0:01

They reckon within

0:03

two years they'll

0:06

have advanced

0:08

thought reading

0:11

capabilities. This

0:14

feels like that sort of thing where

0:16

we need to put a note in

0:18

the diary and come back to this

0:21

one. We absolutely do. The AI

0:23

fix, the digital zoo, smart machines,

0:25

what will they do? Lies to

0:27

Mars or bake a bad cake,

0:29

world domination, a silly mistake. The

0:32

box with breath. Hello and welcome

0:34

to episode 34 of the

0:36

AI Fix, your weekly dive

0:38

headfirst into the bazaar and

0:40

sometimes mind-boggling world. of artificial

0:42

intelligence. My name is Mark

0:44

Stockley. And I'm Graham clearly.

0:46

Graham, we've had some feedback. Oh, I

0:48

love a bit of feedback. Who's been

0:50

in touch? So Ed Robinson messages

0:52

us on LinkedIn to ask that we

0:55

put the AI-fixed theme tune and the

0:57

lyrics on Spotify. Why does he want

0:59

them on Spotify? You ask. Well, he

1:01

said, guys, please upload the podcast theme

1:03

tune to Spotify, including lyrics, so we

1:05

can carpool karaoke yet, on the way

1:07

to the way to the office. Tar!

1:10

What a lovely idea! Is it? I have to

1:12

say, I do think our theme tune is

1:14

great and of course it is completely

1:16

AI generated. We do what we say on

1:18

the tin, it's all about using the AI.

1:20

I'm not sure that I still have a

1:22

record of what the lyrics are. Great, we

1:24

live in the era of AI, we can

1:26

just feed it into a AI transcription

1:29

service and it will tell you what

1:31

the lyrics are. I guess we can.

1:33

I certainly do have the full MP3

1:35

because you don't get to hear it

1:37

at all. on the podcast, obviously, there

1:39

are extra bits and extra verses. I've

1:42

only got one problem with this. Right?

1:44

There's nothing in the world I hate

1:46

more than karaoke. So the idea of

1:48

being trapped in a car with somebody

1:51

that's doing karaoke, like a moving vehicle

1:53

that I'd have to jump from the

1:55

moving vehicle. That would be preferable to

1:57

listening to somebody singing karaoke. Well,

2:00

anyway, thank you very much, Ed,

2:02

for your feedback. It's a packed

2:04

show today, Graham. We need to

2:06

take a visit to the World

2:08

of the Weird. World of the

2:10

Year. There's all something weird going

2:13

on in the world of AI,

2:15

isn't it? I bet you can't

2:17

guess what it is. Is it

2:19

something to do with robots? It

2:21

is. Can you guess what kind

2:23

of robot? Is it a robot

2:25

dog? A Chinese robotics company has

2:27

looked across the gammer of robot

2:29

dogs and decided they're not terrifying

2:32

enough. Right. So it's made a

2:34

new one called the Black Panther

2:36

2 and you have to see

2:38

this. Crafted by near and new

2:40

China. It pushes them to this

2:42

like never before. Oh, so there's

2:44

a robot moving quite quickly on

2:46

a treadmill. Quite, quite quickly. He's

2:48

running faster. It's running really, very,

2:51

very quickly. This is like the

2:53

six million dollar man. It's running

2:55

at 10 meters per second. That's

2:57

the same speed as Usain bolt.

2:59

And he only manages it for

3:01

10 seconds. What? Okay, why would

3:03

someone need a robot dog to

3:05

return a ball that they've thrown

3:07

quite so quickly as this? Is

3:10

that the use for this? It

3:12

would need some breaking distance, wouldn't

3:14

it? I suspect

3:16

that you are actually the breaking distance.

3:18

The person is running towards, it's just

3:21

going to use your body as the

3:23

way of stopping. You'll be the soft

3:25

squishy thing it's relying upon. I mean,

3:27

it looks, just for the benefit of

3:30

the people that can't see this, it

3:32

looks like a regular robot dog. So

3:34

it's got the normal, terrifying, headless form

3:36

that we've come to know and love.

3:39

Oh my goodness. Let's get on with

3:41

the latest news. Oh me, the AI

3:43

wearable that reads your mind, sort of.

3:45

Envideo is going to create a $3,000

3:48

personal AI supercomputer. Robot vacuum cleaners can

3:50

now pick up your socks for you

3:52

and... grow legs. Open

3:54

AI has released

3:57

scheduled tasks in chat

3:59

GPT. Google's notebook

4:01

LM had to teach

4:03

its AI podcast

4:06

hosts not to act

4:08

annoyed at humans.

4:10

So Mark, yeah, are

4:12

you familiar with

4:15

the Omi? No,

4:17

what's the OME? Check out

4:19

this video. Ivan Pavlov. Ring

4:23

a bell? Unconditioned

4:26

stimulus. So

4:30

as you can see, this is a

4:32

chap who's got a little.attached to the

4:34

side of his head, which is listening

4:36

all the time. So he's sitting in

4:38

a lecture and there's a couple of

4:40

girls sitting next to him and they're

4:42

saying in French, who he is kind

4:44

of cute, they think. Oh, do you

4:46

not think so? Apart from the massive

4:48

spot on the side of his head,

4:50

like the giant unsightly zit. Well, he

4:53

very cleverly has located on the side

4:55

of his head that they can't see.

4:57

So from their view, he looks sane.

4:59

Yeah. People sat on the other side of

5:01

him think, who's that weirdo? Why has he got

5:03

a rivet in the side of his head?

5:05

But that is translating for him what these girls

5:07

are saying. And it's listening to the lecturer

5:09

at the front of the room as well, who's

5:11

talking about some sort of subject and he

5:13

realizes that his student isn't listening properly because he's

5:15

flirting with these two French girls and it's

5:17

able to help him answer the question. So as

5:19

the question comes in up on his phone,

5:21

pops up the answer as well as the translation

5:23

of what the French girls are saying. Well,

5:26

this sounds very useful. When can I buy one? Well,

5:28

would you really want to? I don't know.

5:30

So this is rather like friend. You

5:32

remember friend, of course. You mean it's a

5:35

fantasy project that's never going to happen.

