0:00

It's time for to make this

0:02

week in Google. Will talk about

0:05

a I of course both both

0:07

a good and bad and the

0:09

ugly. Googles gotta seven hundred million

0:11

dollar fine and to pay and

0:13

Is Cox marketing Actually listening to

0:16

your private conversations will answer that

0:18

and a lot more coming up.

0:20

Next on the last twig of

0:22

Twenty Twenty Three. Podcasts

0:27

you love from people you

0:29

trust. This

0:31

is tweet. This.

0:38

Is Twig this weekend Google

0:40

Episode Seven Hundred Forty Seven

0:42

Recorded Wednesday, December Twentieth. Twenty

0:45

Twenty Three. In an abundance

0:47

of caution, There's.

0:49

We get Google is brought to

0:51

you by Secure my emails year.

0:53

My email provides easy encryption for

0:55

your current personal and business email

0:58

addresses. Set up only takes minutes.

1:00

Start your free account or enjoy

1:02

a thirty day free trial of

1:04

a premium plan no payment info

1:06

require and they have a special

1:08

offer for Twit listeners. Visit Secure

1:11

My email.com/to It and years ago

1:13

Twitter Checker an by our friends

1:15

and I T Pro Tv now

1:17

called a See I Learning. A

1:19

see as new cyber skills as training

1:22

for everyone, not just the pros. Visit:

1:24

Go Down Easier learning.com/to it. As a

1:26

twit listener you can get up to

1:28

sixty five percent off and Id Pro

1:31

Enterprise Solution Blame is complete. The farm

1:33

is based on your team size. Find

1:35

out how much you can save it.

1:37

Go Daddy! See I learning.com/to It It's

1:40

time for doing this. We get Google

1:42

featuring. The. Most attractive, festive. His

1:44

paw I think we've ever had. I

1:47

mean for a little man poll, it really does.

1:50

Does. It. Brings. a whole

1:52

house together ah hello

1:54

there paris mart know

1:56

merry christmas know they're

1:58

happy holidays Is it

2:00

okay to say Merry Christmas? I feel like I want

2:02

to say that. I don't know why. You

2:05

can say whatever you want. Happy Hanukkah, happy

2:07

Kwanzaa. Merry Christmas, happy holidays. Doesn't matter to

2:09

me. Happy Festivus,

2:11

Paul. Festivus. The

2:13

solstice is tomorrow, shortest day of

2:15

the year. So it's a good

2:17

day to light a

2:20

candle and curse the darkness or something.

2:23

Happy holidays. I do that every day. Yes. It

2:26

is far better to light a candle than

2:28

to curse the darkness, said somebody

2:31

once on a bumper sticker. And Mr.

2:33

Jeff Jarvis, who celebrates

2:36

his final hours

2:39

as the director of the Town Light

2:42

Center for Entrepreneurial Journalism at the Craig

2:45

Newmark Graduate School of Journalism at the City

2:47

University of New York. Are

2:49

we retiring the singers after this show? And

2:52

boy do I have grievances. That's great emphasis. Well,

2:57

I'm basically retired, but I don't

2:59

actually become emeritus until next August.

3:02

That's just a technicality. You

3:04

ain't working. You know more classes is the bottom line,

3:07

right? You don't have to work. Well, I might teach

3:09

some in the executive program, but yeah, no more classes

3:11

there. Are you going to

3:13

miss the students? I bet that was nice to

3:15

come in and see their fresh young faces excited

3:17

and eager to become journalists. There

3:19

were exceptions occasionally, but yes, all in

3:22

all, yes. All

3:24

right. He's young, as they say. Did

3:26

I see, Jeff, that you had a going away party

3:28

where all of your students came and told you

3:30

how great you are? Our

3:32

engagement and lums threw a little

3:34

dew for me, yes. Oh,

3:37

that's nice. Was it on Twitter or

3:39

where did you put it? There were

3:41

pictures. On Facebook. On Facebook. Good Lord.

3:43

I have to load Facebook. Definitely

3:47

not just on Facebook, because I saw it somehow.

3:49

Yeah, I put it on Twitter too, but he

3:52

is going to go back and back and back

3:54

and back and back. Yeah, he's

3:56

been tweeting. He's so active on the Twitter.

4:00

scrolling scrolling Do

4:03

you find it probably go to the media tab? Yeah,

4:05

you guys down there four days ago. Oh I'm

4:09

still in hours one is called what is

4:11

called another tough farewell. Oh You'll

4:14

see that If

4:16

you go down Then

4:18

we'll use you down Only

4:21

four days We

4:23

went on average 30 times. I

4:25

was doing a Facebook go to Facebook. Oh, no, I'm looking

4:27

at Here's one one

4:30

celebrated Jeff Jarvis's academic legacy

4:32

at CUNY That's

4:35

my colleague and former student just said he had to

4:38

boy. Yeah, Correa, but you would boys you look at

4:40

is more Okay You

4:44

just got into Facebook it'll be over now That's

4:49

not how Facebook works Jeff I'm sorry You

4:53

just it'd be over by now Leo

4:55

didn't want to click on Facebook Look

4:59

at all this stuff Look at

5:01

all it. Yeah. All right, you want to see my Facebook?

5:03

Let's see if it's cleaned up I

5:05

have been very aggressively deleting

5:09

Accounts that show me. Oh, I'm not even logged

5:11

in That's

5:13

how long it's been since I since

5:15

I use this thing. All right, I

5:18

still have it. I didn't kill the account Enter

5:21

your six-digit code. Oh They

5:24

want my Alright guys, I've

5:27

already found it and I'm gonna post it

5:29

in discourse. Oh, yeah, Patrick also just posted

5:31

it I Feel

5:33

like I am Now there we

5:35

go the old man you are how

5:37

does it now? How

5:42

does Facebook work Is

5:44

it is it now? Okay now

5:46

I'm on Facebook. Let's see here. Yeah,

5:49

remember my password. Thank you. All right

5:51

There's somebody I know. Oh Here's

5:54

Paul Theron Mary Joe Foley having oh, there you

5:56

go. It is like the very

5:58

first thing in my Facebook

6:00

feed look at you and

6:02

and you're in a bar and you're

6:05

having fun Is this where George Washington

6:07

got inaugurated this part? That's

6:11

launches tavern spot. Yeah, it looks

6:13

pretty cool Very

6:15

nice. Well, you guys know one more you'll

6:17

find another farewell pictures. It's not something. Yeah.

6:20

Oh, I have to go to you That's the problem

6:23

has to go to you But

6:26

you know what it does look like I have eliminated

6:29

all of the Bikini

6:31

shots, which is good. Oh, is this

6:34

you? Let's be where

6:36

these people I tweeted that I

6:39

Facebook. Oh god. I don't like this

6:41

at all Pain.

6:43

All right enough. I'll leave this as an

6:46

exercise But while you were

6:48

there you went by I didn't get a chance

6:50

to plug it last week because it wasn't done

6:52

yet The audio now my audio book is on

6:54

yay Oh, well, I'm

6:56

gonna have on the now here my

6:58

dulcet tones talking very slowly In

7:03

they got that out fast audio book. There.

7:05

Yeah, they did. Should we listen? You can

7:07

do a little sample? Here

7:10

we go. He texts were identical consistent

7:13

No longer subject to the idiosyncratic

7:16

edits amendments wins and errors of

7:18

scribes. No, this is good That's

7:21

exactly the right tempo. I Think

7:23

so, especially since most people listen at one and a half

7:25

to two. Well, yeah, you're gonna speed me up It's

7:28

interesting. They've done something in the editing

7:30

to where it sounds more Mechanical

7:33

than your actual speaking voice like it

7:35

sounds a little robotic because

7:37

I'm sure they edited it to

7:39

remove the spaces or something in between your words

7:41

or make They do some of that something. Yeah,

7:43

we're in my case add spaces between the words.

7:45

Does it really do that? They actually Mess

7:48

with it that much. I don't know

7:51

whether they do or not I don't know

7:53

but it sounds like there's something going on

7:55

there It's different than how Jeff normally sounds

7:57

when I first did an audio book When

8:00

I screwed up, they'd just say, keep going

8:02

and do it again. And then they would edit it

8:04

no more. Now when

8:06

you screw up, they back it up three

8:08

or four words, start it again. And

8:10

then you've got to pick up right at the moment.

8:12

Oh no. It sounds like a pain in the butt.

8:15

That's just lazy on their part. That's

8:18

standard now. Yeah. Absolutely standard. And you said,

8:20

is this your, is this

8:22

your favorite student? This guy from Maine.

8:25

He doesn't, he looks a little, a little

8:27

stiff. Jeff

8:31

is hugging a mannequin in a

8:33

photo for those. Newmark, J.

8:35

School teacher. Oh, look. And now I've got

8:38

Monica V. That's an ad. That's not me.

8:40

And he's now getting an ad for a scantily clad

8:42

woman. Yeah. Well, there's not as many as there used

8:44

to be. I think it's really, well, there's another one.

8:47

Well, there's still the one. Oh, this is a group

8:49

I really want to join. Really

8:51

want to join that one. That's a group I

8:53

want to be part. Why do they think I

8:56

am in, I am not a king of the

8:58

golf course. I don't, I don't

9:00

want to see this. I don't anyway.

9:02

Okay. Fine. Most, I have

9:04

to say it's gotten, I've gotten rid of

9:06

a lot of lovers, sponsored content for you.

9:08

Are you a trumpet lover? Well, here it

9:11

is. La Fio Shiver. Okay.

9:20

Wow. Um, boy, Facebook,

9:22

just the gift that keeps on giving.

9:25

I must say. I'm,

9:28

if you're into jazz trumpet, the talented

9:30

Jeff Jarvis is a jazz musician. Oh,

9:32

there's another Jeff Jarvis. There's

9:35

a Jeff Jarvis is a, is if you Google,

9:38

there's also a very talented pianist. That's a Benito Gonzalez.

9:40

That's a jazz musician too. Maybe they should form a,

9:42

maybe if there were only the others, they'll call it.

9:46

Yeah. There isn't like a Paris Martineau who plays

9:48

a piano. We have a whole, we

9:50

gotta get, you know, it can't be me. You

9:53

gotta get the one else out there. There's a Martine Paris out

9:55

there who's also a tech. journalist.

10:00

Maybe she can play the piano. You have a

10:02

good unique voice. I mean, name. Your

10:05

name is... I'll take both compliments. Both.

10:08

Google has to pay some

10:11

money to 38 states

10:13

attorneys general and the District

10:15

of Columbia. I

10:18

love that. To 102 million people,

10:20

we'll each get less than $7.00.

10:22

Yeah. But Google, it's almost

10:24

a billion, $700 million in

10:27

the Play Store settlement.

10:29

They're settling a claim

10:31

from all of those states that

10:34

Google operated its app store as

10:36

an illegal monopoly stifling competition from

10:39

other app distributors and devices using

10:41

the Android operating system. It's

10:43

funny because Google allows sideloading.

10:45

I really still don't get this.

10:48

This is, of course, immediately following their

10:50

loss in a different case with Epic

10:52

Games. That will

10:54

be appealed, but they have settled this one. This

10:57

is the one they say, yeah, we'll pay that.

10:59

And they're changing behavior here, Leo. What I wanted

11:01

to ask you is, does that have any impact

11:03

on Epic or vice versa?

11:06

What was the behavior that they got in trouble for

11:08

at Epic? Was that the same thing? It's the

11:11

same thing. They were a monopoly in

11:14

the store and the jury there agreed.

11:16

Now, under appeal, it goes

11:18

to judges and justices who

11:20

probably read the Wall Street Journal, but I

11:22

don't know how much they're supposed

11:24

to be influenced by that. So it's unknown whether

11:29

that'll make a difference or not, just because

11:31

they settled with the FTC. They, I'm

11:33

sure, agreed to no wrongdoing. Never

11:36

folks agreed to wrongdoing. Yeah,

11:40

never do that. Never, never. I

11:42

mean, in 2021, I think when

11:45

this suit first came into

11:47

existence, the South Korean government had

11:49

also passed a law forcing

11:52

Google and Apple to allow app

11:54

makers to charge customers directly and

11:57

they complied with that. And since then, Google

11:59

has offered alternatives. billing options in

12:01

South Korea and it

12:03

also preemptively introduced a pilot program

12:05

in the US to do

12:08

something similar, give users a choice in how

12:10

they're billed in the US. This is before

12:12

the settlement even took place. So I really

12:14

kind of laid the groundwork. Yeah I mean

12:17

I get it I don't understand why Apple's

12:19

so adamant about not they're

12:21

gonna be forced to let other app stores on

12:23

there which is exactly you know that I would

12:25

agree is a bad idea because you know they

12:27

could have bad software pirate software that kind of

12:30

thing on there. Whereas

12:32

if they would just open up loosen up a little

12:34

bit on the payments thing maybe give people a better

12:37

deal I don't know you know they

12:39

do secret deals with Spotify and Amazon

12:43

as has Google I don't know I it

12:46

just does not seem to be the

12:48

number one crisis in our nation at

12:50

the moment. You don't say. Yeah

12:52

so speaking

12:57

of which the

13:00

the folks at

13:02

CMG the Cox

13:05

marketing group kind of got caught out

13:08

by 404 media.co

13:12

maybe. Is this the

13:15

big story you were working on Paris?

