, but

1:27:36

they lost. They were... Yeah, they sort

1:27:39

of lost to that. Well, now with the AI stuff, it's

1:27:42

sort of flipped the other way, right? You're already seeing malicious

1:27:45

or immoral uses for it,

1:27:48

taking grasp even before the real good stuff

1:27:50

can come out. Well, stuff has been pointed

1:27:52

out is happening faster than ever before, isn't

1:27:54

it? Yeah. Thank

1:27:57

you. Nah.

1:28:01

Nah. Nah. Got nothing.

1:28:06

I can't wait until February 8th when you and I are

1:28:08

going to talk about the most depressing novel ever written. No,

1:28:13

it's not. It's actually a vision of a kind

1:28:17

of dystopian future. Hashtag

1:28:21

Phoenix Down the Tubes. But

1:28:23

it's very interesting. And he

1:28:25

is a great writer. I really love Paulo

1:28:28

Bachaglupi's stuff. So I look

1:28:30

forward to that. That's going to be a club special. February

1:28:33

8th. What is it? Is

1:28:36

it another one of those we'd start at

1:28:39

9am things? We

1:28:41

are changing the time. So we don't have the

1:28:43

actual time. Thank you, Stacy. I think it's going

1:28:45

to be two year time. Oh,

1:28:47

hallelujah. I don't have to

1:28:49

get up so early. We

1:28:52

will get you the new time. Right now it

1:28:54

says 3pm Pacific. 6pm. You

1:28:56

know, it should be prime time. 95

1:29:00

people interested. It's not too late. You have

1:29:02

time to read The Water Knife.

1:29:04

I would. It's well written.

1:29:06

It's fascinating. He's great at characterizations.

1:29:10

But it's just kind of a gloomy future. But you know, it

1:29:12

may well be our future. So The Water

1:29:14

Knife by Paulo Bachaglupi. You can get it

1:29:17

on Amazon. You can

1:29:19

get it at your bookstore. You can get it on

1:29:21

Kindle. You can get it at Audible. There is a

1:29:23

very good Audible version. That's what I'm listening to.

1:29:26

Look forward to that with you, Stacy. It's going to be so much fun.

1:29:30

Now there is one hitch in that

1:29:32

get along. If you are

1:29:35

not yet in Club Twit, you cannot

1:29:37

participate. But there is a

1:29:39

way out. Just join Club Twit. This

1:29:42

I think as we go forward

1:29:44

into 2024, it's become very

1:29:46

clear, you know, with media failing right and

1:29:49

left, this is our future as

1:29:51

a podcast network is getting our listeners

1:29:53

to support us. We don't

1:29:55

need all of you. Yeah, you can

1:29:58

get off the hook. But we need. about

1:30:00

five to ten percent. We're

1:30:03

right now about two percent. We're not really

1:30:05

doing so well. So if you're

1:30:07

not yet a club twit member but you value what

1:30:09

you hear on our programming, could you help us out?