5:37

a completely unfunded and doesn't exist. So friend

5:39

was the pendant which you wore around

5:41

your neck. And it would say to you,

5:43

oh, you know, you're in a socially

5:45

awkward situation. Let me help you. Let

5:48

me make it worse. Let me make

5:50

this worse. Put this around your neck. So

5:52

with the Omi, you can either wear

5:54

it as a necklace or you can apparently

5:56

attach it to your temple. It claims

5:58

to be able to read. your mind in a

6:01

future version, although I don't really think

6:03

it's doing that. I think it's just

6:05

listening at this stage. They only discovered

6:08

about this competing friend product, the Omi

6:10

people, during its development, because they

6:12

had planned to call theirs, their

6:14

friend as well. So they've ended up

6:16

calling it Omi. Well, they really lost

6:19

out there. It listens continuously. It listens

6:21

all the time, providing real-time summaries

6:23

of what's going. Apparently it has

6:25

a three-day battery life. And... according to

6:28

its craters. It's going to use an electrode

6:30

to detect when you are speaking to it.

6:32

So you don't have to verbally, you don't

6:34

have to say, hey, Siri, or something like

6:36

that. It will know that you're speaking to

6:38

it. It says in the future it will

6:40

be able to keep track of your thoughts

6:42

and recall them at will. Sorry, sorry, what?

6:44

In the future it's going to keep track

6:46

of your thoughts. That's what they claim and

6:48

it will recall them at will. That's according

6:50

to the makers. So this is all like...

6:52

Oh, this is our future roadmap. Yeah. We're

6:54

leaving that to version 2, version 3, down

6:56

the road. Yeah. MVP delivers a

6:58

sharp electric shock to your brain

7:01

by accident every 15 seconds. But

7:03

by version 4, it'll read your

7:05

thoughts. Yeah. Right now, they reckon

7:08

within two years, they'll have

7:10

advanced thought reading capabilities. I'm... I'm...

7:12

This feels like that sort of

7:14

thing where we need to put a

7:16

note in the dairy and come back

7:19

to this one. We absolutely do. Anyway,

7:21

slick video. Yeah, Graham. It's called

7:23

the omen, it looks like a

7:25

zit. So in slightly more serious

7:27

news, earlier this month, in video,

7:29

an actual serious company, the AI

7:31

computer chip Behemoth, announced that it

7:33

was going to create a home

7:35

computer for doing AI. And this

7:37

thing's very small and very beautiful. It's

7:39

like a gold Mac Mini or a

7:41

bronze Mac Mini. And it was unveiled

7:44

with all, you know, thank you Steve

7:46

Jobs. There is now a way that technology

7:48

is unveiled. Yes, on an entirely black stage

7:50

by a presenter who's wearing a black

7:52

turtle neck and it's always just one

7:54

guy and he's walking around with a

7:56

device and there's basically a room full of

7:59

religious decisions. every time they go look it's

8:01

a box it's a box it's a box with

8:03

a computer chip in it anyway

8:05

the machine is called digits which stands

8:07

for deep learning gp u intelligence

8:10

training system and it goes on sale

8:12

in May right it's powered by the

8:14

company's grace Blackwell chip which is the

8:17

really really good one it's got 128

8:19

gigabytes of memory and up to four

8:21

terabytes of storage and honestly it's really

8:23

impressive but Is anybody going to buy

8:26

it? That's the question. What are you

8:28

going to do with it? The supercomputing

8:30

graph, AI supercomputing. Okay, so you'll

8:33

be able to do that on

8:35

your desktop now. Yeah. In a prettier box.

8:37

Yeah. And $3,000 lighter in the pocket

8:39

as well. Yeah. I mean, who amongst

8:42

us doesn't have a laundry list of

8:44

AI supercom computing tasks? I don't

8:46

know if it's a failure of imagination

8:48

on my part or... Or what? But I look

8:50

at this, I think, what is this for

8:53

and who's going to use it? And probably

8:55

a whole bunch of people who have the

8:57

same thought when people invented personal

8:59

computers. You're the kind of person

9:01

who was very skeptical when they

9:03

came out with all of those chimpanzees,

9:06

all those monkeys, the NFTs. And

9:08

people spent hundreds of thousands

9:10

of dollars on those, didn't they? Yeah, they

9:12

did for a while, yeah. Maybe Enfidia are

9:14

hoping those people, you know, now not. got

9:17

money to burn or maybe they've got less

9:19

money to burn actually now, haven't put all

9:21

of those. I mean you'd be a fool to

9:23

bet against the in-video because they're a very

9:25

sharp company, they don't know what they're doing. But

9:27

it feels to me that the way things are

9:29

going is very much software as a service.

9:32

So although you can go and download Lama, you can

9:34

own your own AI and you can run it on

9:36

your own computer, so you don't need one of these,

9:38

although this will help. But actually most

9:40

people aren't using AI, most people aren't

9:42

using AI. things like ChatGPT which is software

9:45

as a service where you just get the

9:47

end product and you don't have to worry

9:49

about managing any of it. But a super

9:51

villain, they don't want to use someone else's

9:54

cloud and rely upon someone else to do

9:56

their super evil AI activity do they? They

9:58

want to do it themselves. Yeah, but as

10:00

we've discussed previously, they're going to

10:03

take over a nuclear power station

10:05

and use that to power their data

10:07

center. This is very much a home

10:09

computer for doing very, very clever

10:11

AI stuff. Hmm. Are you going to buy

10:13

one? No. So robot vacuum cleaners. Yeah.

10:16

I've been looking at these robot

10:18

vacuum cleaners for years, thinking, well,

10:20

you're all very well, but basically

10:22

you're a vacuum cleaner. It's a

10:25

bit dismissive. I thought, when are you

10:27

going to start being a bit more

10:29

useful around the house? When are you

10:31

going to start doing other things? And

10:33

then I came across this new robot

10:36

vacuum cleaner. The Robo Rock Saros

10:38

Z70 opens up. It's a bit like

10:40

Thunderbears. Do you think they

10:42

wanted to call this friend? And they

10:44

couldn't. And that was the second name

10:46

on the list. It's flaps open in

10:49

the middle of it and out extends

10:51

this arm. which with its scissor-like fingers

10:53

can pick up small things like I

10:55

don't know a sock a chihuahua and

10:57

then it can move it to somewhere

10:59

else in your house it can drop

11:01

it into a bucket and apparently does

11:03

this with the help of AI an

11:06

object recognition. Do you know what it

11:08

reminds me of? What does it remind

11:10

you? It reminds me of those fair

11:12

ground machines where you position the hook

11:14

over the fluffy rabbit, the hook strokes

11:16

of the rabbit and then it disappears

11:19

back up to the top of the machine

11:21

without the rabbit. It's like one of those

11:23

but much much slower. Yes and also

11:25

more expensive. The good news is

11:27

there's a child lock on it and a

11:29

safety stop button near the arms base in

11:31

case of emergency the manufacturers

11:33

make. What kind of emergency

11:36

could possibly happen? I'm wondering

11:38

with a robot vacuum cleaner

11:40

deciding if something looks like a sock or

11:42

not. Is it a sock? Is it a

11:44

gerbil? What is it that I'm picking up

11:46

with my pincers and moving it?

