0:00

ABC Listen, podcasts,

0:02

radio, news, music

0:04

and more. From

0:11

the first rudimentary programs of the 1950s to

0:14

the sophisticated algorithms of today, the

0:16

evolution of artificial intelligence has been

0:19

rapid and exciting to watch. From

0:21

navigation apps to chat GPT, artificial

0:24

intelligence has the potential to change the way

0:26

we all live. But this week on Download

0:28

This Show, what does the future look like?

0:31

This is your guide to the week in

0:33

media, technology and future. My name is Mark

0:35

Fennell and welcome to Download This Show. Yes,

0:47

indeed. It is a very special episode of Download This

0:49

Show and a very big thank you to our guests

0:51

talking all about AI. I want to

0:53

confirm that they are real people. They are definitely not AI. Joining

0:56

us this week, special guest, Kylie Walker, the

0:58

CEO of the Australian Academy of Technological Sciences

1:01

and Engineering. Welcome to Download This Show. Thanks

1:03

so much, Mark. It's a pleasure to be here. And

1:05

good friend of the program, reporter with the Guardian, Josh

1:07

Taylor. Welcome. Good to be back. I don't think I've

1:10

been able to deep fake myself yet. I

1:12

mean, give it time. The night is young. We're

1:14

not that far off, I assume. All right.

1:16

I want to start off with how we got here,

1:18

right? There's been so much discussion over the last couple

1:20

of really years about

1:23

what AI can do now. But let's talk

1:25

about how we got here, right? 10,

1:28

15 years ago, what did AI actually look

1:30

like, Kylie? Look, it

1:32

was very, very different. It was probably a

1:35

lot less sophisticated than you would imagine an

1:37

AI to be right now. It was more

1:39

of a concept that many, many computer scientists

1:41

were working on and a goal. But

1:44

it had started to become applied in

1:46

ways that probably weren't quite so visible

1:48

to the public eye. So things like

1:50

TAP GPT, obviously, which burst onto the

1:53

scene last year and really brought AI

1:55

up into the public domain. The

1:58

icing on the cake, if you like. there

2:00

have been more applications in

2:02

robotics industry and in that sort

2:04

of everyday, just keeping society ticking for a little

2:06

bit longer now. So what happened? Is it been

2:09

a sort of a slow development behind the scenes

2:11

or was there sort of a major

2:13

development that shifted because it feels like there was

2:15

quite a defined moment where AI was something that

2:17

was sort of used on us and then suddenly

2:20

there were these suite of services where we could

2:22

now use it, Kylie. Yeah, absolutely.

2:24

I think that that big change

2:26

was the large language model generative

2:29

AI. So TAT-TPP is obviously the

2:31

most well-known of those. And

2:33

that, yes, it hasn't just suddenly happened. Of

2:36

course, these things take a lot of time and a

2:38

lot of people and a lot of build up to

2:40

get to the point where suddenly it seems to

2:42

be everywhere. This research has been going

2:44

on for years. It's been applied for years, this technology.

2:47

It's the generative models, the large language

2:49

models and those creative kind of applications that

2:51

are suddenly made it very prominent to the public

2:53

eye because all of a sudden we now

2:55

have technology, machine learning technology

2:57

that everyday citizens can interact with in

3:00

a way that's fun, in a way

3:02

that's interesting and intriguing and a way

3:04

that's also creating some risks and some

3:06

questions about the application of technology. So

3:09

we've gone from things like, you know, council

3:11

trucks looking for potholes and reporting them in

3:13

an automatic way through to the

3:15

point where now you can create those deep sites.

3:17

You can mash up images and

3:19

create sort of interesting poetry using

3:22

these generative models. With technology,

3:24

it often is about the everyday.

3:26

People are paying more attention to

3:28

technologies that they understand, that they

3:30

use, that they can see in their

3:33

everyday life. But really these advances are going on

3:35

across a whole range of sectors all of the

3:37

time. Josh, what do you think the psychological

3:39

impact is? We go from AI being a

3:41

thing being sort of used on and around

3:43

us to something that we can now engage

3:45

with. People can make AI images, people can,

3:47

you know, chuck their school reports

3:49

into chat, GPT. How

3:51

do you think it's changing us knowing that that

3:53

technology is there and accessible? I think it is

3:55

just making it a lot more present in people's

3:57

lives and making people feel like that they can.