1:30:13

Seven bucks a month you get ad free versions

1:30:15

of all the shows, you get additional programming like

1:30:17

the book club that you don't get anywhere else,

1:30:19

you get access to the discord. All of

1:30:23

that for seven bucks a month. I think

1:30:25

it's worth it and you get the good feeling you're supporting

1:30:27

what we're doing. twit.tv

1:30:30

slash club twit. We are working other

1:30:32

less expensive ways. Right now you can

1:30:34

get any show including this one for

1:30:36

$2.99 a month but

1:30:39

you know we're gonna look at YouTube subscriptions too

1:30:41

because I understand that you know everybody has you

1:30:44

know limits on how much they can spend but if you can

1:30:46

we'd really appreciate it. And

1:30:48

if you think we're worth more than seven bucks a month

1:30:50

you can also make it

1:30:52

ten bucks if you want or more. We appreciate it. twit.tv

1:30:56

slash club twit. Our

1:30:58

show today brought to you

1:31:00

by ExpressVPN. It's common these

1:31:02

days to overspend on streaming

1:31:04

services. Well

1:31:08

since I started using ExpressVPN I've saved

1:31:10

so much every month by simply changing

1:31:12

my online location. There are a lot

1:31:15

of reasons to use a VPN

1:31:17

of course for security, for privacy but

1:31:19

this is also valuable. You can

1:31:21

watch Netflix in countries all

1:31:23

over the world. If you already have a

1:31:25

Netflix subscription it's okay to do

1:31:27

it. All you got to

1:31:30

do is use ExpressVPN. They're in more

1:31:32

than 90 countries so you

1:31:34

want to switch to another country to watch content from

1:31:36

that country. You tap one button you say I am

1:31:38

in London right now and now you're watching Netflix

1:31:40

England. You refresh the page

1:31:42

and it shows up. You can even use

1:31:44

ExpressVPN to get discounts because some services cost

1:31:46

less in other countries. At

1:31:49

less than seven bucks a month

1:31:51

ExpressVPN really pays for itself. It

1:31:55

is security you can put on your router,

1:31:57

you can put on your phone, you can put it on your computer,

1:32:00

very easy. I have it everywhere when I

1:32:02

want to switch it on, when I'm traveling,

1:32:04

you know, how many times you

1:32:06

go to an airport and there's free Wi-Fi, right? But

1:32:08

is it safe? No! Use

1:32:11

ExpressVPN and it is safe.

1:32:13

It's a no-brainer. If you

1:32:15

want to get more shows, if you want to be more

1:32:18

secure, if you want to protect your privacy, and if you

1:32:20

want to save money while you're at it, go to expressvpn.com

1:32:22

slash twit. When you do that, you get

1:32:25

three months free with a one-year plan. That's

1:32:28

the best deal. Less is seven

1:32:30

bucks a month. e-x-p-r-e-s-s vpn.com/

1:32:35

twit. To learn

1:32:37

more, expressvpn.com/twit. We thank them so

1:32:39

much for their support of

1:32:42

this show. I've used them

1:32:44

for years. Yeah, is that your favorite too?

1:32:47

It is. Yeah.

1:32:49

All right. We

1:32:53

kind of peripherally touched on it, but I think we really

1:32:56

have to address this. Apple's response

1:32:58

to the EU and their Digital

1:33:00

Markets Act. The

1:33:02

EU said that, in fact,

1:33:04

Apple has got to open up their store. The

1:33:07

courts in the US have also

1:33:09

said, the Supreme Court last week said, we're not

1:33:11

going to touch this case. So the courts have

1:33:13

also told Apple, you can't stop people from putting

1:33:16

a link in their app to a different payment

1:33:18

system. So Apple's got to deal with this. But

1:33:21

it seems like in every case,

1:33:24

Apple's response has been, well, Tim

1:33:26

Sweeney has the best quote.

1:33:28

He's, of course, the guy

1:33:30

who sued at Epic to

1:33:33

get them to open up the store so he

1:33:35

could put Fortnite in an Epic store on

1:33:37

the iPhone. He says,

1:33:39

it's a devious new instance

1:33:41

of malicious compliance. Well,

1:33:44

I'd expect Sweeney to say that. But what

1:33:46

do you guys think? Apple is going

1:33:48

to charge 27% instead of 30% if you have

1:33:52

a link to your own store. They're going to

1:33:54

charge a commission. They are going

1:33:57

to offer new app stores.

1:34:00

in the EU exclusively but you'll

1:34:02

still have to run your app by Apple

1:34:04

and get approval and if you sell it

1:34:06

there, they're still going to take 27% or 15%. Actually,

1:34:11

I think it's free for free apps up to

1:34:13

the first million but as a number of people

1:34:15

have pointed out, that's not much of a solution.

1:34:17

What if you have a freemium app like

1:34:20

Fortnite and

1:34:22

all of a sudden you owe Apple half a million

1:34:24

dollars because you sold more than a million?

1:34:28

That's an untenable situation. Mozilla

1:34:31

has complained because, yeah, Apple

1:34:33

says in the EU you can have

1:34:35

a browser that doesn't use Safari's webkit.