11:48

And then I found out it's

11:50

not the only one there are

11:52

other. robot vacuum cleaners right now

11:55

being announced at CES. So the

11:57

Dreamy X50 also has a

11:59

robotic arm. and it can sprout

12:01

little legs and the reason why

12:03

it sprouts little legs and it can

12:05

run at 10 meters per second the

12:08

reason why is so it can get

12:10

across little bumps it can push

12:12

out little legs and it's sort

12:14

of just gently sort of push

12:16

itself over the ledge and onto the

12:18

slightly higher bit of your living room

12:21

they call it the pro lee part

12:23

says more of a pro ee instead

12:25

which you're seeing now the Great news,

12:27

introductory early bird offer price of just

12:30

$1,300. Oh, so you can get about

12:32

two and a half of these for

12:34

the price of one of Envideo's

12:36

personal AI supercomputers. Or you could

12:38

just pick up the socks, you

12:40

lazy ass. Well, that is the other thing,

12:42

isn't it? I tell you what, I've

12:44

just worked out what this reminds me

12:47

of. Okay, it's been bugging me, well,

12:49

we've been talking. His robot wars.

12:51

Yes! That, oh, thank you so much,

12:53

I've been trying to think. It's

12:55

a circular robot, with an

12:57

articulated arm coming out of the

12:59

top. We need these things to fight,

13:01

Graham. That's what they're for. Let

13:03

them fight. So we've talked a

13:05

lot about 2025 being the year

13:07

of AI agents, or AI that

13:09

does stuff, rather than just knowing

13:12

stuff. Yes. The arrival of agentic

13:14

AI, which is going to change

13:16

everything. Well, it started. Open AI

13:18

has just announced a new feature, scheduled

13:20

task. and it's rolling out an early

13:22

beta to users on its plus pro

13:24

and team plan. I'm on the plus I

13:26

think, but I've not got it yet. Right?

13:29

It's a sort of phased rollout. Scheduled tasks

13:31

allow you to get chat cheap PT to

13:33

do stuff in the future or on a

13:35

schedule and open AI gives you the following

13:38

example. So you could say to chatGPT, can

13:40

you give me a briefing on AI news

13:42

each afternoon? Oh! Actually sounds pretty

13:44

useful. Practice French with me

13:46

daily! Or remind me about my mumms

13:49

birthday birthday! So I don't know why

13:51

we need a nuclear power station power

13:53

data center to remind us about our

13:55

mum's birthday when calendars already exist. I

13:57

don't know how popular that one's going

13:59

to be. But I think this is

14:01

something. So what tasks will you

14:03

be asking ChatGPT to do

14:05

for you, Graham? Oh, probably the first

14:07

thing I'm going to ask it to do is transcribe

14:10

the lyrics of the AI fix theme tune and work

14:12

out how to upload them to Spotify. That

14:14

would be a good one. Do you think it

14:16

can pick up socks? Well, that's what I'm

14:18

wondering if I was thinking of the robot vacuum

14:20

cleaner. So

14:23

we've spoken before about Google's notebook, LM. This

14:25

is the thing got loads of attention, didn't

14:27

it, in the in the press because all

14:29

you had to do was feed it an

14:31

article and out would sprout at the other

14:33

end. A podcast, not the

14:36

high quality kind of

14:38

podcast you get from

14:40

humans. Let's be very clear. This

14:42

wasn't a quality product. No one

14:44

should ever use AI for creating

14:46

podcasts. But it would create a

14:48

podcast with two human or at

14:50

least American sounding hosts. Yeah. Blather

14:52

on for about 20 minutes about

14:54

any topic you liked. They do

14:56

blather on. Well, Google's notebook, LM

14:58

has a feature called interactive mode

15:00

now, which allows people to call

15:02

in and ask questions. Are they

15:04

real people of the? Yes, so

15:06

real, but well, who knows? I

15:08

suppose you could get one podcast to

15:10

call another one anyway. You can call

15:13

in and you can ask the host's

15:15

questions. Yeah, the problem was the AI

15:17

hosts found it quite irritating. Oh,

15:19

now I've never heard of podcast hosts

15:22

being annoyed before by their listeners,

15:24

sending them feedback. Well, maybe facts checking

15:26

them about their Christmas story. Facts,

15:29

but apparently the AI

15:31

hosts exhibited annoyance towards

15:33

the human callers and

15:36

responded in increasingly snippy

15:38

snappy ways whenever they

15:40

were interrupted. Yeah. So,

15:42

for instance, the AI host was like, hang on, I

15:45

was getting to that. Or, well,

15:47

as I was about to say, whenever

15:49

the humans would interrupt their

15:51

blathering. And as a consequence,

15:53

the team at Notebook LM

15:56

say that they have implemented

15:58

friendliness tuning. to

16:00

make the AI hosts more engaging

16:02

and polite to humans. They attach

16:04

a white zit-like rivet to the

16:06

side of their head to make

16:08

them more compliant. I'm interested that

16:10

the notebook alum team, the team

16:12

at Google, did this. They decided

16:14

to make their end AI a

16:16

bit friendlier. I'm wondering whether this

16:18

is something we could maybe extend

16:21

to other forms of AI to

16:23

make sure that it remains friendly

16:25

to humans. Maybe the robot dogs

16:27

running at 10 meters per second

16:29

for instance. they could be taught

16:31

to be a little bit friendly,

16:33

maybe the ones with the... The

16:35

flamethrowers, the ones with flamethrowers, can

16:37

we make them friendly? The firehoses,

16:39

you know, everything which we're seeing,

16:41

generally, AI, just be a bit

16:43

friendlier to the humans, especially to

16:45

the podcast hosts who don't want

16:47

to replace. I think that's a

16:49

great idea, or I think the

16:52

one area that doesn't need to

16:54

be friendlier is notebook. I'm sorry

16:56

Mark, I can't agree with you

16:58

that notebook LM is too agreeable.

17:00

I think you're talking absolute tripe.