13:19

Nope. She knows real stories. CMG. Oh okay

13:22

so we spent some time talking

13:28

about this on both twit and security

13:30

now CMG is a

13:32

division of Cox the big internet

13:34

service provider and cable company and

13:37

they had the Atlanta Journal Constitution. Oh

13:39

really I did not know that. They

13:42

had some pages which has since been

13:44

taken down but are visible on the

13:46

Internet Archive. Aimed

13:48

not at us not you and me but aimed at

13:51

advertising buyers saying have you ever

13:53

wanted to know what people

13:55

are saying in the privacy of their own home

13:58

well now you can. active

14:01

listening and they claim

14:03

that they can

14:06

identify customers based on quote

14:08

casual conversations in

14:10

real time through

14:13

smartphones, smart TVs. They

14:17

don't say exactly what devices.

14:19

Yeah. They apply Amazon echos

14:21

and Google Nest Hubs but I

14:24

can tell you I'm sure that that's not happening

14:26

because if it were we would see that traffic

14:28

being sent out from your house, you know, all

14:30

of the sound. The

14:33

company says it's a marketing technique

14:35

fit for the future available today.

14:37

They even have, which is hysterical,

14:39

a disclaimer

14:42

to their overview in

14:44

their FAQ that says

14:46

is this legal? We

14:50

know what you're thinking. Is this even

14:52

legal? The short answer is yes.

14:55

It is legal for phones and devices to listen

14:57

to you when a new app

14:59

download or update prompts consumers with a

15:02

multi-page terms of user agreement

15:04

somewhere in the fine print active

15:06

listening is often included. It's such...

15:08

Well that makes it okay. Yeah.

15:10

They don't know we're listening but

15:12

they said it was okay. They

15:14

agreed. They don't

15:17

they don't say exactly, you know,

15:19

how it works or who they're

15:21

listening in on. They do, you

15:23

know, claim partners with

15:25

some of the biggest companies including

15:27

Microsoft and Amazon. Okay. I'm curious

15:32

what is your thought on this?

15:35

Do you think that after reading this

15:37

article that this actually exists in the

15:39

way it's being described? First of all,

15:42

I think it's important to note that marketers

15:44

never lie. They've never said anything

15:47

inaccurate ever. You're an honest group of

15:49

people. So, you know, this is coming

15:51

from a marketing group probably

15:54

written by, you know, some young guy

15:56

who doesn't really... Probably written by chat,

15:58

GPC. And

16:00

I think we know that some of the

16:02

capabilities they claim don't exist,

16:04

but we also know for a fact

16:06

that TVs, you know, my TV has

16:08

a camera and a microphone built into

16:10

it. They say, well, that's

16:12

so you can zoom from it. Nobody's ever zoomed from

16:15

it. There are other reasons they

16:17

may have a microphone and a camera

16:19

on my TV. Your phone, if you

16:21

have an iPhone, if the microphone's turned on,

16:23

we'll have a little red light. Same thing

16:25

with the camera. Android device is the same thing. And

16:28

every expert who's looked at these devices

16:30

say it's impossible to bypass

16:32

that. So I don't think your phone's

16:34

listening to you. I think your

16:36

TV might. We saw that Mozilla

16:39

report on automobiles that say that

16:41

all the big manufacturers not only

16:43

collect information from all of the

16:45

listening devices in your car, but

16:48

actively sell it to data brokers and others,

16:51

including CMG. CMG did

16:53

pull all of this down and

16:55

then later said, we aren't doing it. We are partnered

16:57

with other people who are doing it. We just are

17:00

aggregating the anonymized information from

17:02

these other people that

17:04

are doing it. I think they're overselling

17:07

their capabilities. But the most important point

17:09

to be made is they don't need

17:11

to do this because you're telling them

17:13

every time you go out on the internet.

17:16

Exactly. You

17:18

go inefficient. Yes. So the example they

17:20

give, which really cracked me

17:22

up, are, you know, conversations

17:25

like, boy, we really ought

17:27

to do something about the mold in our house.

17:29

Or gee, my lease expires in two

17:31

months. I wonder if I should look at a

17:33

new car. You

17:36

know, much easier to just look at your web browsing

17:38

history and say, oh, he's been looking at a car

17:40

dealers. We can tell which

17:42

one. And since

17:44

it is Cox, who

17:46

is in fact an ISP and

17:49

almost certainly is gathering that information. It's completely

17:51

legal for ISPs to do this. Can

17:53

we talk last week about Xfinity? No,

17:56

no, it wasn't Xfinity. It was

17:59

another phone. people

20:00

say, I'm turning off, I'm throwing

20:02

out my Amazon Echo. I wouldn't

20:05

worry about your Amazon Echo. And I wouldn't worry about

20:08

your phone because you'll know if the phone microphone is

20:10

on. You're just sending those signals all

20:12

the time. Do I have Facebook on my phone? Does it

20:14

know everywhere I go? Yes. They're

20:16

trying to act like they're smart. Marketers

20:19

are just all the time trying to say that

20:21

they can do more than they can do. I'm

20:23

starting to say that Cambridge Analytica acted like they

20:25

were so damn smart and every researcher I know

20:27

says it was complete BS from the beginning. They

20:29

never could do what they insisted in their marketing

20:31

material. We

20:35

do magic with ads and find them when they want

20:37

to buy the stuff. Yeah, and sales is what they're

20:39

doing. I think that probably the most realistic

20:41

thing that is happening here, if we want to

20:44

believe that any part of this is

20:46

true, is that maybe they

20:48

bought a bunch of

20:50

data from some third party broker that

20:52

consists of ad keywords

20:55

or something assigned to a

20:57

user advertising ID that came

20:59

from, I don't know, your

21:01

off-brand air conditioning

21:03

unit that has voice control that you

21:06

signed up, didn't really think anything

21:08

of, clicked the I agree with the

21:10

terms and conditions and it was collecting

21:13

that. It's some probably third party device

21:15

in people's homes that might be collecting

21:17

a small amount of data whenever a

21:19

voice prompt is used. It's

21:22

not even in perhaps the

21:24

most best case scenario for this company's

21:26

claims. It's going to be a fraction of

21:28

what they're describing. So

21:31

should we worry? No. More

21:34

things to worry about. Lots of other things to worry

21:36

about. I do like this

21:38

particular paragraph. Don't leave money on the

21:40

table. Claim your territory now.

21:43

Our technology provides a process that makes

21:45

it possible to know exactly when someone

21:47

is in the market for your services

21:49

in real time, giving you a significant

21:51

advantage over your competitors. So it's geolocated.

21:53

Territories are available in 10 to 20

21:55

mile radiuses. So you click

21:58

this button and you will claim your territory. It's

22:00

a sales pitch. They're

22:04

overstating their claims. Not

22:08

that people shouldn't worry about their privacy, but you're giving

22:10

that information up all the time in a variety of

22:12

ways. I don't know why people

22:14

are so worried about the microphone listening to them. I

22:17

mean, I think it is creepy. Because I think

22:19

it is the most tangible version

22:22

of a privacy invasion that someone could think of.

22:25

When you're thinking of privacy, you think,

22:27

oh, there's someone with their ear against

22:29

my door listening in. You don't think

22:31

of the fact that everything you do

22:33

online is way more

22:36

invasive. That's why when

22:38

Microsoft has those ads about Gmail man,

22:40

or lately I've seen somebody else, a

22:42

VPN doing it, it's always a person

22:45

looking over your shoulder at your

22:48

Gmail. And that's not exactly

22:50

what's happening at all. The

22:53

other thing is, as Paris was saying, there are

22:55

data brokers out there that have huge data. I've

22:58

seen that many times on the show. This

23:02

year, the FTC went after

23:04

Kochava for

23:06

staggering sales of consumer

23:08

data collected from mobile apps, revealing

23:11

location, revealing other things. And

23:13

so that's what these marketing firms will

23:15

say, well, we know so much. You should see what we can

23:18

do. And they're just trying to

23:20

improve their odds by 1%. Now,

23:24

this is valuable, because there's two things I

23:26

think will be worth watching. One,

23:28

when are they going to get the letter from Ron Wyden? Because

23:32

you know they will. Or

23:34

there'll be a letter from a senator saying,

23:37

well, what can you do and

23:39

what's going on here? But

23:42

for those who say, oh, maybe

23:44

this will stimulate some sort of

23:46

legislation from Congress to protect

23:48

our privacy, to cut back on data

23:50

brokers. I would point out, we've known

23:53

about this data broker business. We've

23:55

known about the innovations of privacy for

23:57

years. It's only getting worse.

24:00

and Congress has done nothing about it,

24:02

they're far more likely to go, well

24:04

we'll talk about this in just a

24:06

bit, to go after, you know, Facebook

24:08

and TikTok than they are against

24:11

the data broker industry which is far more of an

24:13

invasion of privacy. And I have a theory on that.

24:16

I think they don't want to shut it down because they use it. So

24:21

remember, we know the FBI

24:24

and the NSA buy

24:26

data about American citizens from these

24:29

data brokers. This is

24:31

a really valuable tool for law enforcement.

24:34

So they're not going to, they can't shut it down. They don't

24:36

want to shut it down. Yeah,

24:39

I think that it, honestly, it is that.

24:41

It's the fact that not even, it's not even just

24:43

as simple as like the FBI and these

24:46

various agencies buying it directly. There

24:48

are so many major industries in

24:52

American, you know, in our

24:55

American corporate system that are

24:58

in some way using all of this sort

25:00

of data. It's

25:02

probably never going to have the political will

25:04

to actually be fixed because why

25:06

would any of these institutional powers

25:08

move to fix it?

25:11

Joe Esposito is on fire. I think you must have

25:13

had a couple of cups of coffee this morning. Claim

25:16

your territory now and it's a dog peeing

25:19

on a hydrator. Okay.

25:21

Yeah, that's one way

25:23

to claim your territory. So,

25:27

you know, watch carefully to see

25:29

exactly what the reaction is. Maybe

25:32

run wide and smart enough to know as

25:34

we have asserted that this is just, you

25:36

know, add. Well, he has very smart people working

25:38

for him. He does. He does, including, I

25:41

didn't realize this was Chris Segoian. Segoian.

25:43

He was making a video. Really, really.

25:45

Segoian, there's a few who are tougher

25:48

on privacy than Chris Segoian. Yeah, exactly.

25:50

But he also doesn't want to pursue

25:53

BS. But keep your powder

25:55

dry, so to speak. Google

26:00

did take advantage of

26:02

a Ron Wyden letter, as

26:05

did Apple. Google

26:09

just killed warrants that

26:11

give police access to location data.

26:15

Remember that the police have been, for years

26:17

we've talked about it, using geofence warrants. Who

26:20

was in the neighborhood of

26:22

that crime when it was committed? And

26:25

go to Google and they say, we want the location

26:27

data of everybody in the vicinity. I mean, talk about

26:29

a fishing expedition. They

26:31

did it during the Black Lives Matter

26:34

protests. Who were those

26:36

protesters? That's

26:38

pretty shocking. And

26:42

they didn't do it willingly. They were forced to

26:44

do it. Google didn't, but the local and federal

26:46

authorities did it quite willingly and

26:48

would like to keep doing it. Apple's

26:52

decision to end access to

26:54

location data is, I hope,

26:56

going to put a crimp in that. A

26:59

Google employee who was not authorized

27:01

to speak publicly told Forbes that

27:04

along with the obvious privacy benefits of

27:06

encrypting location data, Google made

27:09

the move to explicitly bring the

27:11

end to such dragnet location

27:13

searches. Cops are going to just

27:15

hate those. They hate it. And

27:18

in Apple's case, Apple has been

27:21

handing over push

27:23

notification information. And

27:26

because these are national security

27:28

letters, they couldn't tell

27:31

anybody until,

27:33

I love this, Ron

27:36

Wyden busted it wide

27:38

open, writing him a letter, and then it was

27:40

public at which Apple

27:43

could say, yes, now we can tell you

27:45

it's happening and we're going to stop it.

27:48

We're going to require a subpoena, a

27:50

judge, to order us

27:52

to hand over push notification information.

27:55

But in effect, they were admitting they'd been doing it

27:57

all along and they couldn't tell anybody until...

28:00

Why couldn't they tell anyone? Because

28:02

in the case of national security letters, one

28:04

of the things is we want this information

28:06

and you may not tell anybody that we're

28:09

getting it. And the reason- Especially

28:11

the object of the- Yeah, that's the thinking is, well,

28:13

you don't want Tony Soprano to know that you've got

28:15

a wiretap on him. But

28:17

that's been extended to the point where we don't

28:19

want any of these protesters to know we know

28:21

exactly who they are. So Apple

28:25

has long, I think, chafed under this.

28:27

As Google with these geofence warrants. And

28:29

the fact that now it came out in

28:31

public because Ron Wyden wrote a letter, they were

28:34

able to say, okay, we're changing our policy. They didn't announce

28:36

it, by the way. But people

28:38

noticed they changed their legal ease to

28:40

say, we will not hand this data

28:43

over without a warrant, which is good.

28:46

So both Apple and Google are on the

28:48

right side of the law. I think in

28:51

this case, some might say, well, they're coddling

28:53

criminals. But I

28:55

would say, if you

28:59

give law enforcement all the power

29:01

it wants to fight crime, we're

29:03

going to be sucked in along with that

29:05

dragnet. And many innocent people will suffer. And

29:09

just all you need is one

29:12

authoritarian to get into Congress or

29:14

become the president. And

29:16

suddenly those powers become really scary

29:20

when the government's doing it. Yeah.

29:24

Honestly, it doesn't even take that for it

29:26

to be very scary. Right. You

29:28

have your local law enforcement agency just

29:31

decide to go off. And

29:34

suddenly you're in a very bad

29:36

place. I'm much more worried about that

29:39

than I am TikTok knowing

29:41

that I like girls in bikinis or whatever

29:44

it is they think I like. That's

29:48

kind of de minimis compared to a government, an

29:51

organization that has guns. The

29:54

right to use them. Yeah.