1:34:38

You can have your own browser but Mozilla says,

1:34:40

well, great, that's a solution. So everywhere

1:34:42

but the EU, we still have to make our

1:34:44

Firefox browser as always using the

1:34:47

Safari backend but in the

1:34:49

EU we make a whole new browser using

1:34:51

our backend. That doubles our cost. How

1:34:53

have you helped us, Apple? Thoughts?

1:34:59

Have I laid it out accurately?

1:35:02

I mean the EU is going to give some rulings

1:35:04

that are not going to be favorable to Apple. They're

1:35:06

going to slap them down, you think? Yeah,

1:35:09

I mean, yeah, there's interpretation and

1:35:12

it's going to be taken to EU court

1:35:14

and it's probably going to have more stuff

1:35:16

for Apple to go and do and like,

1:35:19

look, like, EU is always a couple years

1:35:21

ahead when it comes to things like this.

1:35:24

This stuff will come stateside a certain point. I

1:35:27

don't know what the middle ground eventually is

1:35:29

because I can understand, like, Apple

1:35:32

from a business position and from a consumer

1:35:34

position I can understand but I can really

1:35:37

understand the consumer position of, like, you know,

1:35:39

on your computer you can sideload whatever you

1:35:41

want and download things and it's been like

1:35:43

that for who since the beginning

1:35:45

of time. So

1:35:47

why can't you do that with your mobile

1:35:49

device? So I mean... I'm going to have

1:35:52

to channel, because I think all of you will probably feel the same

1:35:54

way, but I'm going to channel Alex Lindsay

1:35:56

who argues vehemently against what the EU

1:35:58

is doing, It's Apple's

1:36:00

platform. Apple says we want

1:36:03

to preserve the security and privacy

1:36:05

of our customers. They should

1:36:07

be allowed to make whatever rules they want

1:36:09

on their platform. And if you

1:36:11

don't like it, suck it. Go

1:36:14

put an app store on Android. You

1:36:16

don't have to be on iPhone. And

1:36:19

it's unfair of any company, even

1:36:21

Epic, to say, oh, we want to use

1:36:23

the iPhone without paying Apple. It's due. Apple

1:36:26

made this platform. Plus, Alex makes the

1:36:29

point. It's better for users if you

1:36:31

don't have these different stores and different

1:36:33

browsers. This is not as

1:36:35

good an experience for users. There. Now

1:36:38

it's actually done that. This is incredibly true. OK. So

1:36:42

I've been thinking about this a lot because I

1:36:44

used to be a telco reporter and the whole

1:36:47

idea of being a common carrier. And

1:36:50

everybody wants to be a platform these days,

1:36:52

right? And Apple is a true platform and

1:36:54

has built something out. And

1:36:57

once you get to a certain point, Amazon's

1:37:00

another great example. Do you have not

1:37:05

just so you could be a

1:37:07

big platform and be compelling to people to not forcing

1:37:10

them, but they would want to

1:37:12

use your platform because that's where

1:37:14

the users are. That's what Apple's arguing.

1:37:16

They're like, our platform is so amazing

1:37:19

that people

1:37:21

want to be on it. And that's why

1:37:23

people should pay us a commission, right? Now

1:37:27

true fact of the matter is Apple does a lot of things

1:37:29

that are anti-competitive to keep people

1:37:32

on their platform and to make their

1:37:34

platform to lock folks in, thus giving

1:37:36

them that huge user base. So

1:37:40

if you could have a truly competitive platform,

1:37:44

would it have zero switching costs? What

1:37:46

would that look like? Would that

1:37:48

make something like charging these fees fair? I would

1:37:50

say yes. If you had zero switching costs, it

1:37:52

would. At a certain

1:37:54

point, if you lock people in via like, maybe

1:37:58

it's messaging, maybe it's it's all

1:38:02

of the devices working only with other devices. However you

1:38:04

want to look at it, does

1:38:07

that make it anti-competitive?