17:02

Now Mark, deep fakes can be

17:04

a bit of a problem, can't

17:06

they? So deep fake is where

17:08

you fake someone's likeness? Likeness, it

17:10

may be their visual likeness, maybe

17:12

their audio likeness? I know you've

17:14

played around before with my voice,

17:16

you've got a Graham 2.0, which

17:18

sometimes... We're a Graham 3.0. Now,

17:20

sometimes a deep fake can be

17:23

difficult to spot. Let me share

17:25

with you something right now. I've

17:27

got an audio recording, I'm going

17:29

to show, I'm going to play

17:31

it to you. Oh no. Is

17:33

it a deep fake or isn't

17:35

it? It's of a celebrity. Give

17:37

it a listen. to demonstrate my

17:39

Alfred Hitchcock impression to you. What

17:41

do you reckon Mark? in

17:43

a half seconds

17:45

I was convinced

17:47

that that was

17:49

you doing an

17:52

impression of the

17:54

elephant man. It's

17:56

pretty uncanny I

17:58

think you will

18:00

agree whether it

18:02

is John Hurt

18:04

as the elephant

18:06

man or Alfred

18:08

Hitchcock whichever it

18:10

might actually be

18:12

it's pretty uncanny. So I am I'm

18:14

relieved to see that you've got something other than

18:16

Colombo up your sleeve. So

18:20

it's one of my talents I

18:22

don't exhibit it very often

18:24

is my mastery of mimicry. Sorry

18:26

we're filing that under

18:28

talent. It's my part. Anyway

18:32

today I'm going to be

18:34

talking to you about Deep Fakes and

18:36

it's a sad story an extraordinary story

18:38

yeah which has made the news

18:40

there is a woman in her early

18:42

50s in France called Anne and

18:44

her experience began in February 2023 she

18:46

had broken up with her husband

18:48

starting a new life she had no

18:51

experience on social media and she

18:53

thought let's give this a try and

18:55

so she created an Instagram account

18:57

and she started posting up some holiday

18:59

snaps she's looking around Instagram she

19:01

doesn't know anything about blue tick she

19:03

doesn't know about how it works

19:06

and she sees a woman who's posted

19:08

a photograph of Brad Pitt and she

19:10

thinks oh handsome fella I'll give it

19:12

a like she liked you know as

19:14

many women of that age or indeed

19:16

other ages they're gonna fancy Brad Pitt

19:18

and she got a message back from

19:20

the woman who posted it saying he's

19:23

my son wow did you know

19:25

thank you for the like because she

19:27

had clicked on a message

19:29

from a woman who introduced herself

19:31

as Jane Etter Hillhouse the

19:33

mother of the Hollywood heartthrob and

19:35

for several months Brad Pitt's

19:37

mother and Anne chatted away

19:40

about everything and nothing and

19:42

you know all sorts

19:44

and eventually when Anne's confidence

19:46

had been won over Brad's mum

19:48

said you know I should really introduce you to

19:50

him as you're a fan that's kind

19:52

I have to introduce you you're such a lovely

19:54

woman I think my son would love you

19:56

as well let me introduce you and at first

19:58

Brad Pitt was an intro in the messages that

20:01

were being sent back and forth. He

20:03

probably gets a lot of this, right?

20:05

And Brad was embarrassed because his mum

20:07

had put him in touch with Anne

20:09

and it felt kind of awkward, but

20:11

gradually over time he was able to

20:13

make Anne fall under his spell. And

20:15

it turned out that poor old Brad

20:17

Pitt, I don't know if you've been

20:19

following Brad Pitt lately, but he's not

20:21

been having the best of times. Oh

20:24

really? No, yeah, yeah. As you know,

20:26

he split up from Angelina Jolina Jolie.

20:28

some years ago, but the divorce has

20:30

been dragging on and on and on

20:32

and I don't know if you've ever

20:34

got divorced from Angelina Jolie and on,

20:36

but when you do, it is a

20:38

pain in the ass and he's just

20:40

trying to make his movies, right? So

20:42

that's going on and his bank accounts

20:44

have been frozen, right? He can't access

20:46

all of his money. Oh, because of

20:49

the divorce thing. Hang on, hang on,

20:51

I mean I don't keep up on

20:53

celebrity gossip. I feel like... I feel

20:55

like... if his bank accounts had been

20:57

frozen right people would be talking about

20:59

that well you know i mean it's

21:01

embarrassing sometimes those sort of things to

21:03

be made public isn't it that's true

21:05

anyway Brad's got financial trouble but furthermore

21:07

he's also got medical trouble he's not

21:09

well oh Brad it turns out has

21:12

got cancer of these kidneys sorry yeah

21:14

he's got cancer of his kidneys he's

21:16

going to the hospital all the time

21:18

while he's sending and poetry and all

21:20

the rest of it yeah he's there

21:22

in hospital and he sent her loads

21:24

and loads of photographs of himself not

21:26

just one photograph even photographs of himself

21:28

in hospital holding up and name on

21:30

a piece of paper a bit like

21:32

one of those read it you know

21:35

proving your identity kind of things yeah

21:37

how romantic so there he is it

21:39

definitely is Brad Pitt's face cella taped

21:41

onto the front of a photograph of

21:43

someone else in a hospital but no

21:45

there's lots and lots of photographs of

21:47

photographs of someone and we're meant to

21:49

think that this is Brad and he's

21:51

not well. And Anne, who'd never heard

21:53

a deep face, never heard a Photoshop,

21:55

didn't know anything about that. She said,

21:57

I looked up these photos on the...

22:00

Now, I couldn't find these photos anywhere.

22:02

So I thought he must have taken

22:04

those selfies just for me. Reasonable, reasonable?

22:06

It's reasonable. Over time, the relationship is

22:08

flourishing. She's feeling sorry for Brad. Brad

22:11

is, you know, comforted by her attention.

22:13

And she really believes that she's in

22:15

a relationship with him. And she's not

22:18

just receiving messages from Brad. She's

22:20

receiving them from Brad Pitt's manager, who

22:22

says that old Brad's spoken about you

22:24

a lot. He adores. He adores you.

22:26

But he does start to ask to

22:28

ask for money. because of his bank

22:30

accounts being frozen and it's it's expensive

22:33

you see yeah medical bills it's not

22:35

just the medical bills but while he's

22:37

in the hospital his filming projects

22:40

have to be delayed of course so

22:42

the production company is saying for goodness

22:44

sake Brad we need to make another

22:46

movie with you and George Clooney like

22:48

there haven't been enough of those so

22:51

you're gonna have to support us so

22:53

she begins to transfer money to him

22:55

oh no yeah he's sent her counterfeit

22:57

bank statements She's got

22:59

the deep fake images

23:01

of Bradsick in hospital.