29:56

And I always say, I've said it

29:58

since I wrote a book on the topic. Them it.

30:00

Portrays. Itself as a protect her privacy

30:02

when it is the most dangerous enemy attacks

30:05

ace. Exactly.

30:07

I lists and this has been a very nice show. Thank

30:10

you very much! Hope you go off and have a one

30:12

on. I guess we have. A few

30:14

more things to talk about. Somebody. Somebody was

30:16

his organization five months ago. Drill read:

30:18

eggnog. I You know, I drink my

30:21

eggnog unadulterated. I know, I don't want

30:23

to dilute whiskey. I did my street

30:25

from the cars read: cream and egg

30:27

and right from the car. and you

30:29

bet. Yep, you. Dude

30:31

on and I had a dogs I'm

30:33

not a now years people hated knew.

30:36

You'd Leo: Yes, I love and I'm in

30:39

my second court disease and. I

30:41

haven't had a d? Why not? It's

30:43

the best part of holidays. Egg.

30:45

Nut Bag radio on Nut Bag not bags.

30:49

Yup, Of eggs. Yeah, I can't

30:51

really do dairies. Eggnog is Not for me. It's

30:54

is very dairy. yeah it's incredibly their

30:56

yeah it really. When I figured out

30:58

I was lactose intolerant, all the things

31:00

that I've always is generally than disgusted

31:02

with your body was traded to ignore

31:04

this one of them your be one

31:07

hundred limestone Not Us Harris Pizza. I.

31:09

Mean. listen. I. Sacrifice

31:12

a lot for to is. Okay, As

31:14

as a. Side be

31:17

having Should I love to consume

31:19

Sees straight off the block every

31:21

got gosh darn Daves know? do

31:24

I do it anyway? Yes and

31:26

no. My favorite! My favorite moment

31:28

of Christmas as I go to

31:31

the of Bryant Park. Oh

31:33

I love magazine! Obvious market

31:35

is so cute! I go

31:37

to Big Cheese house. Yes,

31:40

Oh did they do the

31:42

Rak Clinton Flat? Oh My.

31:44

God. Ah right. So we're

31:46

now. he can explain what

31:48

rec letters not everybody knows

31:50

his Swiss franc. Okay, Rak

31:52

Clad essentially is where you

31:54

take a ah. Some.

31:56

portion of a cheese we'll you put

31:59

it up to heat generating

32:01

kind of object it gets it all nice

32:03

and gooey and then you slice off kind

32:05

of the top part of it where it's

32:07

just a layer of gooey cheese and you

32:09

put that on something and it's fantastic. They

32:12

have a special raclette machines

32:15

designed for the heating of the

32:17

cheese and then they

32:19

slicing the thin little

32:22

bit of the cheese off like

32:24

that the goo. It's molten. Now

32:27

he's putting a plate with a pickle. Well that you

32:29

can do too that's perfectly legitimate. Yeah but that's probably

32:31

the real raclette but that's not what I want. If

32:34

you go to the if you go to the discord

32:36

you will see me in raclette

32:38

heaven. I would like to see you in raclette

32:40

heaven. Alright. This is me. I

32:42

do it once a year. I allow myself this. I'm

32:45

a cardiologist or my wife. Oh there's something.

32:47

That's not you. That's a good one. They're

32:49

scooping on sausages. That's a

32:51

good one. Oh where is that?

32:53

Where is you on right now? I put it in

32:55

there. It shouldn't be in there. No. No. In

33:00

the discord? There it is. There it is. Now

33:03

it is. Oh it takes a while. At the

33:05

baked cheese house. Now that looks like a baguette

33:08

with raclette cheese. With

33:10

a few cornichole. Oh a little ham if

33:12

you like. Ham. Mustard.

33:15

And then they always do the extra lump of cheese on

33:17

the top to make it drip down. Oh it

33:21

is just. Lactose

33:24

intolerant paradise. I

33:27

think Paris even seen that picture as

33:29

having a reaction. Listen I'll

33:31

give it all up for raclette. I'm

33:33

literally searching through my Instagram trying to find

33:35

a photo. Let's go to Bryant Park

33:37

right now. There is. Right there. I

33:39

mean I will. Yeah. There is a wonderful

33:42

French restaurant not too far from my

33:45

apartment cafe Paulette if anyone

33:47

likes Brooklyn. They

33:49

often during the winter I guess I haven't

33:51

seen us back. We'll have raclette and

33:54

one time I was there with two friends we were all just

33:56

gonna get some raclette. They were like oh we've

33:58

actually got a raclette machine. They

34:00

end up bringing into this tiny

34:02

French priest. Oh a giant raclette

34:04

machine Plop it on our table

34:06

and bring out a whole half cheese

34:08

wheel and leave it with us now Plates

34:13

and plates of potatoes Yeah,

34:17

and all like little meats and My

34:20

god, I think we roll out

34:22

of the wheel home. Yeah, basically. Yeah, that's

34:25

just disgusting and you don't think eggnog is

34:27

good I Alright

34:32

I'm going to cafe. I have some I have

34:34

a date in New York cafe

34:36

Paulette. Yeah, Bryant Park Christmas. Yeah, you

34:38

do fair Yeah, yep, and

34:42

Oops all cheese It

34:45

is a special cheese, right? I mean, it's

34:47

not just any old Jesus that she's designed

34:49

for this for rec lighting I believe Yeah,

34:52

but in case you want to just kind of take

34:54

your walking. Well now we got your wheel apartments on

34:56

home and scrape it Used

35:00

for a raclette no, it's okay. You don't

35:02

do search Swiss

35:04

cheese or great greener good. Yeah, it

35:06

would be good beer Yeah,

35:09

okay good sounds delicious Wonderful, I

35:12

assume fontino would also work well

35:14

fontina any any met low melting

35:16

point cheese cheese This

35:22

week and cheese By

35:25

the way, I did want to mention, you know, I've been

35:27

talking with Lisa about the show and you

35:29

know It's about Google a little bit, but it's about everything

35:32

else. We thought for a while me we call it this

35:34

week in general But that's so

35:37

I think honestly given that we had to give

35:39

up our AI show in the club This

35:41

could be almost this week in AI

35:44

Really? I have we have lots to talk about an

35:46

AI. I guess we don't want to be

35:49

limited though Do we know you don't we don't because

35:51

how else do you talk about cheese? Right? That's true.

35:53

We should talk about AI heads

35:55

in here being like what is this? Where's

35:57

my AI news? I came for a

36:01

Well, I got that in here too. I

36:04

took a walk with an

36:07

accelerationist, an AI

36:09

go-go-go guy, and I'm

36:11

going to tell you the tale of that in a little bit. He

36:15

kind of convinced me. Was it a speed walk? Was

36:17

it a fast walk? No. It

36:19

was relaxing. It was quiet. He

36:21

might have been microdosing. I don't know. But

36:24

I have a tale to tell.

36:26

He was very convincing. But

36:29

before we go there... Oh, God. I've

36:32

gone basically from AI skeptic to

36:36

AI like go-go-go. It's

36:39

time for humans to give

36:41

it into the AI the next...

36:43

He said it's like first contact.

36:46

It's an alien species we are giving birth

36:48

to. Oh, no. Leo. We

36:50

can't be back on this. We had

36:52

such a good week last week. We

36:55

walked the whole Gemini thing. I

36:58

believe. I

37:00

think there's got to be a point where

37:03

you stop describing yourself as I'm an AI

37:05

skeptic, but... Not anymore. I

37:07

am an AI believer. This

37:10

is the point right here. The church

37:12

of Teschre. Yeah. You

37:14

know, you're kind of minimizing it

37:16

when you say Teschre. It's so

37:18

much more. It is

37:21

the beginning of the next era in

37:23

humanity, in humankind.

37:25

Bigger than the

37:28

internet, bigger than the personal computer revolution.

37:30

This is... Yes. He said, and

37:33

I love this, he said it's going to get really

37:35

weird in the next 10 years. Okay.

37:38

Wow. Great prediction,

37:40

dude. Wow. What a prediction.

37:42

Huge. I'm a futurist. Why you didn't say

37:44

that for attribution? Well, it

37:47

was probably the most he could get out while

37:50

he was dealing with all of the acid visualizations.

37:52

You know? Okay. I

37:54

will say, I'll give you exactly one little

37:56

bit, which is today, for the first time

37:59

ever, I use... GPT for

38:01

something just in my day-to-day and

38:05

I needed a way to describe it

38:07

succinctly and I kept getting confused. So

38:10

I had ChatGPT summarize it for

38:12

me and a number of... I

38:14

asked like summarize it in plain

38:16

English, two sentences, two sentences but

38:19

plain English and it worked quite

38:21

well actually. Did you see

38:23

the semaphore reporter who has

38:25

created a ChatGPT GPT

38:27

with her articles in

38:29

there? That's... Gina Chu

38:32

is the like top editorial person. It really

38:34

tracks for semaphore. I mean

38:36

it also tracks for Axios. I love

38:38

it. I love

38:40

it really works for semaphore. I'm not

38:42

hating but also come on. Gina

38:46

is a brilliant journalist, a former

38:48

top editor at Reuters and

38:52

unlike others who just are talking about this

38:54

stuff, Gina sat down and created things that

38:56

might be useful for journalists. Yeah.

38:59

Basically what we've seen is one is like network

39:01

LM. That's could I query a set of data.

39:04

She said... That's so last week you saw

39:06

my... I've been using because

39:08

I've been doing the advent code and

39:10

using my common Lisp expert system. It's

39:13

phenomenal. I don't have it write the

39:15

code necessarily but it's really... It's

39:18

like having a Lisp expert sitting

39:20

next to me and I could say,

39:22

well how do I do that and it tells you. Then

39:25

you ask it to go for a walk and it can't

39:27

and you're so disappointed. I don't need it to go for

39:29

a walk. I can go for a walk all by myself.

39:32

A lot of accelerationists. Well, I

39:34

walked with him. He walked too. He had two

39:36

legs. He did have two heads

39:38

but we'll talk about that in a bit. Did you

39:40

look down at the sand and you saw only one

39:42

pair of footprints? Yes. And you

39:45

realized he was carrying you? I saw the whole

39:47

world in a grain of sand. It's

39:49

a beautiful thing. I

39:51

have seen the light my friends and you

39:54

laugh but in

39:57

five years... We do! When

39:59

everything... Going to be real weird. It's.

40:02

Gonna be real weird. He said there's like

40:04

a be a money and in a couple

40:06

of decades minorities or oh god oh my

40:08

god, what are you. Know

40:11

that Crypto? No no worse than that.

40:13

it's you be. I slices going to

40:15

be so much surplus oh that, oh

40:17

jesus that we don't need money. Every

40:19

but I have whenever they need and

40:21

always Sam Short or tall. A

40:24

sister who's Sam.

40:26

Open. Oh no was it was

40:28

fun for you could have been referring to differences

40:30

banks and for you know to the my cats

40:32

are wasn't with it was his. I am going

40:34

to preserve his anonymity but he works in an

40:36

Ai company is where to Microsoft and Google he

40:38

knows all of the year. Of. The

40:41

players. And. You know, obviously

40:43

there's some disagreement the world of Ai

40:45

about what this all means, but he

40:47

would. I am not. Look, I, you

40:50

know I completely acknowledge that this is

40:52

a cookie. Kind. Of point

40:54

of view but I it's possible. He

40:57

said he says we're seeing emerge are

40:59

racing emergent. Properties. Emergent

41:01

behaviors in Ai. And please do.

41:03

You really think that someone who has

41:05

a history of working for the largest

41:08

set company. Is and exists so he less

41:10

because working. In a industries

41:12

that is so saturated,

41:14

the. Sensor Capital that it is

41:16

like the only bright designs in is

41:19

so true. I admit. Elites. Do

41:21

you truly think that he

41:23

venture capitalists. Is working

41:26

to bring about a system with capital

41:28

ceases to exist? Yes, which is this

41:30

is why I am. I respect him

41:32

because he's talking about that. A world

41:34

that is going to be very from

41:36

his point of view very alien. Not

41:38

what he would is been working his

41:40

whole life works but he's you know

41:42

he is. He's working out the company

41:44

that he started and was on the

41:46

board of and is working at that

41:48

company. But. Snack.

41:52

Of the send him. I. Thought

41:54

I just I'm not And I'm not even

41:56

saying that I am now seen the light.

41:59

But. What he but? when? talked about is very

42:01

credible from a person who's actually on the ground

42:04

with this is very credible. And I

42:06

do think there is a possibility. Is it one

42:08

in a thousand? Is it one in a hundred?

42:10

Is it one in ten? I don't know. But

42:12

there is a possibility that we're going to see

42:15

a massive jump in the capabilities of

42:17

AI as it starts to self-generate

42:20

over the next few years

42:23

that will make everything we've been talking about

42:25

kind of moot. But

42:28

big tech is not going to be where

42:30

it's happening anymore, I

42:32

think. How would it not be where it's happening

42:35

if tech is increasingly advancing now? Because most of

42:37

the people who really care about this don't want

42:39

it to be in the hands of Google or

42:41

Microsoft. They want it to be open source. And

42:44

believe me, I want it to be open source. You

42:46

should want it to be open source. You don't want

42:48

Microsoft to own this. Absolutely. Absolutely.

42:50

That much I agree with. Yeah. Yeah.

42:54

No, it should be open source. That's where all the

42:56

advance... It's running fast, mid-journey's

42:58

open source, stable diffusion's open source. Actually, there's some

43:00

stories there. We'll get to that. I

43:02

wanted to take a break about ten minutes ago. You got

43:04

me started, you see? No, no, no,

43:06

no. No, you pushed in. You pushed in my buttons.