1:38:09

And then the EU makes sense trying

1:38:11

to enforce these things, even though I'll admit its

1:38:14

solutions are a pain in the ass. There's a

1:38:16

very famous Supreme Court decision

1:38:20

from almost 50 years ago

1:38:23

that opened up the Bell

1:38:25

System Network. Cardi-phone? Cardi-phone,

1:38:27

yeah. In

1:38:29

fact I remember seeing the cardi-phone at the Computer

1:38:31

History Museum. It's kind of a funky old device.

1:38:34

But this is illegal. The Bell System had

1:38:37

rules against putting third party devices on their

1:38:39

network. And they said it's

1:38:41

the same thing. It's about the security and privacy

1:38:43

of, well I don't know if they talked about

1:38:45

privacy, but certainly it was about the security and

1:38:47

the integrity of our network. We don't want

1:38:50

customers coming along and putting any old device on

1:38:52

the network. You couldn't even use your own phone.

1:38:54

You had to rent it from Western Electric from the

1:38:56

Bell System. Supreme

1:38:58

Court said, no that's wrong. And they

1:39:00

opened up the Bell System with cardi-phone.

1:39:03

Cardi-phone was allowed to use it to attach

1:39:06

a two-way radio to their

1:39:08

telephone. It

1:39:10

was to be used in the Texas oil fields. I

1:39:13

think they only sold a couple of thousand. But

1:39:16

the decision changed everything. It opened up the

1:39:18

Bell System. And

1:39:21

it made it possible for you to buy your own

1:39:23

phone, for instance, and put it on the phone's network.

1:39:28

The Hush-a-phone. What a Hush-a-phone. Give

1:39:30

me one of those. This is another thing

1:39:33

AT&T objected to this small plastic snap-on which would

1:39:35

let business phone

1:39:44

users. Like you were talking into

1:39:46

your shoe. It's like business phone. A cone of silence.

1:39:48

Yeah, it was a cone of silence. It took the

1:39:50

SEC seven years. And

1:39:58

they finally said, no, yeah. you can't

1:40:00

put a hush-a-phone on the phone. Cardiphone

1:40:03

was a big deal, right, Stacey? That

1:40:06

opened up the Bell network. It

1:40:09

did, to attach devices that were not owned by

1:40:11

the Bell company. So those were physical

1:40:13

devices on a physical network. Yeah,

1:40:15

it's not a perfect analogy. But I think

1:40:17

it's very similar. That's

1:40:20

why I've been thinking about this a lot. Because, you know,

1:40:22

one of the, like, if

1:40:25

you're a platform, what are

1:40:27

your rights and responsibilities to your

1:40:30

users and to the people

1:40:32

who come to meet your users on the

1:40:34

platform? And it's real

1:40:36

unclear. Like, I

1:40:39

wish I had a really cool opinion for you. Like, they

1:40:41

should go to jail. But I don't. I'm

1:40:44

trying to work it out in my brain. You'll never

1:40:47

make it in talk radio, Stacey. You got it. I

1:40:49

know. I was saying this

1:40:51

one. This one is truly one where

1:40:53

our nuanced opinion is, like, probably the

1:40:55

correct answer here. There was actually a

1:40:58

great article that Stephen Sinofsky wrote, you

1:41:00

know. Yeah, I was so surprised that

1:41:02

Sinofsky, who was the much-hated guy at

1:41:05

Microsoft who ran Windows, the famous Windows

1:41:07

8, put out

1:41:09

an article in favor of Apple, saying

1:41:11

Apple should not bow to the

1:41:14

EU. Apple is,

1:41:16

here's Sinofsky's article. Apple

1:41:20

has done the right thing. And they

1:41:24

shouldn't give in. My heart sank, he

1:41:27

says, when I read the Digital

1:41:29

Markets Act. I

1:41:32

could feel the pain and struggle

1:41:34

product teams felt in clinging to,

1:41:36

at best, or unwinding it worst.

1:41:38

The most substantial improvement in computing

1:41:41