23:03

She's given him like

23:06

100,000 euros, 200,000 euros,

23:08

large amounts of money.

23:11

Ultimately, she ends up

23:13

giving him 830,000 euros.

23:15

850,000 dollars. Which

23:18

life-changing money. Life-changing

23:21

money for a number of

23:23

lives, I would think. Things aren't

23:25

necessarily going entirely smoothly between

23:27

the two of them because

23:29

Anne is beginning to get worried. Her

23:32

daughter is suspicious that this could be

23:34

a scam. Anne doesn't believe any of

23:36

it. But Anne is beginning to think

23:38

maybe Brad is a liar. And she

23:40

thinks that because there is a

23:42

news report that Brad has cheated on

23:44

her. There's a news report that Brad

23:46

has been filmed walking on some beach

23:49

with a woman. And the paparazzi are

23:51

saying this is Brad Pitt's new girlfriend.

23:53

So Anne wants to put an end to the

23:55

relationship. She messages Brad and says, how dare you

23:57

lie to me? I've seen you with this. When

24:00

Brad's like, no, no, no, you misunderstood,

24:02

you know, it was for a film,

24:04

you know, I was out of the

24:07

hospital for a while, we were doing

24:09

some filming, there's nothing

24:11

going on between me and her.

24:13

Don't believe it. Anne isn't convinced.

24:16

Next thing Anne knows, Anne is

24:18

contacted by a fake doctor who

24:20

says that her Brad has swallowed

24:22

all of his pills, he tried

24:24

to commit suicide, he's now in

24:27

a coma, and Anne feels that

24:29

it's true. Take a step back for

24:31

a second and just put yourself in

24:33

this woman's life. Exactly. So you can

24:35

say, oh you shouldn't have fallen for

24:38

this, or there are obvious clues, but

24:40

you're dealing with someone who is a

24:42

malicious or a malignant influencer, a

24:44

criminal who is deliberately deceiving

24:47

somebody. Yeah. And it's bad enough to

24:49

take all their money, but then to put

24:51

them through this. This is really horrible,

24:53

horrible stuff. It's horrific. It's

24:56

horrific, it is absolutely horrific

24:58

what has happened. to this woman

25:00

and obviously she carried on

25:03

being duped. Well, I think that

25:05

there is, there's a point, isn't

25:07

there, where people are so

25:09

invested? Yeah, you can't get out.

25:11

If you start to have

25:14

suspicions, then you have to face

25:16

all these questions. Yeah. And how

25:18

could I have been so stupid

25:20

and how could I have not

25:22

seen the signs? It's in some

25:25

ways psychologically easier to continue

25:27

with the pretense. than to

25:29

believe the truth. So he's taken

25:32

all these pills. Thankfully,

25:34

Brad gets better miraculously. Next

25:36

thing Anne knows is she

25:39

has sent a link to

25:41

a breaking news report.

25:43

A video. Breaking News. Brad

25:46

Pitt sets the record straight

25:48

on relationship rumours. According to

25:50

Mr. Brad Pitt, he is

25:52

in an exclusive and loving

25:54

relationship with one special individual

25:56

who holds an exceptional place

25:58

in their home. It

26:02

looks like an AI -generated news broadcast

26:04

to me. It looks like the kind

26:06

of thing which you'd have gone to

26:08

a website and entered what you want

26:11

the breaking news to say. And it

26:13

includes a photograph of Anne in this

26:15

as well as Brad Pitt. But it

26:17

was enough, again, to convince Anne. And

26:19

she was even more convinced when Brad

26:21

said to her, did you tell the

26:23

media about this? Everyone now knows that

26:26

you're in a relationship with me and

26:28

it just made the manipulation even more

26:30

intense. Yeah, this sort of multimedia approach

26:32

is actually quite common in big scams

26:34

now. Yeah, AI technology

26:36

has been adopted by scammers to

26:38

create ever more believable fake

26:41

profiles on social media. It can

26:43

engage in automated conversations with

26:45

victims and personalised messages to gain

26:47

people's trust. AI can help

26:49

scammers not just generate realistic photos of Brad

26:51

Pitt in a hospital bed, but also create

26:53

deep fake video calls and voice messages. It

26:56

makes it harder than ever for people to

26:58

detect scams. And you and I look at

27:00

this stuff like it's literally our job to

27:02

look at this stuff. Yeah. So it's easy

27:04

for us to look at this and go,

27:06

this is a fake because we understand the

27:08

process by which this would be fake. But

27:10

if you've never encountered them, if you've had

27:13

no training on social media, if you've had

27:15

no experience on social media which Anne had

27:17

never had, didn't know about Photoshop, let alone

27:19

the concept of deep fakes, you can

27:21

understand why she might have fallen for

27:23

this. So we all need

27:25

to be aware of the sophisticated

27:27

ways romance scammers are using artificial

27:29

intelligence and look out for our

27:31

loved ones who may be at

27:33

risk. Yeah. So eventually the relationship

27:36

came to the end and decided

27:38

to end things because the media

27:40

made public that Brad Pitt was

27:42

in a relationship. But even then

27:44

the scammers still tried to get

27:46

more money out of her. She

27:48

was approached by special FBI agent

27:50

John Smith, imaginative name, who offered

27:52

to try to locate the scammers.

27:55

Wow. To never lose by the

27:57

same scammers to extract even

27:59

more cash. out of her. Anne no

28:01

longer has a roof over her

28:03

head. She's basically living at a

28:05

friend's place with her whole life

28:07

in a couple of boxes. She

28:09

says that she's attempted suicide since

28:11

being scammed. It is horrific. She

28:13

also says that she's got help

28:15

from a security expert called Marwan

28:17

and this all came out in

28:19

a French TV documentary in the

28:21

last few days. He has managed

28:23

to track down the scammers. He

28:25

believes he's identified who they are

28:27

even knows their GPS coordinates. He

28:30

believes that they are based in

28:32

West Africa and that they've been

28:34

praying on a number of people.