43:08

We were ending cheese. You had a nice kind of...

43:12

And then you decided to talk about your walk with

43:14

Jason Calacanis. No, no, no, no. It

43:17

wasn't Jason. Jason, actually, I don't think he

43:19

is a AI... I

43:21

think he's more of an AI-jumber. He knows what he is.

43:23

Because he's Elon's... You know, pal. But

43:25

they're the AI-doomers. They're the ones who

43:27

say we've got to defend humanity against

43:29

AI. This is Ed Riesen and company. Isn't

43:32

this the guy who just released Grock? Yeah.

43:35

Grock, by the way, which is

43:37

a very unimpressive piece of fluffery,

43:40

is not what I'm talking about. I

43:43

don't think we want AI to do dad jokes. Can

43:45

I just say? This is what

43:48

you can Google. It's brought to you by

43:50

Secure My Email. I want to talk about

43:52

something. Actually, I've been advocating for years. Email.

43:55

When you use email, you're sending a postcard.

43:57

Basically, anybody along the way can read it.

44:00

That's why it's so important to

44:03

use email encryption. But it's unfortunately

44:05

up to now been incredibly difficult.

44:08

Along comes something called Secure My

44:10

Email. It's an email

44:12

encryption service that's actually easy

44:14

to use. It

44:16

has the security and privacy you require using

44:20

strong encryption, OpenPGP, ChaCha 20

44:22

and other strong ciphers. But

44:26

it makes it easy. It hides the

44:28

ugly details away. So I've

44:30

been talking about for years S-MIME and PGP

44:33

but those are frankly decades old

44:35

technologies which are very hard to

44:37

implement. They require kind of

44:39

one user at a time exchanging keys. You've

44:41

got to have either a key server or

44:43

a personal exchange of keys. This

44:46

is not going to fly. And I know because I've

44:48

been pushing it for a long time and it never

44:50

took off. But here's the easy way

44:52

to do it. Secure My Email allows you to enjoy

44:54

the simplicity and utility of

44:56

email just as you know it today

44:58

but with the privacy and security of

45:00

modern encryption. It works on Mac, Windows,

45:02

Android, iOS. It

45:05

uses strong open technologies

45:08

like OpenPGP and ChaCha 20. It

45:11

can make your email fully HIPAA and

45:13

GDPR compliant. If you're an attorney

45:15

or maybe you're a tax preparer you're

45:17

doing tax returns and mailing them out

45:20

to your customers, yikes. It's

45:22

a postcard. You need Secure My Email.

45:24

You don't have to change anything. You

45:27

can encrypt your current email addresses both

45:29

personal and business. You don't have

45:31

to change email addresses or providers. You

45:34

can use Secure My Email's apps to manage your email

45:36

but you can also keep using your current setup and

45:38

just use Secure My Email when you want the encryption.

45:41

It's very easy to set up and use. You

45:43

are not using a different server. This is the

45:45

key. You're just using an updated client that

45:48

hides all the complicated

45:50

encryptions. Your recipients, and this is really important,

45:52

do not have to use Secure My Email. They

45:55

don't have to register. They don't need to know

45:57

passwords. Even Companies

45:59

like... Pro Tamil are out of ban passwords

46:02

for non user recipients not secure my email.

46:04

When. Recipients Response: Their email attachments

46:07

to you are encrypted through Secure

46:09

My email systems. And

46:12

here's another great part: Secure. My

46:14

email has a free for ever

46:16

plan. The. Legend Crypt: a single

46:18

consumer email address from female Yahoo, Microsoft,

46:20

and more. Instant download an activation. You

46:22

don't have to give them any payment

46:24

info, you don't have to register, they

46:26

don't have to call you to verify

46:28

who you are for paid plans. Pricing

46:30

is very affordable. Three dollars, ninety nine

46:32

cents a month, Twenty nine, ninety nine

46:34

a year per user. And that. Let's

46:36

use eight email addresses, business or persons

46:39

of free forever for one. Less.

46:41

Thirty bucks a year for eight email

46:43

addresses business or person. Start your free

46:46

account or enjoy thirty day free trial

46:48

of a premium plan. No payment info

46:50

required. Certainly. Worth taking a

46:52

look. And. They've get an

46:54

extra special offers sense. Says

46:58

Smith said secure months it's

47:00

visit secure My email that

47:02

com/twit use the code twit

47:04

at checkout Secure my email.

47:07

Dot. Com smashed to at the

47:09

offer Code is T W I.

47:12

T. Finally I can. Tell.

47:15

People. Encrypt: Without having

47:17

explain. The. Complicated, ridiculous steps enough

47:19

to go through to do it sucks

47:21

it's as simple, secure my email that

47:23

comes less to offer code. It. Was

47:26

saying of so much for their supports. Of

47:28

this week in Leo's A

47:30

I. Revel,

47:33

A hostess this helps the colts

47:35

have a i you're gonna be

47:37

here I'll be in a podcast.

47:39

The focus is great because every lead to

47:41

get to see you descend. Further, And further

47:44

into the can't imagine is a. Of

47:47

eyes. I mean I literally or as

47:49

keeps tried to pull you back six

47:51

weeks ago I was gone. This is

47:53

by See Auto Correct. it's a Parlor

47:56

Directs It's nonsense. But you have to

47:58

admit. It. Has beginning. and

48:00

better and better and most impressive ways and very interesting.

48:02

It still doesn't know a single thing about facts. It

48:06

doesn't believe anything. It doesn't

48:08

know anything. It is a prediction

48:10

machine, full stop. But if you want it

48:12

to be human, it's not. Because it sounds

48:14

like us. I

48:17

think it's pretty useful, but okay. There's

48:20

uses for it? Yeah, it's definitely useful. It's

48:22

just not going to

48:24

totally revolutionize human society

48:26

and bring about the end

48:29

of capitalism. It's got a lot of optimism in five short years.

48:32

Or maybe it will. It

48:35

takes someone like Andrew Ng, who is a

48:37

founder of DeepMind, who very much believes in

48:39

AI, who believes in open source, and he's

48:41

mocking the thinking. Yeah, because he comes from

48:44

the... So there's schisms in

48:46

this, which I also learned about. And

48:49

people like Anthropic and DeepMind were

48:52

separated out from Google, because Google is

48:54

more on the test-gréal side. Larry Page says,

48:57

you know, to Elon Musk, you're a speciesist.

48:59

Let's let the next species come. Let's

49:02

let it come on. And these

49:04

companies are the ones that split off, because

49:07

they said, oh, no, we have to be safe and cautious.

49:10

Why be safe and cautious? Full

49:13

speed ahead. I

49:16

actually kind of believe that, because I

49:18

think that the caution stuff is all

49:20

under the flag of test-gréal and is

49:22

BS. And it's actually the ultimate accelerationism

49:25

is, oh, my God, we have such power we can

49:27

destroy mankind. What's the worst of the BS?

49:30

No, no, no. It's layered down BS. It's going to be

49:32

our little friend. Well,

49:34

and I think this is... I've quoted this

49:36

paper often, 1998, Rand

49:38

Corporation, Paul Doerr, we need to

49:40

get to the unintended consequences sooner.

49:43

Yeah. And I think that we

49:46

know what AI can do, and we must regulate it all

49:48

today, and it's a dumb deal. We

49:50

don't know. We can't. It's

49:52

hubris. Yes, that's the hubris of the president. We cannot

49:54

regulate it, because we don't even know what it is. So you

49:56

know, that happened with the internet in the early days.

49:58

Here's the difference. We've

50:01

seen what happened with the internet, so we

50:03

understand you've got to let this stuff grow, but you

50:05

should have some care

50:08

about how

50:10

it's growing and what's happening and the

50:12

consequences. But I also am asserting

50:14

to be of the opinion that as

50:17

many bad things has come out of the internet ... oh boy,

50:19

I'm sounding like Jeff Jarvis. It

50:22

has been an absolute net

50:24

positive, and had we

50:27

regulated it, we wouldn't have the

50:29

benefits. It would be sad. It wouldn't

50:31

be what we have today. Now, this

50:33

is an example of AI. Earlier

50:36

today, Mary Jo Foley, our

50:39

long-lost host from Windows Weekly, joined

50:41

Paul and Richard on Windows

50:43

Weekly. We

50:46

were talking about ... Microsoft Bing has

50:49

announced some extensions, including one from

50:52

a company called Suno, SUNO.AI,

50:54

that lets you write songs.

50:57

I thought, well, let's write a song for

51:00

Mary Jo Foley. You're welcome here back

51:03

for the holidays. I just thought

51:05

I'd play it for you and get your

51:07

opinion. This is called Christmas

51:10

with You. The AI wrote the entire

51:12

thing. The only thing I did is a

51:14

prompt saying, let's welcome Mary Jo Foley back.

51:17

I think I said, do this in a

51:19

hip-hop style like Jay-Z, which it

51:21

failed at miserably, but here it is. It's

51:26

okay. Full

51:46

mark. It's

51:56

okay. Does

52:00

sound like perhaps the sort of

52:02

intro song that you'd get at

52:05

a Hallmark lifetime like movie?

52:08

Yeah. It sounds no worse than the eight

52:10

million other great songs you're being subjected to

52:12

on the radio this week. I

52:15

don't know if that's true. You think it's

52:17

worse than the barking dogs? I

52:20

don't know what that is. Oh, what?

52:23

Oh, you had to say that, Paris. You had to say that.

52:25

Now you're going to get educated. I had to say. I'm

52:29

just trying to make sure that the show

52:32

is taken down for copyright. It feels like

52:34

a copyright strike. It's definitely a copyright strike,

52:36

so don't play this out loud. But

52:40

I worked in radio for many years in

52:42

which the barking dogs were

52:46

a major... Oh, okay. Yeah, I do. Yeah, of

52:48

course you know what this is. 1971,

52:51

the barking dogs produced in

52:53

Denmark. I don't know. I think

52:55

that that's probably a little bit better than

52:58

the AI song. Okay,

53:00

but you've got to admit it's better than Celine

53:02

Dion or Mariah Carey's

53:05

Christmas songs, yeah? No? She

53:08

sent the hate mail to Leo

53:10

LaPorte at Alooga, California. Mariah

53:14

Carey Hater. All

53:16

right. Anyway, you're right. AI

53:19

has not yet become

53:21

a great musical player. What

53:23

did this guy say that's so convinced

53:25

you? I don't

53:30

know if he convinced me, but I think he

53:32

gave me food for thought. How about that?

53:35

And then I'm no longer completely

53:39

skeptical about the potential for

53:41

AI to become something

53:44

much more than the

53:46

mediocrity that we see so far. When

53:48

it's doing protein folding?

53:50

Yeah, amen. Phenomenal.

53:53

And part of that process was writing this, you

53:55

know, writing. I didn't really write it, but creating

53:57

this lisp expert.

54:00

which is incredibly useful. I made

54:03

an Emacs expert similarly. I really see

54:05

the potential for this. You know,

54:07

one of the things I'm trying to do is

54:10

there is a open source project

54:13

called GPT for All that lets you

54:16

run this stuff on your own computer

54:18

using open source models from

54:20

companies like Microsoft and Facebook

54:23

and Google. And

54:25

it's really interesting because this is

54:28

only running locally. And

54:31

it's really, you know, it's

54:33

not quite as good as some of the

54:35

big giant models that we are using with

54:37

chat GPT and Bing chat

54:39

and so forth, but it's getting

54:41

there. And I think if you look

54:44

at stable diffusion and mid-journey, both of which

54:46

are similar projects, we're

54:48

moving along quite rapidly here. I am

54:50

no longer of

54:53

the opinion that, oh, this is all it's ever going

54:55

to be able to do. I think we are, we

54:57

could very well be on the... Oh, I think you're

54:59

totally right. I think

55:02

there's, I don't think that Jeff

55:04

and I are saying, oh, AI is just a parlor

55:06

trick. It's never going to be anything other than

55:08

mediocre. I think what we're saying is, yeah, it's

55:10

going to be a useful tool. Do you

55:13

think it's possible that it will

55:18

somehow emerge as an intelligence?

55:23

I mean, it's

55:25

possible that I could grow a

55:27

second head. It's highly unlikely.

55:29

It's not, it's highly unlikely

55:32

that I think... I

55:36

don't know. I mean, yeah, I think it's a non-zero chance. Yeah.

55:41

And I don't think that we're

55:43

going to achieve human intelligence. And

55:46

the human is the wrong word. Human is the wrong

55:48

word. It isn't good. No, no, because we're human and

55:50

we have different, you know, it's not going to have,

55:52

you know, the blood vessels

55:54

and the emotion. There's not going to have a lot of

55:56

things we have. We will never know what hungry means. And

56:00

if we said hungry to it, it won't understand it.

56:02

That was profound. That was very good. Yeah, Benito Gonzalez.

56:04

Benito, that's the show. He really should be doing this

56:06

show and I should be sitting there on the board.

56:10

But I do think it's a mistake. Yeah,

56:12

I make a hash of it, so they

56:14

better keep me on this side of the

56:17

board. I think it's a mistake to

56:19

say that it won't achieve something

56:21

that is of equal value. And

56:25

I think it might well be a human-machine

56:27

partnership. I'm not saying it's going to replace

56:29

us. So Leo, I'm

56:32

working on my

56:34

prototype research. And

56:38

when this machine could suddenly spit out whole lines

56:40

of type, it freaked out Mark Twain. He

56:42

wrote a... He thought it was going

56:45

to change the world. Yeah, he lost his entire

56:47

fortune on this. Exactly. At first he

56:49

loved it, but then he hated it. And

56:52

the interesting thing is what Mark Twain said

56:54

is any machine that can set type must

56:56

be able to think. Yeah, see that's a

56:59

mistake. And so we always come along and

57:01

we think that a machine could do something

57:03

we couldn't imagine it could do must be

57:05

and could possibly replace us. Must

57:08

then, at any task, must be

57:10

like us. And

57:12

so we set that as our bar.