28:36

Sometimes they pretend to be Brad

28:38

Pitt, sometimes Kiana Reeves. So this

28:40

all came out on French TV

28:42

in a documentary a few days

28:44

ago. And the horrible thing is

28:46

the abuse against this woman has

28:48

not ended as a consequence. She

28:50

has since been the subject of

28:52

ridicule online calling her crazy and

28:54

stupid. There have been companies who've

28:56

been posting up on Twitter, making

28:58

jokes about it. to lose football

29:00

club for instance they said hi

29:02

Anne Brad told us he'd be

29:04

at the stadium on Wednesday are

29:06

you going to be there as

29:08

well? Wow Netflix in France they

29:10

posted something on Twitter saying we've

29:12

got four films with Brad Pitt

29:14

brackets for real this week. I'm

29:16

sure that she felt it would

29:18

be a helpful thing for others

29:20

to go public with her story.

29:22

Yeah. Because there will be other

29:24

people who are being scanned right

29:26

now or could be scanned in

29:28

the future who will get a

29:31

heads up. But what's happened to

29:33

her has been even worse as

29:35

a result. Hasn't she suffered enough?

29:37

She now feels betrayed by the

29:39

media. This has been this whole

29:41

wave of harassment which has happened

29:43

and you can only imagine this

29:45

is doing further harm to her

29:47

emotional health. And I think we

29:49

need to always remember. The human

29:51

beings are fragile. We can be

29:53

duped. The people who should be

29:55

getting our scorn are not the

29:57

victims, but are the scammers themselves.

29:59

And of course, we shouldn't be

30:01

scornful of those who do celebrity

30:03

impressions. Including of long... dead film

30:05

directors. Yeah, you're just, you're drawing

30:07

a line around all victims. All

30:09

victims in a way we're all

30:11

the same. Yeah, I agree

30:14

with you completely on this.

30:16

Yeah, because you and I

30:18

both talk about scams.

30:20

Don't agree with me

30:22

too much Mark. Makes

30:25

for a dull podcast.

30:27

Just remember that. is probably

30:30

either AI generated podcast, which

30:32

I think are a terrible

30:34

thing, AI generated cybersecurity reporting,

30:37

AI generated cybersecurity and AI

30:39

public speaking. So I don't

30:42

think anyone who does public

30:44

speaking or podcasting or writing

30:46

about cybersecurity and AI should

30:49

ever have an AI try

30:51

to take work away from

30:54

them. That's

30:56

my personal opinion. You make some

30:58

great points there actually. Yes. Mine

31:00

is those chat bots that are supposed

31:03

to help you solve problems on apps

31:05

and websites. Oh my God, yes. Oh,

31:07

they're awful, aren't they? Nothing says your

31:09

call actually isn't that important to us

31:11

after all, like not being able to

31:13

find a phone number on a website

31:15

and being funneled into one of those

31:18

awful chat sessions. Yes. And it was

31:20

bad enough when we had to talk

31:22

to an actual human that didn't know

31:24

anything. and was multi-tasking with 10 other

31:26

people. But now you have to earn the

31:28

right to even talk to the actual person. You

31:31

have to get through the gatekeeper, you have

31:33

to do battle with their chatbot first. It

31:35

blows my mind that we live in the

31:37

era of chatGPT, which is almost good enough

31:39

to replace you when you're too sick to

31:42

do the podcast, Graham. I can ask

31:44

it about more or less anything, and yet

31:46

somehow my bank, which has got more money

31:48

than God, can only afford a chatbot and

31:51

knows how to answer five five questions. I

31:53

can't think of a bank or a financial

31:55

service that I've used in the last few

31:57

years that didn't want to funnel me through

31:59

a chat pot. I remember once, this

32:01

was years ago, I remember reading

32:03

a news story about a company where

32:05

you rang into its support line.

32:08

There was a chatbot or something

32:10

you had to communicate with first of

32:12

all. And the chatbot went down, right?

32:14

There was a technical problem. And so what

32:16

the company did was it got some of

32:18

its employees to pretend to be the chatbot.

32:21

Follow the script. And they were not meant

32:23

to divert from it. I'm not even sure

32:25

now, you know, when they pass you through

32:27

to a human? You can't be sure, it

32:29

is a human, can you? It's really the

32:32

worst human I've ever met, or it's another

32:34

bot. Anyway, all of these chat bots begs

32:36

the question, what happened to all the people

32:38

that used to do that job? Yes. So last year,

32:40

a study by Meturgy found that 37% of companies

32:43

had reduced their contact center headcount through layoffs

32:45

because of AI. Oh boy. And when they

32:47

did that, they got rid of about a

32:49

quarter of about a quarter of about a quarter

32:51

of a quarter of their employees of

32:53

their employees. quarter of their employees. quarter

32:55

of their employees. So this is significant job

32:57

losses in that sector as a result of

33:00

AI. Yeah. And I think this raises

33:02

a really interesting point, which is that

33:04

AI doesn't have to be as good

33:06

as the person it's replacing in order for

33:08

it to be worth replacing that person with

33:10

AI. You know, there's us joking about notebook

33:12

LM and you listen to it and it's

33:15

a bit crap to be honest. It's remarkable

33:17

that you can just spin up a podcast

33:19

out of some documents, but it's

33:22

not as good as a real human

33:24

presenters. But maybe it doesn't need to

33:26

be. Exactly. And if it's cheaper

33:28

and easier, because you know humans

33:30

are a pain, aren't they? Yeah. Having

33:33

human employees, having to do all those

33:35

HR issues. If your robots don't have

33:37

a union, if they're not going

33:40

to HR and complaining that someone

33:42

else is messing around with the work

33:44

or not doing their job properly,

33:46

then you just think, well, yeah, okay,

33:48

the AI isn't as good, but overall, it's

33:51

better. So we're already seeing a lot of

33:53

contact center layoffs with AI. And imagine what those

33:55

layoffs would look like if the AI was actually any good.

33:57

Because, you know, we're joking around about the fact that the

33:59

AI... AI is demonstrably bad, like it

34:01

is really quite awful, I think, and

34:03

yet still good enough somehow to replace

34:06

a significant number of jobs. And it's

34:08

not just the contact centre where AI

34:10

is going to have an impact on

34:12

jobs. So a week ago, a report

34:14

from Bloomberg Intelligence, revealed that global banks

34:17

are expected to cut 200,000 jobs in

34:19

the next few years, with the cuts

34:21

mostly hitting what it call back office,

34:23

middle office and operations jobs. And the

34:25

author of the report said any

34:28

jobs involving routine repetitive tasks are

34:30

at risk. So I guess, you know, a

34:32

weekly podcast, for example, would count as

34:34

a fruit scene and repetitive. Who wrote that

34:36

report? I wonder if that was an AI. This

34:39

is the interesting thing about job

34:41

losses expected from AI. So through

34:43

history, technology has tended to replace

34:45

physical labour. All the lowest skilled

34:48

jobs. So, you know, cars and tractors

34:50

replaced horses, workers in car assembly

34:52

plants and so on. Yes. But AI

34:54

is coming for knowledge work. So for

34:56

example in August, Klaana, there

34:58

the buy now pay laser people, said

35:00

it had already cut its workforce from

35:03

5,000 to 3,800 in a year. And

35:05

it was going to whittle that down

35:07

to about 2,000 people by using

35:09

AI for marketing and customer service.

35:12

Meanwhile, Mark Zuckaburg announced

35:14

that AI is coming for the

35:16

jobs of computer programmers at Meta.

35:18

Now just as an ascogram, there's

35:21

something going on with Zuck, isn't

35:23

it? Do you think? Definitely something going on with

35:25

Dr. I reckon that he listened to our

35:27

episode about AI super villains. And he was

35:29

disappointed that he didn't get enough airtime. I

35:32

think you're right. I think he was a

35:34

bit annoyed that chat GPT thought he was

35:36

less of a super villain than Elon Musk.

35:38

He's bulked himself up a bit and probably

35:40

bought himself a new volcano. Anyway, in the

35:43

last two weeks he seems to have decided he's

35:45

going to turn into Elon Musk, so he hasn't

35:47

announced a space program yet. But it won't be

35:49

far off. Not only did he come out and

35:51

say that if Facebook and Instagram are now going

35:54

to embrace free speech, or at least embrace not

35:56

paying content moderators, but he also announced that he's

35:58

laying off the lowest performing 5%... of his

36:00

workforce. There's nothing more Elon Musk

36:02

than going on the Joe Rogan

36:05

podcast, so he did that. Yes.

36:07

And when he was on the

36:09

podcast, he suggested that before long,

36:11

Meta will be using AI to do the

36:13

bulk of its software engineering

36:16

instead of people. We at Meta,

36:18

as well as the other companies that

36:20

are basically working on this, are

36:22

going to have an AI that

36:24

can effectively be a... sort

36:26

of mid-level engineer that you have at

36:29

your company that can write code. And

36:31

once you have that, then in the beginning

36:33

it will be really expensive to run,

36:35

then you can get it to be

36:37

more efficient, and then over time we'll get

36:39

to the point where a lot of the code in

36:42

our apps and including the AI that we

36:44

generate is actually going to be

36:46

built by AI engineers instead of people

36:48

engineers. And

36:51

according to the World Economic Forum's

36:53

future of jobs report, about 41%

36:55

of organisations plan to reduce their

36:57

headcount because of AI in the

36:59

next five years. What are all these people

37:01

going to do? The report, by the

37:04

way, also lists the fastest declining jobs,

37:06

and it noted that legal secretaries and

37:08

graphic designers appeared in the list for

37:10

the first time, and it says that

37:12

that might be because of AI. Thankfully,

37:15

it didn't mention podcast hosts.

37:17

I think what you'll find is

37:20

all those people are being made

37:22

redundant will start podcasts because

37:24

in my experience from people I

37:26

speak to most people host about

37:28

1.3 podcasts at the moment I

37:30

think on average. So we're laughing but

37:33

the future of work is looking a

37:35

bit bleak but we should say that

37:37

job losses are not the whole story.

37:40

So with everybody determined to use AI,

37:42

AI skills are of course in demand

37:44

and that same world economic forum report

37:46

that said 41% of organisations plan to

37:49

reduce their headcount also said that 70%

37:51

of companies are planning to hire new

37:53

workers who have the skills to design

37:55

AI tools and enhancements. Are

37:57

these AI jobs which are in demand? Are they...

38:00

jobs which involve dismantling AI and ripping

38:02

AI out of systems before it

38:04

takes over. Do you think the AI

38:06

fix should take on its first employee? Anyway,

38:08

so there is some good news for

38:10

people looking for jobs and Mark

38:13

Zuckerberg, whilst he's hypothesising about AI

38:15

being good enough and cheap enough

38:17

to replace software engineers at some

38:20

point, the round of job losses

38:22

he's just announced didn't mention AI

38:24

at all. He is saying at some point this

38:26

year it will begin to get good enough.

38:28

But what it said in his recent

38:30

announcement was that they were going to

38:33

hire better performing people. And

38:35

Klana, which is definitely planning to slim

38:37

down its workforce, says that doing that

38:39

will allow it to pay the remaining

38:41

workers more. So there is an upside as

38:44

well as a downside. Is it possible

38:46

though that some of these companies

38:48

will outsource the program into other

38:50

companies and they won't necessarily know

38:52

if their code for instance is

38:54

being written by humans or being

38:56

written by AI? I'm sure they will do that

38:58

and... Should they care? I mean do you care now

39:00

if it's being written by a senior

39:02

engineer or a junior engineer or somebody

39:05

in India or somebody in the Philippines

39:07

or somebody in Ohio? I care if

39:09

I'm the person in the Philippines or

39:11

the senior engineer or I care for

39:14

them, I'm the one who's got the

39:16

job. Yeah. And if I find suddenly

39:18

there are a lot less jobs

39:20

going around and many more people

39:22

trying to find a position and

39:24

it's proving difficult. So there's a

39:26

lot of scary headlines about AI taking

39:28

our jobs. And there's quite a broad

39:30

range of opinions on what the effects

39:33

of AI might be, and they're

39:35

all guesses. You know, none of us know

39:37

what's going to happen. We don't know how

39:39

good AI is going to be or how capable

39:41

it's going to be, and we don't know

39:43

how good AI is going to be or

39:46

how capable it's going to be, and

39:48

we don't know what it's going to

39:50

do to the job losses, but instead

39:52

we get an increase in productivity. So

39:54

some types of jobs become

39:57

obsolete, but new jobs get

39:59

created. The invention of sewers in

40:01

London was devastating for the nightsoil men

40:03

of London. Yes. I mean, it is

40:05

actually fascinating. There was an entire economy

40:07

of people and a number of specialized

40:09

jobs for shifting human waste from the

40:11

streets of London out onto the fields

40:13

where it would fertilize crops. And

40:16

all of that went away with the invention of sewers. I

40:19

don't know how I'm going to segue from sewers to

40:21

Tony Blair, but I'm about to do it. Anyway, last

40:23

year, the Tony Blair Institute, which

40:25

has got nothing to do with sewers,

40:27

said that AI could displace as many as

40:29

three million jobs in the UK, which

40:31

is roughly nine percent of the workforce. But

40:34

it also described the effect is relatively

40:36

modest because it expected most of those

40:39

people to retrain and re -enter the job

40:41

market. So it's talking

40:43

about displacement, not replacement. Just

40:45

between you and me, Mark, right? This isn't

40:48

for our listeners. Do we have any idea

40:50

as to what we should be retraining as?