57:14

We're doing the same thing now with

57:17

programs and algorithms that predict words for

57:19

us. It's just, as

57:21

Paris said, it's just a machine, it's just a

57:23

tool. It can do amazing things and that's fine.

57:25

But we're on this weird path of trying to

57:28

discuss it as if it's like us. Same

57:30

thing happened with Gutenberg. Same thing happened

57:32

with the

57:35

line of types. Same thing happened with steam powered

57:37

machines. When the steam powered press

57:39

came into the Times in London, the Times

57:42

the next day wrote about how it was this... It

57:44

could almost think because it amazes us

57:47

that it can do something that

57:49

we all had to do with muscle before. So

57:53

yeah, it'll do amazing things. It will do great

57:55

things. I just don't think

57:57

that the scale is that it's going to

57:59

replace. It's going to destroy mankind.

58:02

It's going to be beyond anything we can imagine.

58:04

No, we're going to make them. I'll give you an example. We'll

58:06

imagine it. Somebody brought this up on

58:08

one of our other shows. Maybe it was

58:11

MacBreak Weekly. The

58:13

notion that Elon has that we're going to

58:15

colonize Mars is absurd. Our

58:18

physical bodies just don't do well outside

58:20

the planet. But in AI,

58:22

working on our behalf, we

58:25

could easily become an extra planet species. Oh,

58:27

no. You're hit for long termism here.

58:29

Oh, wait a minute. No, no, no, no. It's not

58:31

long termism. I'm just saying, I don't

58:33

think we're going to explore the stars. I just don't think so. But

58:36

AI might because it doesn't have a body.

58:38

It doesn't have to worry about long term

58:40

effects of microgravity. It doesn't have to... I

58:43

think you're humanizing AI in a way that is... Exactly.

58:46

No, I'm not. ... particularly relevant. I

58:48

think you're human. I think you're human. I

58:50

think... AI will colonize. AI won't do

58:52

any... AI could be a... it's a

58:54

system that is going

58:57

to be maybe powering a robot we use

58:59

to explore a place. Right. Now

59:01

we're getting down to it. It's not going to be doing it. Now we're getting

59:03

down to it. And I think this is the thing that bothers humans a lot

59:05

is the notion that AI could have free will. That

59:08

really bothers people a lot. And

59:11

I think this is an interesting leap to make. Could

59:14

AI have its own

59:16

free will? We don't know if people have their own

59:18

free will, Leo. Well, we don't.

59:20

Yeah. I don't even know if people think AI could. Yeah.

59:23

No, but that doesn't mean AI can't. Thank you, Benito. Maybe AI

59:26

could and we can't. Honestly,

59:29

this is an interesting... This is kind of a

59:31

fundamental question. If AI explores space, is it doing

59:33

it as remote control for us or is it

59:36

doing it on its own? That's a really interesting

59:38

question. Does a self-driving car have free will? It

59:41

seems that way to us because we can't understand. David

59:44

Weinberger in his book said there's no such thing as an

59:46

accident. Only things we can't explain. Yeah.

59:50

Anyway, I'm just... I don't want

59:52

to argue in favor of or against this

59:55

notion. I want to say that I have... My

59:58

eyes have been opened. I

1:00:00

think there's more here than meets the eye.

1:00:05

And some of these fundamental questions like can AI

1:00:07

have... AGI is

1:00:09

deceptive because it's kind of saying, well, is it

1:00:11

like us? Does it think like us? It will

1:00:14

never think like us. I'm not

1:00:16

worried about that. In fact, it's a mistake

1:00:18

to try to make AI duplicate humans. That's

1:00:21

not what we're talking about. I

1:00:24

think it's going to be very different. That's what he means, I think,

1:00:26

when he says it's going to get weird is

1:00:28

that we are going to have a relationship

1:00:31

with a non-human

1:00:33

intelligence. And that is

1:00:35

going to be very weird. And

1:00:37

we're not going to like it. That's

1:00:40

all I'm saying. And it may not happen.

1:00:44

We may be talking about the latest

1:00:48

features of the iPhone 17 in two years.

1:00:51

If that's the case, I'm just going to quit now. But

1:00:56

I think we may have more to talk about over the

1:00:58

next few years. And it might

1:01:00

be quite surprising. By

1:01:03

the way, the rich kids now

1:01:05

are worried about living forever, of course.

1:01:09

This is just... Of course. This

1:01:12

is from Bloomberg. Silicon Valley's

1:01:14

quest to live forever has,

1:01:16

what a surprise, many warring

1:01:19

factions. Jack

1:01:21

Titans, venture capitalists, crypto enthusiasts,

1:01:23

and AI researchers have

1:01:25

turned longevity research into something... They're

1:01:27

all on Kallikana's podcast. Between the

1:01:30

hottest science and the tragic comedy.

1:01:36

One Saturday in August, Anastasia

1:01:38

Egorova, a 37-year-old chief executive

1:01:41

of a longevity research nonprofit, organized

1:01:44

two dozen volunteers in San Francisco and 10

1:01:46

other cities to get answers from almost

1:01:48

200 passers-by. They

1:01:50

were wearing sweatshirts. They

1:01:53

said, say forever, which

1:01:55

didn't make any sense until you

1:01:58

found out that what they were asking... all

1:02:01

these people is, how long would

1:02:03

you like to live? And

1:02:06

it ranged between, for most people,

1:02:08

80 and 120 years, but what they said is,

1:02:12

say forever. Could

1:02:14

we live forever? I don't think it's a good idea.

1:02:17

If the people on this

1:02:19

planet today live forever, that's gonna cause

1:02:21

a problem. We're gonna have a

1:02:23

couple of issues to figure out for

1:02:25

all living forever. Yeah, they

1:02:28

carry around posters with mantras like, death

1:02:30

is unacceptable, death is

1:02:32

boring, and stay alive.

1:02:34

Just gagged me. I

1:02:37

love the quote at the end of this paragraph, which is

1:02:39

from the 37 year old CEO

1:02:41

of the research

1:02:43

nonprofit. They say, quote, dying

1:02:46

is bad. This is something

1:02:48

humanity doesn't take seriously enough. Wow.

1:02:51

Huge. Thanks so much for

1:02:53

that. Maybe she should do

1:02:55

some more reflection. The ego

1:02:58

of these guys, that's

1:03:00

what, we've got

1:03:02

to demote them. We've got to just make

1:03:04

them irrelevant. The usual suspects, Jeff

1:03:06

Bezos, Sam Altman, Larry Ellison, Larry

1:03:09

Page, and other tech titans, according

1:03:11

to Bloomberg, have pledged hundreds of millions of

1:03:13

dollars towards companies pursuing

1:03:16

longer life. Remember, a long-termism thing

1:03:18

too. They think they're so, what

1:03:21

Emil Torres explained to us on

1:03:23

AI Inside is that they

1:03:26

think that when they get to the super intelligence, that

1:03:28

what it's going to be able to do is make

1:03:30

them immortal. That it's going to be

1:03:32

so smart, it will figure that out. Yeah,

1:03:34

well there's a whole, and I love this, science fiction

1:03:37

series called the Babaverse, which

1:03:40

addresses this. It

1:03:42

starts off the whole world

1:03:44

of Roberts, sort of, yeah. The

1:03:47

guy, Robert, is a tech

1:03:50

billionaire who's made a lot of money, and

1:03:52

he signs a deal to have his brain

1:03:56

scanned and preserved, and

1:03:59

then he gets hit by a car. like immediately and

1:04:02

then wakes up and he's in a machine. But

1:04:05

is it him? And then it turns out they

1:04:07

can clone him. So there is

1:04:09

literally a universe of Bob and

1:04:11

they all deviate slightly over time so

1:04:13

they have some unique personality quirks but

1:04:16

they essentially have his memories before

1:04:18

death. And so it's a very

1:04:20

interesting book. I love it and it's funny. I

1:04:23

recommend it. But yeah, I

1:04:26

don't know if this is really a good idea.

1:04:28

Robert Nelson, a hedge fund manager who

1:04:30

has a stake in longevity focused biotech,

1:04:32

Altos Labs says, aging

1:04:35

is a humanitarian

1:04:37

disaster that kills as many people

1:04:39

as World War II every

1:04:42

two years. The

1:04:44

horror! The horror! He

1:04:47

takes a dozen drugs a day

1:04:49

including rapamycin which has been

1:04:51

shown to increase lifespan in mice. I

1:04:54

remember Ray Kurzweil, the AI researcher

1:04:57

and synthesizer maker who thinks the singularity

1:04:59

is near, takes a fistful

1:05:02

of supplements

1:05:04

every day. His goal is to

1:05:07

live long enough to live forever. He

1:05:13

thinks if he can just get it, then

1:05:15

that's what these playdribbers all believe. Yeah. Give

1:05:18

me a couple of extra decades. It's not just the super intelligence. It's

1:05:20

not just artificial intelligence. It's not

1:05:23

just general intelligence. It's artificial general

1:05:25

super intelligence. I think that that

1:05:27

is probably a perversion of what

1:05:30

AI really could be. I don't

1:05:32

think AI is here to preserve

1:05:34

our lives. Oh man, philosophy and

1:05:36

ethics and religion and psychology and

1:05:39

yeah. I think

1:05:41

it's all tied up ultimately in the

1:05:44

humanity's fear of death. I feel

1:05:47

like especially these illuminant,

1:05:50

like these luminary tech CEOs,

1:05:54

they fear the idea

1:05:56

that one day everything

1:05:58

that they've devoted their. energy to creating

1:06:00

will cease to exist, not their death is

1:06:02

just an expedient to that. So of course

1:06:05

the only option is to try and get yourself to

1:06:07

live forever so that you never have to

1:06:09

confront that eventual reality. Now

1:06:11

you talk about one VC, Martin

1:06:14

Tobias, who's almost 60,

1:06:17

who has half a million dollars

1:06:19

of equipment in his garage, two

1:06:22

saunas, an infrared lightbed, an electro

1:06:24

muscular stimulation suit, and a cryotherapy

1:06:26

chamber. He takes cold

1:06:29

plunges, flies to Central America to get

1:06:31

injections of stem cells, and undergoes treatment

1:06:33

to lengthen his telomeres, chromosomal

1:06:35

proteins that shorten with age. I

1:06:40

don't know. I mean, you know, maybe?

1:06:43

Yeah, I'm consuming. But it seems like it's a waste of

1:06:45

time and money. It sounds like people need to download the

1:06:47

We Croak app. Yeah, We Croak, baby. I gotta be reminded

1:06:49

five times a day, I'm gonna die. By the way, I'm

1:06:51

really annoyed. If you tell me I'm gonna die, that's annoying.

1:06:53

The ego is tremendous, is that the

1:06:56

world needs me. I'm

1:07:00

so special. I should live

1:07:02

forever. Yeah, I don't want to die

1:07:04

when I die, but I don't

1:07:07

mind dying. I think the trick is

1:07:09

to make the most out of every

1:07:11

moment, not be lying in a thermonuclear

1:07:14

bed trying to survive.

1:07:16

So I would like to have a sauna in his garage. A

1:07:18

sauna would be nice. I

1:07:21

can smell those smell of roses. There

1:07:26

are bad things. Rite Aid has now been

1:07:28

banned from using face

1:07:31

recognition by the FTC.

1:07:33

What is Rite Aid using face recognition

1:07:35

to begin with? Does Rite Aid go

1:07:37

bankrupt? They're practically bankrupt. They

1:07:40

use face recognition at stores in

1:07:42

large cities to catch shoplifters.

1:07:45

The system used low quality images

1:07:47

often taken from security cameras to

1:07:49

create a database of alleged shoplifters

1:07:52

and would set alerts to

1:07:54

employees when it flagged a match against somebody

1:07:56

entering the store. Then

1:07:59

the employees would follow the database. By the

1:08:01

way, most often blacks, Latinos, and women, the

1:08:04

employees would then follow customers

1:08:07

around the store, sometimes even

1:08:09

call the police or falsely accuse

1:08:11

people of shoplifting. So

1:08:13

when I came back through the UK and my trip

1:08:15

from Vienna, I was in London for the

1:08:17

night and I turned on the TV. By the way, British

1:08:20

TV sucks. And

1:08:22

there's an entire series on Channel

1:08:24

5 in which these people see

1:08:27

people who look suspicious on the high

1:08:29

street and then follow them in and

1:08:31

think they're going to shoplift. And

1:08:35

in one case they said, well this guy, I'm sure

1:08:37

he's going to do it, oh no, obviously he knew

1:08:39

we were there so he decided to turn the corner.

1:08:41

Is that a reality show? Yes.

1:08:44

It is noxious. We can thank the UK for

1:08:47

some of the worst reality shows. They all came

1:08:49

from the UK. Oh my lord. But

1:08:53

we also, they have shows like QI with

1:08:56

Stephen Fry. They also have

1:08:58

Love Island which is the best show

1:09:00

in which it's ever existed. And

1:09:02

I always forget the name of it, that one where they give

1:09:04

you crazy tasks and you have to go out and... Taskmaster?

1:09:07

Oh that rule. That's a

1:09:09

great show. Taskmaster is phenomenal.

1:09:12

Really recommend it. So there.

1:09:15

That's on Channel 4. The

1:09:17

UK, speaking of the Emerald Isle,

1:09:19

no, what is it? What

1:09:21

is the UK? That little

1:09:24

guy up there above Europe. That

1:09:27

island nation above Europe. Formerly Europe.

1:09:30

Former European says, the

1:09:32

UK Supreme Court says AI cannot

1:09:35

invent things. Can't hold a patent.

1:09:38

This I think comports with what US courts

1:09:40

have said. Well, no,

1:09:42

it doesn't want patents. It doesn't

1:09:45

care about us. It doesn't care

1:09:47

about our silly little government, our

1:09:50

US patent and trademark organization. Those

1:09:52

are picky you and concerns today, I

1:09:54

know. In the long term. In the

1:09:57

short term. They don't care about so

1:09:59

many things. It doesn't care

1:10:01

about nothing. The

1:10:04

only problem with AI is it would be like if

1:10:07

you just took just the intellect, just

1:10:09

the voice in our heads out of us, that's

1:10:12

it. And no heart, no soul. The

1:10:14

intellect, you know, a synonym for that,

1:10:16

intelligence. Yeah. Huh.

1:10:18

Yeah. But I mean,

1:10:20

if you didn't have intelligence moderated by your

1:10:22

feelings, your heart, your empathy, all the things

1:10:24

that make you a human, it

1:10:27

would just be a calculator, right? I

1:10:31

think it would probably just be a, yeah, stimuli

1:10:36

processing. Yeah. And

1:10:38

that's what AI is going to be. We have to

1:10:40

give it the heart, my friends. We

1:10:42

have to tell it what it's like to be hungry. That's

1:10:44

our job. I don't know. Anyway,

1:10:47

you can- Every night, Leo goes to his

1:10:49

GPTs and describes the hunger he feels. I

1:10:51

tell you how I feel. My

1:10:54

stomach is all knotted up

1:10:56

with a lack of food. Actually,

1:10:59

I've never known hunger. I'm sad. Look

1:11:02

at me. Do I

1:11:04

look like I've ever been hungry? A

1:11:07

US computer scientist on Wednesday lost his

1:11:09

bid to register patents over

1:11:12

inventions created by his artificial

1:11:14

intelligence system. He

1:11:18

created a, quote, creativity

1:11:21

machine called DaBus. The

1:11:26

UK's intellectual property office said,

1:11:28

no, you got to be

1:11:30

a human or a company

1:11:33

rather than a machine. So he appealed

1:11:35

to the Supreme Court, which on

1:11:37

Wednesday, this morning, unanimously rejected

1:11:40

his appeal as

1:11:42

under UK patent law, an inventor, quote,

1:11:44

must be a natural person, not

1:11:46

an unnatural person. So what are

1:11:48

the fancy ways this will be used to get around the

1:11:51

law, I wonder? Well,

1:11:54

the bad thing about this is it kind

1:11:56

of encourages people to hide their inventions. Right.

1:11:59

The whole point is- patent law is

1:12:01

that when you invent something, yes, you should be

1:12:03

able to capitalize on it for a limited number

1:12:06

of years, but then it should become public.

1:12:09

Everybody should be able to do it. So

1:12:11

if you invent a rubber tire, sure, for

1:12:13

60 years or whatever the term

1:12:15

is, you can profit from it and no one

1:12:18

else can make one. But after that, we can

1:12:20

all make them because that's good for society. So

1:12:22

if you don't allow patents, I think

1:12:24

that just incents the inventors to just

1:12:26

keep it to themselves. I think

1:12:29

this is a different situation. This judgment doesn't

1:12:31

stop a person from using an AI

1:12:34

to devise an invention. It just

1:12:36

stops listing the AI tool as

1:12:38

the inventor. The

1:12:40

person who used that AI tool to invent

1:12:42

it could patent it. They just have to

1:12:44

be listed as the inventor, not the AI.

1:12:46

So suddenly somebody's going to seem

1:12:49

like they're incredibly prolific patenting

1:12:51

200 things in a month because

1:12:54

they're just Elon Musk brilliant. But

1:12:56

in fact, the AI did it. Well, again,

1:12:59

my point is that's good if they do

1:13:01

patent it because that reveals it in order

1:13:03

to make a patent. You have to be

1:13:05

patent trolling. It'll be a

1:13:07

new form of patent. Well, that's another problem. They'll

1:13:10

try to preemptively put a

1:13:12

mark around everything. Yeah. Yeah.

1:13:16

All right. I think there are good

1:13:18

reasons for them to accept patents from

1:13:22

whoever and make it public.

1:13:25

I don't know. I don't know. You know, when

1:13:27

the AI takes over, it ain't going to

1:13:29

matter because there won't be no more money. That's

1:13:34

true. See, that's where it goes

1:13:36

too far is when it tries to think that

1:13:38

the technology is technological determinism to the extreme, that

1:13:41

is suddenly going to change all of

1:13:43

society overnight. The

1:13:45

internet is a pretty damn big change that

1:13:47

we can all talk to each other. And

1:13:49

it did change a lot, but it

1:13:51

didn't change us. It's

1:13:54

true. Still screwed up. We're

1:13:57

in funny hats. I would argue that this

1:13:59

is not. in the same realm

1:14:02

as the internet. But

1:14:04

we'll see. You know what? You know

1:14:06

what? It will give me great pleasure

1:14:08

in the year 2333 to toddle onto this stage and

1:14:10

say, I told

1:14:14

you it was going to get weird.

1:14:20

There is a dataset so

1:14:23

weird. The

1:14:26

hierarchy of information in AI

1:14:28

is complicated. You have a dataset

1:14:32

that is trained on inputs.

1:14:35

And then you tune that dataset, often

1:14:37

by humans, to

1:14:39

be more useful in certain ways. There's

1:14:42

a whole process involved. There is a

1:14:44

machine learning dataset called Leon 5B. And

1:14:49

apparently a lot

1:14:52

of AIs

1:14:54

use this, including stable diffusion.

1:14:59

Leon 5B has a problem. It contained,

1:15:01

according to a Stanford study, our good

1:15:03

friend Alex Stamos, 3,226 suspected

1:15:08

instances of child sexual abuse

1:15:11

material. At least a

1:15:13

thousand of them were actually validated by

1:15:15

NICMIC. So

1:15:17

Leon on Tuesday told 404 Media

1:15:20

that out of an abundance of caution,

1:15:22

out of an abundance of

1:15:26

caution, I am not going to give

1:15:28

you a Christmas gift this year, Paris. It's

1:15:31

out of an abundance of caution. Out

1:15:34

of an abundance of caution, Leon took

1:15:36

down its datasets, including 5B and another

1:15:38

called Leon 400M, temporarily

1:15:41

to ensure they're safe before republishing

1:15:43

them. Alex Stamos,

1:15:45

the Stanford Internet Observatory, found

1:15:48

the suspected instances of CSAM through a

1:15:51

combination of perceptual and cryptographic

1:15:54

hash-based detection and analysis of

1:15:56

the images themselves. Pretty

1:15:58

sophisticated. It

1:16:03

doesn't mean that you would get that imagery

1:16:05

if you asked Ableton Fusion

1:16:07

to do it. The

1:16:10

paper says while the amount of CSAM present

1:16:12

doesn't necessarily indicate that the presence

1:16:14

of CSAM drastically influences the output of

1:16:17

the model above and beyond the model's

1:16:19

ability to combine the concepts of sexual

1:16:21

activity in children, it likely

1:16:24

does still exert influence. The

1:16:26

presence of repeated identical instances of

1:16:28

CSAM is also problematic due

1:16:31

to its reinforcement of images of specific

1:16:33

victims. I return once again

1:16:35

to the Stochastic Parrots

1:16:37

paper. It says when you try

1:16:39

to make these incredibly large models, you

1:16:42

lose all ability to audit the input, let alone

1:16:44

the output. Right. And that's

1:16:46

why we need smaller models and open source models so

1:16:48

that we can audit them. Or

1:16:51

just scrape every damn thing you can find. The

1:17:02

problem is it's too

1:17:05

big a dataset to manually check every

1:17:08

sample. Exactly. Exactly.

1:17:11

You don't know where you got it because you're scraping everything

1:17:14

and that means that you can't have quality

1:17:16

control of it. It's

1:17:19

like when you're a net fishing and you kill the

1:17:21

dolphins, you know? Say again? Like you're net fishing and

1:17:23

you kill the dolphins. Yeah, the person saying that's, you

1:17:25

just want tuna but you get some dolphins in

1:17:27

there. Yeah.

1:17:30

Okay. It's also, this

1:17:33

is a show entirely today about male ego

1:17:36

because that's part of what Margaret Mitchell

1:17:38

and Emily Bender talked about too and

1:17:40

Timothy Givler with these large models is

1:17:42

they just want it to be BSD.

1:17:45

They want it to be big for the sake of big.

1:17:47

Mine's bigger than yours. And they say

1:17:49

that's just absurd. It makes it worse. It's

1:17:51

not good. But they want size.

1:17:54

Well, a funny thing that you should mention,

1:17:56

Margaret Schmittchel, because she's

1:17:59

now working at hugging. face, the folks that put

1:18:01

out stable diffusion. She's their

1:18:03

chief ethics scientist. And

1:18:05

she tutored, I just wanted to pop in to say that

1:18:07

there's been a lot of time and energy spent

1:18:10

on trying to find sea salmon. None has

1:18:12

been found. Some

1:18:14

people at HF or hugging face have

1:18:17

been attacked as if pedophiles, but it's

1:18:19

just inappropriate cruelty. So,

1:18:21

she's defending them. So,

1:18:28

be careful who you quote. It's

1:18:31

interesting. Now, she's on the other

1:18:33

side. Yeah,

1:18:36

I think we can agree that CSAM

1:18:38

should not exist. It shouldn't be propagated

1:18:40

and certainly shouldn't be used in AI

1:18:44

models. Apple's

1:18:49

going to stop selling the Apple Watch tomorrow, so

1:18:51

run over to the Apple Store if you

1:18:53

want one of these suckers. If you want an

1:18:56

Apple Watch Ultra 2 or

1:18:58

a Series 9 Apple Watch, the reason these are

1:19:00

being pulled off the market

1:19:02

is because they have a blood oxygen

1:19:04

sensor in them. And there's

1:19:07

a little company called Massimo that makes

1:19:09

blood oxygen sensors that's

1:19:11

telling Apple, hey, dudes, that's

1:19:13

our patent. Apple

1:19:16

disagreed, but the International Trade Commission,

1:19:18

the ITC, did not. And

1:19:21

they have said there will

1:19:23

be a ban hitherto

1:19:26

for from now on in

1:19:28

the future in all perpetuity to

1:19:31

bringing those into the United States until

1:19:33

they resolve this dispute

1:19:35

with Massimo. I

1:19:38

mean, I know this is how patent law works,

1:19:40

but how can you patent just a

1:19:42

basic health measurement

1:19:45

tool like a blood health

1:19:47

oximeter? It's unfortunate.

1:19:50

Well, and I think I'll bet, Paris,

1:19:52

this is when you do one, it has

1:19:55

to go through both sides of your finger.

1:19:58

That's all right. So to be able to do it. do it just

1:20:00

on one side or something. Bouncing light off. I mean, there

1:20:02

is the technology involved in this. What

1:20:05

happened was that this company, Massimo, released

1:20:07

a watch with continuous

1:20:09

pulse oximetry and

1:20:12

Apple sued him and said, wait a minute, we

1:20:14

got the Apple watch, you can't do that. Too

1:20:16

much Massimo counter-suits saying, yeah, but we invented it,

1:20:18

dude. So Apple kind

1:20:20

of poked the bear and the

1:20:23

bear so far as one. Apple is hoping

1:20:25

the President of the United States can veto

1:20:28

the ban, but he only has till Christmas

1:20:30

Day. But

1:20:34

I think Brandon Claus might come through on this one.

1:20:36

It's just a possibility. Apple is

1:20:39

pulling the watch off the, in

1:20:41

an abundance of caution, Apple

1:20:43

is pulling the watch off the

1:20:45

shelves at its own stores.

1:20:48

Of course, if other retailers have some in stock, I

1:20:50

guess they could still sell it. But

1:20:53

Apple can no longer import new ones. And

1:20:55

I think Apple is kind of trying to bring

1:20:58

it to a head before Christmas Day so

1:21:00

that Santa Brandon will come through

1:21:03

and say, oh, never mind. I

1:21:06

think Massimo actually has- It doesn't seem like the

1:21:09

plot of a Hallmark movie, like

1:21:11

Christmas movie, trying to get Joe

1:21:13

Biden to veto a ban

1:21:16

on your device right before

1:21:18

the clock strikes Christmas Day. I hope he

1:21:20

signs it in front of the Christmas tree

1:21:22

in the Oval Office with a

1:21:24

little Santa hat on. It

1:21:27

looks like Massimo actually has

1:21:29

some merit in this. They have a light

1:21:31

shining through it and all that stuff. It

1:21:33

looks very much like an Apple Watch technology.

1:21:36

So anyway,

1:21:40

we'll watch that with interest. Maybe

1:21:43

the Prez can get

1:21:45

him out of jail. Take

1:21:48

another break and then we're going to talk wordle

1:21:51

because that's really what matters. Oh no. Oh

1:21:54

no. Do you wordle, Jeff? No. And

1:21:57

I Hate people who do. Just Be on the record. I.

1:22:00

Didn't greens? I

1:22:02

assume rinse do do justice. Syria was

1:22:05

you constantly. I don't tell me that

1:22:07

is very annoying. Senior were to like

1:22:09

yourself. Yes, He want

1:22:11

to do it on your own? Fine? Just stop.