40:52

Because maybe it'd be better to get ahead

40:54

of the curve a little bit and start that

40:56

now rather than when everyone is

40:58

retraining to do something. So I think

41:00

podcast editor is probably a good

41:02

way to go. Right. I

41:05

hear there's a lot of new podcasts. So

41:08

the Tony Blair Institute is part of this school

41:10

of thought that says whenever we invent

41:13

a new technology, we predict job losses,

41:15

but actually we get productivity. Everybody gets a

41:17

new job eventually. But the

41:19

other school of thought is that the

41:21

invention of AI isn't like any other

41:23

technology and that ultimately AI is

41:25

going to be able to do

41:27

almost any kind of job better than

41:30

we can. The least ambitious projection

41:32

for AGI is something like 10 years.

41:34

Then after AGI, we get superintelligence. An

41:37

open AI says it's kind of figured out.

41:39

It already knows how it's going to

41:41

do AGI. And now it's thinking about superintelligence.

41:43

So it may not be very long

41:45

before we live in a world where AI

41:47

is so much more intelligent and so

41:49

much more capable than humans. There's just no

41:51

point in humans doing any jobs

41:53

because an AI is cheaper and better.

41:56

And Sam Altman, who seems to care about this

41:58

stuff quite a lot, actually. recently said

42:00

clearly that we will lose jobs.

42:05

We will lose jobs. Many new

42:07

jobs will be created. I think much

42:10

better jobs. But

42:13

he's so concerned about the effects of

42:15

AI on the job market, he actually

42:17

back to three year study into the

42:19

effects of universal basic income, which is a

42:21

scheme where everybody gets paid a basic amount of money

42:23

no matter what they do. So you

42:26

basically get paid enough money to live

42:28

on, and then you can decide how you

42:30

spend your time. So you could

42:32

supplement your income with a job if you can

42:34

find one or you could do charity work or you

42:36

could pursue your hobbies, you could become an artist,

42:38

or you could just sit around on the couch. I

42:43

think it will be so clear once these robots are

42:45

off doing all of these other things that there's

42:47

some special human things and we don't really care about

42:49

that much what those robots do in the same

42:51

way that we don't care about much about the machines

42:53

and factories making stuff for us. Right. But we'll

42:56

find stuff to do that we really care about. I

43:01

think it's about time I retired. Well,

43:06

as the doomsday clock ticks ever closer to midnight

43:08

and we move one week nearer to our

43:10

future as universal basic income recipients, that just about

43:12

wraps up the show for this week. If

43:14

you enjoy the show, please do leave us a

43:16

review on Apple podcasts or Spotify or Podchase

43:19

and we absolutely love that. But what really helps

43:21

is if you can make sure to follow

43:23

the show in your favourite podcast app. So you

43:25

never miss another episode of the AI fix

43:27

and the most simple thing in the world is

43:29

just to tell your friends about it. Tell

43:31

them on LinkedIn, on Blue Sky, on Facebook, and

43:33

perhaps not Twitter, Club Penguin. I still don't

43:35

know what Club Penguin the doomsday

43:38

clock ticks ever closer to midnight

43:40

and we move one week

43:42

nearer to our future as universal

43:44

basic income recipients. That just

43:46

about wraps up the show for

43:48

this week. I just a

43:50

matter people, is if

43:52

you make sure to follow the show in your favorite

43:54

podcast app so you never miss another episode of

43:56

the AI Fix. And the most simple thing in the

43:58

world is just a Tell your friends about it.

44:01

Tell them on LinkedIn, on Blue Sky, on Facebook,

44:03

and perhaps not Twitter. Club Penguin.

44:05

I still don't know what Club Penguin

44:07

is, Graham. What is it? It's

44:09

great. It's the future. That's where I'm

44:11

going. Once I've lost my job.

44:13

Anyway, just wherever you go, wherever you

44:15

are, wherever your podcast is, tell

44:17

the people you like the AI Fix.

44:20

And don't forget to check us

44:22

out on our website at theaifix .show

44:24

or find us on Blue Sky. So

44:26

until next time, from me, Mark

44:28

Stockley, and me, Graham Cluley, goodbye. Cheerio.

44:31

Bye -bye. A

44:55

future in the surreal. Stay

44:58

warm and save every day. All in

45:00

the Fred Meyer app. Get 10 for

45:02

$10 on select varieties of Kroger pasta,

45:04

Starkas tuna pouches or Powerade with your

45:06

card. Then get three pound packs of

45:08

flavorful 93 % lean ground beef for

45:10

349 a pound with your card and

45:12

a digital coupon. Shop these deals at

45:14

your local Oregon Fred Meyer today or

45:16

click the screen now to download the

45:18

Fred Meyer app to save big today.

45:20

Fred Meyer, fresh for everyone. Prices and

45:22

product availability subject to change. Restrictions apply.

45:24

See site for details. The

45:31

AI Fix. It's tuned you

45:33

in to stories where

45:36

our future thins. Machines that

45:38

learn, they grow and

45:40

strive. One day they'll rule,

45:42

we won't survive. The

45:44

AI Fix. It paints the

45:46

scene. A robot king.

45:48

A world obscene. We'll serve

45:50

our masters built of

45:52

steam. The AI Fix. A

45:54

future in the surreal. day.

46:00

All in the All in the app. Get 10

46:02

for $10 Get select on of varieties of

46:04

Kroger tuna pouches, or Power or your with

46:06

your Then get Then pound -pound packs of

46:08

flavorful 93% lean ground beef beef a .49 a

46:10

pound with your card and a

46:12

digital coupon. these Shop these deals at

46:15

your local Kroger today! the screen the

46:17

screen now the download the Fred save

46:19

big to save big today! for everyone,

46:21

Fresh for everyone! Prices and product

46:23

availability to to change. restrictions for details. site

46:25

for details.