1:22:14

Are. You grinches will talk about and just

1:22:16

a moment my. God. Playing.

1:22:19

The role of Ebenezer Scrooge and or episode

1:22:21

as the airing of grievances how as good

1:22:23

and we got the Fbi got the Paul

1:22:25

You Can Dancer and else or sure they

1:22:28

brought to you by a good friends that

1:22:30

I T Pro T V you might say

1:22:32

whatever happens those guys will there Now I

1:22:35

see I learning they have been our studio

1:22:37

sponsors all year in case you didn't notice,

1:22:39

is that cool in today's I T Talents

1:22:41

heritage? Whether you operate as your own department

1:22:44

or your part of larger team. Gp

1:22:47

scales up to date. Ninety.

1:22:49

Four percent of see Ios and ceases

1:22:52

agree. Attracting and retaining town is increasingly

1:22:54

critical to their roles were a Cia

1:22:56

Learning is the best way to keep

1:22:59

your eye. T team now only up

1:23:01

to date a happy. They. Love

1:23:03

the the love the training. Especially because

1:23:05

a salary never teaches of may already

1:23:07

know it only teaches him the things

1:23:10

they need to know to become more

1:23:12

effective. And who wouldn't want to become

1:23:14

more effective? It's a sailor. Has more

1:23:16

than seventy two hundred hours of contents

1:23:19

at Always fresh. They're adding new content

1:23:21

every day to keep you at the

1:23:23

top of your game and you team

1:23:25

will love. The training is entertaining. Ac

1:23:27

ever is. Structures are passionate. Experts.

1:23:30

In the field, people actually working in

1:23:32

the field they're passing communicate It makes

1:23:35

these really engaging. You learn and you

1:23:37

enjoy is not a surprise. A salaries

1:23:39

completion rate and their videos fifty percent

1:23:41

higher than the other guests. a see

1:23:44

I'm Learning is added to that. I

1:23:46

think you might find useful called Cyber

1:23:48

Skills. It's a solution

1:23:50

to future proof your entire organization

1:23:52

not just the idea, department's like

1:23:55

this is specifically cyber security awareness

1:23:57

training for non I T professionals.

1:24:00

Cyber skill you get. Your employees get a

1:24:02

simple one hour course Overview: They.

1:24:04

Gain Attack Specific training and knowledge Check

1:24:06

assessments based on common cyberthreats they're going

1:24:08

to encounter a daily basis and you

1:24:10

need them to know about You need

1:24:13

to protect yourself by getting them aware.

1:24:16

It covers everything from password

1:24:18

security, phishing, scams, malware prevention,

1:24:21

Network. Safety And again because these

1:24:23

are engaging well produced videos. isn't your

1:24:26

boring the of flash slide shows had

1:24:28

so many companies have. This is good

1:24:30

stuff. It's they're going to be engaged

1:24:32

that be motivated. They'll learn and then

1:24:35

after the one hour offers you the

1:24:37

that access to bonus courses. They got

1:24:39

documentary style episodes. Your place will learn

1:24:42

about cyber attacks and breaches whatever way

1:24:44

they want. So. Important

1:24:46

that they learn this stuff cyber

1:24:48

skills from a Cia. Learning is

1:24:50

one more way. Salary helps, investing

1:24:53

your team and trust them to

1:24:55

thrive while increasing the entire security

1:24:57

of your business. Booster Enterprise Cyber

1:24:59

Security Tough as they were, they

1:25:01

see I learning. Be bold. Trains

1:25:04

Smarts is a go daddy See I learning.com/to

1:25:06

it as a twit listening to get up

1:25:08

to sixty five percent off a mighty pro

1:25:11

enterprise solution. Plants Israel deplete. The form is

1:25:13

depends on the size your team. Fill out

1:25:15

this form and find out how much you

1:25:17

can save. Right now a Go. At

1:25:20

a see I'm learning.com. To

1:25:24

it! Great sponsors, great friends. Been with

1:25:26

his for many years or so. Glad

1:25:28

they were with this and Twenty Twenty

1:25:30

three Ac I Learning! Go Daddy! See

1:25:33

I learning.com/ Twits.

1:25:37

Ah, edward old during the outbreak.

1:25:39

Oh wow, you're a fast wertmuller.

1:25:42

So. What is your? This is the

1:25:44

key here. What is your first word?

1:25:47

everybody? It. Is always I

1:25:49

rates. I are able high

1:25:51

rates are I rates irate. It

1:25:53

has to be a of as

1:25:55

a five letter word Right to

1:25:57

see I raise has. A

1:25:59

good. amount of vowels, you've got I, A,

1:26:01

and E, and it also has R and T.

1:26:04

Yep. So, as we all know, the

1:26:07

most common letters in the English language

1:26:09

are E, T, A, I, O, N,

1:26:11

S, H, R, D, L, U in

1:26:13

order, right? As we all know.

1:26:15

As we all know. So, the one...

1:26:17

Unshroodly. Yes, that's my

1:26:19

friend. And

1:26:21

as a result, that would make sense. You certainly got to have

1:26:24

E in there. It wouldn't hurt to

1:26:26

have a T in there. And I, you know... I

1:26:29

use tiers. I don't know

1:26:31

why. It works very well for me. The

1:26:33

folks at the New York Times have

1:26:35

analyzed the half a

1:26:37

billion wordles people did this year

1:26:41

and published an article, these seven things

1:26:43

we learned while

1:26:46

analyzing these words. They said

1:26:48

the number one first word is

1:26:50

adieu. A-D-I-E-U. Yeah,

1:26:53

that was... I believe early

1:26:55

on when wordle was becoming popular,

1:26:58

maybe the Times or someone wrote that

1:27:00

adieu was the smartest opening guess.

1:27:02

So, a lot of people started doing.

1:27:05

Audio is another one, very close. But

1:27:08

in their analysis, taking

1:27:11

a look at all the first words, because that's really about all

1:27:13

you can do is the first word, they

1:27:16

found adieu is a terrible

1:27:18

guess. People

1:27:20

who start with adieu need about a third

1:27:22

of a turn more to solve their wordles

1:27:24

compared with players who started with slate, for

1:27:27

instance, which they use as a baseline. That

1:27:29

means 132 extra turns over the course of

1:27:32

a year. The

1:27:34

worst words, adieu, audio, traits. Ooh, irate is

1:27:36

on there. Oh, yeah, of course it is.

1:27:38

That's so fun. And that's a good one,

1:27:41

although 47 extra guesses over the year. Steam,

1:27:45

house, aisle, heart, train,

1:27:48

irate, arise, arose, raise,

1:27:50

stare, least, crane and

1:27:52

slate. Tears is not on there. Sounds like a

1:27:54

boggle list. It is kind of a

1:27:56

boggle list. It's all five-letter words.

1:27:59

You've played wordle. If you haven't, you get

1:28:01

six guesses to figure out what

1:28:03

a five-letter word is. The first word

1:28:05

really makes or breaks. You

1:28:09

did well with yours, Irate. I'm

1:28:11

going to try. I'm not going to give you

1:28:13

any ... Cheers here. The way

1:28:15

it works, Jeff, is

1:28:18

you try your starting word and then it'll tell

1:28:20

you with a green tile

1:28:22

that that letter is in the right

1:28:24

place and correct. The yellow tile tells

1:28:26

you the S is there but not at the end. We

1:28:29

know that there is no R, E, or

1:28:31

T in there, so you could continue on.

1:28:34

Leo, do you play Wordle on hard mode

1:28:36

or do you play normal? I

1:28:39

didn't even know there was a hard mode until

1:28:41

recently, but I play as

1:28:43

if I'm playing hard

1:28:45

mode because hard mode ... Go to the next one. Go to

1:28:47

the next one. Go to your next one.

1:28:49

Hard mode keeps you from ... What is

1:28:52

it? You have to use

1:28:54

... Essentially, using this example ... It's

1:28:57

shown us that A is the middle

1:28:59

letter. On hard mode, you

1:29:01

couldn't have any future guesses. They didn't

1:29:04

have A as the middle letter and

1:29:06

didn't include S, which we know is in there

1:29:08

somewhere. Isn't that an easier mode? No.

1:29:11

No. It would be easier

1:29:13

because if you want to just brute force it

1:29:15

and be like, let me then, I know that

1:29:17

A is there. I know that S is in

1:29:19

the word. Let me use a totally

1:29:22

different word that doesn't have anything to do with

1:29:24

that, try and figure out what are the other

1:29:26

three letters in the wordle. That

1:29:28

would be easy mode, but I

1:29:30

think that that's cheating in my opinion. I can't use

1:29:33

an E. I know E ... Yeah, see,

1:29:35

I always use it as if I'm in hard mode. I

1:29:37

don't know if it's on turned on. E-T-A,

1:29:41

shirdlu. I should slash

1:29:44

... This is where I hate it is where

1:29:46

they repeat a letter because you

1:29:48

don't know ... Oh, yeah, that's the worst. That's the worst. Sometimes

1:29:51

they do that. Totally brave for playing

1:29:53

wordle live. World live

1:29:55

wordle for the first time ever. I think I'm

1:29:58

going to get it right now. Because

1:30:00

I know there's an S and

1:30:02

a right in the middle and that L has to

1:30:05

occupy one of the last So

1:30:07

how about we do S? Oh, no, I know I

1:30:09

can't have an E that makes it hard That

1:30:12

makes it hard Snail would

1:30:14

work right? Let's try snail No,

1:30:18

you don't like snail. Oh Yeah,

1:30:21

that has all the right letters. It has

1:30:23

all the right stuff. Oh So

1:30:27

the only these two are wrong, right? S

1:30:31

oh Boy

1:30:34

how many how many guesses did you take? Mine

1:30:37

took four, but it should have taken three. I was

1:30:40

just I got distracted by your ad reading Love

1:30:44

hearing that Can't

1:30:46

use a T. It's just those host red

1:30:48

ads. They're so in there. You know good

1:30:50

aren't they? Aren't they

1:30:53

wonderful? a

1:30:57

blank L You

1:31:00

know it John What

1:31:03

what do you mean I'm killing you Is

1:31:05

it because you knew it already or just because

1:31:07

you're looking at and you're going that's by the

1:31:09

way the success of wheel of fortune Murph Griffin

1:31:11

designed it that way. Yeah, is everybody home knows

1:31:13

the answer Well, the people on

1:31:15

stage looks stupid because they don't know that's the

1:31:18

point of all the television is to laugh at

1:31:20

your fellow human beings Oh, yeah Mocking

1:31:24

There was a clue. Oh That's

1:31:28

the clue Fleh

1:31:33

Scrant fire Flynn

1:31:38

A-blank good clue actually really good.

1:31:41

You're killing me. Oh, I

1:31:44

don't know I give up complete it Do

1:31:46

I have to finish the movie quote? It's a

1:31:49

movie color. Yeah, you're killing me smalls M-a

1:31:54

L. I hate it when they repeat letters.

1:31:56

I hate it pretty messed up

1:31:58

when they repeat. I hate that So that

1:32:00

was it electric audio. Yeah. Yeah so

1:32:05

We have a problem because we only have one New York

1:32:07

Times subscription between the two of us Lisa and I we

1:32:09

both like to Do wordle so usually I have to do

1:32:11

it In incognito mode

1:32:14

so that I know because she can't do

1:32:16

it now because I did it logged in

1:32:18

so she has to do it incognito Anyway,

1:32:21

they're really so sorry Lisa. Well, you did it

1:32:23

for the show. What else did they learn? What

1:32:26

else do they have people like holiday

1:32:29

words party heart bunny and ghost? What

1:32:33

does this teach us about humankind

1:32:35

the top opening words that jumped

1:32:37

in popularity Christmas Eve, Mary Christmas

1:32:40

Day, Mary gifts and peace new

1:32:43

year's. These are actually smart because they do

1:32:45

do seemed word. Oh, they do often Oh,

1:32:47

so it is. Yeah, sometimes they do. Oh

1:32:50

On Coronation

1:32:53

day Charles 3rd and Camilla May

1:32:55

6th crown and royal were the

1:32:58

best the most guest top words

1:33:01

I don't change my once you get a good first

1:33:03

word. You shouldn't change it Yes,

1:33:05

same I don't because I'm always chasing that high

1:33:07

of I don't play wordle that often I'm talking

1:33:09

I'm talking a big game like I do. I

1:33:12

forget about wordle for months on time But I'm

1:33:14

always chasing that high of when I put in

1:33:16

my opening word there Could it be that that's

1:33:19

the word of the day? That would

1:33:21

be fantastic. It's never happened. Here's an

1:33:23

amazing one It's like a hole-in-one more

1:33:25

people solve wordle on their first guess

1:33:27

then can be explained by chance One

1:33:32

game in every 250 a reader gets the answer right

1:33:34

on the first drive you ever got it on the

1:33:36

first try Not

1:33:39

I know that was their their phones

1:33:41

are listening to the New York Times

1:33:45

I do think people are cheating. Yeah, that's

1:33:47

Google It's got to

1:33:49

be good. Okay. Oh, they cheat because

1:33:51

you can't look at the word a land

1:33:53

you your word Oh, it's Google the word

1:33:55

will answer. Oh, that's terrible Yeah,

1:33:58

it says some maybe

1:34:00

re-entering a solution they found on a different

1:34:02

device to maintain a streak. Oh, you dare

1:34:04

that. Or to test a technical issue.

1:34:06

Could be that. Could be that. Others

1:34:10

may have had the answer spoiled or yes, may

1:34:12

have looked it up. Slate

1:34:15

and Stair are on the rise. Crane

1:34:17

is getting less popular. Have

1:34:21

you pissed off a whole bunch of people now

1:34:23

who probably, I guess, spoil the world for them.

1:34:26

Yeah, we really, there were a lot of people in

1:34:28

the chat saying, spoilers for wordles. Close

1:34:30

your eyes. I mean, that's

1:34:32

the thing. By the time this posts. Right.

1:34:34

It'll probably be tomorrow. You can't play the

1:34:36

game. It'll be a different word. Right. The

1:34:39

hardest words to solve, start with

1:34:41

a J, N and Y, and

1:34:43

have a double letter. Jazzy

1:34:45

was a very hard one. Oh, Jazzy

1:34:48

is just difficult. I was pissed when

1:34:50

it was Jazzy. That pissed me off.

1:34:53

Joe S. B. N. O. just did a great one. Oh,

1:34:57

yeah? Joe has one for us.

1:34:59

In our Discord. Let me look. Joe

1:35:01

S. Pizzito's got a wordle for

1:35:03

us. Started

1:35:05

with tears, then he wrote Jeff, and

1:35:08

then truly, and then hates this

1:35:11

crap. You

1:35:15

should post that on your Twitter there,

1:35:17

Jeff. That's a good one. Yeah. There's

1:35:19

another one above. You can also, by

1:35:21

the way, and people hate

1:35:23

this, put your wordle

1:35:25

results up on Twitter. That's what I

1:35:28

hate. Yeah. I could not possibly care.

1:35:31

That is pretty bad. And it was at the beginning, it

1:35:33

was just awful. It was a constant. What

1:35:39

else did the media ask? Well, since we're on

1:35:41

light breaks, I have found, I think,

1:35:44

Hank's perfect match

1:35:47

in life. Oh, well, he's looking for that. You're

1:35:49

talking about my 78 and 79. My

1:35:52

son, Saul Hank, who is a TikTok-

1:35:54

Jeff, are you doing a matchmaking services?

1:35:56

He's doing a matchmaking. For Hank. For

1:35:58

Hank is available. This

1:36:01

is a tick-tocker with similar tastes. She's

1:36:10

putting bacon. I've done this. I've

1:36:12

baked my bacon on a little

1:36:14

parchment. There's some brie. Yeah, well that's the problem

1:36:16

with that. Peppers. No,

1:36:19

no, he's not going to like this. It's too slow. Oh,

1:36:22

she's going to roast the peppers. She's classy.

1:36:24

A little feta cheese and olive oil. She's

1:36:26

going to put that all together. Oh, God. It's

1:36:29

called the red bowl. I've already lost interest.

1:36:31

This is a bit slow. Oh, come on.

1:36:34

It's a minute and 11 seconds. Forget it.

1:36:36

Goodbye. Oh, geez. She's

1:36:38

a reject. Let's see what the other ones. That's

1:36:40

the same one. Chicken

1:36:42

sandwich. Can I show you? Look

1:36:45

at this. Can I show you

1:36:48

what Saul Hank does with

1:36:50

a scene? Oh, okay. Wait

1:36:52

a second. She has an ASMR hashtag. She's

1:36:54

a totally different thing. Oh, she's

1:36:57

making sounds. She's... her TikToks

1:36:59

are about to sink. This is what

1:37:01

a TikTok should be. There

1:37:03

we go. Sizzle, goo,

1:37:07

yum, yum, yum. Dip

1:37:09

it, taste it. That's soup, by the way. I don't know

1:37:11

if that's in the shot. I'm going to hand it over.

1:37:14

That's how a TikTok should be. This is really good. This

1:37:16

is kind of cute. This is very good. This

1:37:19

is that long-sink piece, but how

1:37:21

many seconds? That's

1:37:29

how you make a

1:37:31

good... It's kind of a

1:37:34

violent ASMR. It is, yeah. Notice,

1:37:37

by the way, TikTok has changed its

1:37:39

ways a little bit. They

1:37:41

now put the date on the

1:37:43

TikTok, not the views. You have to... Oh.

1:37:46

The views are still there. Yeah, but they're not in the same

1:37:48

spot and they're kind of hidden. Here

1:37:53

is an 11 million view salt

1:37:56

hank that the title is,

1:37:58

I desperately need a hug. don't

1:38:06

you love him ladies he's

1:38:09

available don't take

1:38:11

it a miss that he lives with

1:38:13

his mom it's not he does it

1:38:15

for her who doesn't who doesn't yeah

1:38:19

I don't probably why

1:38:21

did I got a question for you guys yeah

1:38:23

so whatever anything is it all crispy they have

1:38:25

to run the knife over it everybody does why

1:38:28

is that it's a sound yeah it's

1:38:30

a video and they want to show that it's

1:38:32

crispy in a way that is a

1:38:34

see audio you should ask her she's

1:38:36

brilliant she knows she knows she know

1:38:38

that as someone who loves crispy food

1:38:40

and also went through a brief period

1:38:42

where I really liked hearing

1:38:44

that sound I know this you

1:38:47

grew out of it maybe it was a brief

1:38:49

period I'm not really I'm not a big

1:38:51

I mean I just don't use headphones to

1:38:53

watch tiktoks or anything and

1:38:55

so why would I listen

1:38:58

to ASMR 20 billion dollar

1:39:00

acquisition of figma up in smoke adobe

1:39:02

is gonna pay figma a billion they

1:39:05

were concerned because the regulators in the

1:39:07

UK the US and elsewhere were

1:39:10

looking at scans they

1:39:13

proposed for instance the CMA in

1:39:15

the UK proposed that adobe if

1:39:18

they were out of the have this merger should get

1:39:20

rid of all the products that overlap which

1:39:23

would mean by my

1:39:25

Photoshop by by illustrator

1:39:28

you know just a few small little things

1:39:30

adobe said we're not gonna do that and

1:39:32

so that deal which was one of the

1:39:35

largest ever and

1:39:37

and certainly worth a lot more than figma was

1:39:39

actually worth on paper is

1:39:41

is done it's over it had to

1:39:43

pay a one billion dollar breakup fee

1:39:45

yeah yeah it's

1:39:47

got to be I mean rough week

1:39:50

to be a figma employee you

1:39:52

thought for the last whatever

1:39:54

it was evening mommy you're

1:39:56

gonna be a millionaire

1:39:58

hope that in I put

1:40:01

a pool and I put down a down payment on

1:40:04

a pool in my backyard. Then

1:40:06

I got a subscription to the Jelly of the Month

1:40:08

Club. That's not right. Now

1:40:11

here you are, penniless. Penniless.

1:40:14

Do you know what pig butchering is? Yes,

1:40:18

it's a type of scam.

1:40:21

You get them all the time in your text messages

1:40:23

where somebody just says, hi, I got

1:40:25

one the other day. I'm going to be

1:40:27

in town in a couple of weeks. You want to get together? It's

1:40:33

called pig butchering because it refers

1:40:35

to the fattening up process where a

1:40:38

scam role put in potentially months and

1:40:40

months of work trying to gain your

1:40:42

trust before then pivoting to

1:40:44

the scam. Federal

1:40:47

prosecutors have indicted

1:40:50

four people arrested two to disrupt

1:40:54

a so-called pig butchering seem to cost victims

1:40:56

more than $80 million. Four

1:41:01

people. $80 million. The sad

1:41:03

thing is that money usually comes from

1:41:06

retirees. It's often their entire

1:41:08

life savings. Lu

1:41:11

Zhang, Justin Walker, Joseph Wong,

1:41:14

California resident allegedly conspired with Illinois resident

1:41:16

High Long Ju to launder the illicit

1:41:19

proceeds of their scam. Zhang

1:41:22

and Walker have been arrested. I guess

1:41:24

the other two are at

1:41:26

large. It comes

1:41:28

from the Chinese phrase, sha ju pan,

1:41:32

cold messaging victims, building

1:41:34

a rapport, and

1:41:36

then a variety of scams. Did

1:41:39

either of you guys see the fantastic New

1:41:42

York Times piece that came out

1:41:44

a couple of days ago, an

1:41:46

interactive about essentially seven

1:41:48

months inside one of these

1:41:51

online scam labor camps? Because

1:41:54

often the people that

1:41:56

power these pig butchering

1:41:58

scams are... people

1:42:01

who've basically been kind of abducted

1:42:04

from their home countries and put

1:42:06

in a camp and forced

1:42:08

to do this or else they will

1:42:11

you know be severely

1:42:13

beaten. The New York Times got

1:42:15

a message from a man who had

1:42:18

been a had thought he

1:42:20

was leaving China for a

1:42:22

legitimate job and spent a lot of time kind

1:42:24

of talking to his new employer once he gets

1:42:26

over the border of wherever he was going gets

1:42:28

taken to the scam compound

1:42:30

in Myanmar and he after

1:42:33

a couple of months in there you know

1:42:35

tries to get them to let him out

1:42:38

they won't they decided to put him he

1:42:40

was like an accountant or something by trade

1:42:42

put him in charge of the accounting eventually

1:42:44

once he gets access to his phone he

1:42:46

starts taking photos of everything

1:42:48

inside this scam center and all

1:42:51

the financials and sends them to

1:42:53

the New York Times and other places

1:42:55

as well. So it's a phenomenal look

1:42:58

inside one of these camps

1:43:00

and also specifically how the

1:43:02

business works it is a

1:43:06

huge operation. They're basically it's slave

1:43:08

labor they can't leave. So is

1:43:10

it government run and owned? No

1:43:15

this one is kind of a

1:43:17

camp that I believe was in

1:43:21

a certain part of Myanmar that was

1:43:24

by a couple of different local gangs

1:43:26

that kind of operated as their own

1:43:28

little government entities. Look at all the

1:43:30

phones they have attached

1:43:33

to Iraq because you have to have different

1:43:35

phone numbers and different phones. And

1:43:38

so they would have a lot of the people

1:43:40

in that camp they would have

1:43:42

to go on those phones every single

1:43:45

day and scroll through their WeChat feeds

1:43:47

of all the different phones interact

1:43:49

like normal so that they could get

1:43:51

around WeChat's anti-spam measurements that would

1:43:53

be one of their daily tasks. Wow.

1:43:58

And what's interesting is this is us all. I

1:44:00

mean we've all received these great

1:44:19

lengths to avoid asking their families for help

1:44:22

or reporting the fraud to police out of

1:44:24

fear of being accused of infidelity. The

1:44:27

group had taken in more than 4.4 million

1:44:29

dollars in five months from

1:44:32

214 victims. It's

1:44:35

so sad. It's just

1:44:37

terrible. And of course the people

1:44:39

who are doing this are

1:44:42

not only enslaved

1:44:45

but they're tortured to some degrees.

1:44:48

I mean this is just awful. Yeah

1:44:51

very powerful, very powerful piece.

1:44:55

And you know the way to stop it is not to be

1:44:57

suckered. So this is a good time

1:44:59

of year by the way because you're gonna see family

1:45:02

and friends, people who aren't as technically sophisticated as you

1:45:04

or listeners are. Don't forget to tell them

1:45:06

about stuff like this. Be

1:45:09

proactive. Say you know if you get a message

1:45:11

from somebody you don't know saying hi don't respond to

1:45:14

it. Don't get... Or you

1:45:16

know if you get to something where

1:45:18

someone you've met in the internet, if

1:45:20

someone you trust is asking you to send the money, reach out

1:45:22

to me. You know tell your loved ones like that

1:45:25

you'll always be there to just check.

1:45:27

Give a quick once over. Make sure that it's

1:45:29

a situation where they don't feel embarrassed talking

1:45:31

about it. Yeah my car is almost at the

1:45:34

Niagara Falls one with our son.

1:45:36

Oh I'm stuck in Niagara Falls. I lost my

1:45:38

wallet and I need a car fare to

1:45:41

get home. That kind of thing. Yeah. Yeah. That's

1:45:44

the thing. Maybe they know enough to

1:45:46

actually make it sound credible. Well no

1:45:48

that goes back to our ad story

1:45:50

at the very beginning. No. We quizzed

1:45:52

my father who was

1:45:54

almost headed to Walmart to buy you know

1:45:56

cash cards there. Well

1:45:59

he didn't call me pop up. He called me, you

1:46:01

know, grandpa. He never calls me grandpa. You know,

1:46:03

kind of afterwards should have known. He should have

1:46:05

known. Yeah. But the

1:46:07

fear is so great you look over all

1:46:09

those things because you don't want to be guilty and say, oh

1:46:11

my God, I didn't rescue my grandson. Google

1:46:16

has decided it doesn't need sales people.

1:46:18

You know how many people work at

1:46:21

Google selling ads? Sales. This

1:46:23

amaze me. This one stat just amazed me. 30,000

1:46:28

people work in the ad sales

1:46:30

unit at Google. That's

1:46:33

not all sales people, but still

1:46:35

30,000 people because of that revenue.

1:46:38

But not anymore because... That's after all these cuts. That's

1:46:41

right. Because machine learning techniques

1:46:45

and artificial intelligence can replace 90% of

1:46:47

these people. No, not

1:46:49

90. No. The

1:46:52

planned reorganization comes as Google is relying more

1:46:54

on machine learning techniques to help customers buy

1:46:56

even more ads on its search engine. So

1:46:58

that's what the sales people do. Let me

1:47:00

help you buy some ads. This

1:47:03

is kind of pig butchering in another form

1:47:05

or fashion actually. Well, what it's amazing about

1:47:07

this so much is that we... I

1:47:11

think we still probably have a presumption that ad

1:47:14

buying was automated a lot of the years.

1:47:17

Right. Well, it wasn't. It's

1:47:19

highly manual in sales. Maybe

1:47:24