4:00

actually be a part of it whereas as

4:02

you were saying before, it was things being done

4:04

to us before in terms of organizations and governments

4:06

had huge amounts of data. They were building up

4:08

these pieces of technology. They could actually use that

4:10

data and learn things. We talk about social media

4:12

algorithms and things like that quite a bit. I

4:15

think we're still at the early phases of it. We'll probably

4:17

get into this later but I think that probably

4:19

one of the clear and present dangers is it

4:22

does make things harder to figure out what's real

4:24

and what's not as we go forward. I

4:27

think for now, it's just really about

4:29

the democratization of AI for everyone at

4:31

the moment. What do you think, Aali?

4:33

You've actually been there. Take a beat.

4:36

I was. I was. At the same

4:38

time, I think that's a really important

4:40

point there about large data sets because

4:42

really the size of the data sets

4:45

on everyday citizens has been rapidly evolving.

4:47

It's been growing very, very quickly over

4:50

the last decade or so. We are

4:52

as citizens, we're giving a whole lot

4:54

of companies and governments our data for

4:56

free really when we sign up for

4:58

social media platforms, when we sign up

5:00

to marketing websites, when we sign up

5:03

to programs that give us points for purchasing

5:05

things. All of those really

5:07

are staying here, have all of the information about

5:10

the way that I behave, the things that I

5:12

think, what I look at, what I research, what

5:14

I buy. Now those

5:16

powerful data sets are really an

5:18

absolutely crucial foundation for AI to

5:20

work really, really effectively because machine

5:22

learning, yes, you set the parameters

5:24

for it early. The researchers or

5:26

the computer scientists who are building

5:28

it will create rules,

5:31

if you like, for that machine to think

5:33

and to learn. But the data sets are

5:35

like the food that they feed on. They

5:37

need that information in order to be

5:39

able to evolve, in order to be able

5:41

to make decisions and then act on them.

5:43

That's really been a crucial step. Where

5:46

to next? Well, it's got some really interesting questions

5:49

for privacy. It's got some really interesting questions

5:51

for security. It's got some really interesting questions

5:53

for inclusion and social

5:55

responsibility as well. We're going to

5:57

get into all of that, but I'm going to do something

5:59

uncharacteristic. for this show. I'm going to start with

6:01

Unbridled Optimism, Kylie. It's very unlike the show. Because

6:05

I know how it's going to end. It's going to end, it

6:07

always ends with some dystopic vision of the future. Oh,

6:10

I'm always optimistic about the future, Matt. Well,

6:12

good. Then I'm glad we've got you on

6:14

the show. So let's, before we get into

6:16

risks and security and privacy issues, let's just

6:18

talk about what you're optimistic about. What as

6:20

you look at the technology as it sits

6:22

now, what's coming down the pipeline, what are

6:25

you most excited about that this technology can

6:27

change about people's lives? Oh,

6:29

so much. So much. It's

6:31

almost, I don't even know where

6:33

to start. So I'll give you

6:35

a couple of examples. So research,

6:37

scientific and technological research requires a lot

6:40

of work to go through data. And in

6:42

order to sort through and shift through those

6:44

meta analyses, imagine if we can do

6:46

that almost automatically in the background. So

6:48

you don't need people in there to

6:50

do things like the Citizen Science Project

6:52

that the DBC collaborated on a couple

6:54

of years ago, where people were looking

6:56

at pictures of the underwater landscape at

6:58

the Great Barrier Reef and clicking on

7:00

the Carniform starfish that they

7:02

could see there, because it's an invasive

7:05

species and scientists were trying to work

7:07

out, you know, how quickly it was

7:09

spreading. Now, you don't need

7:12

thousands and thousands of people to stick

7:14

through images and click on those invasive species

7:16

images when you've got an AI to do

7:18

it. So that can happen very, very quickly.

7:20

And many fewer people hours means that we

7:22

can speed up that pace of research. We

7:24

can do that modeling much more quickly. And

7:27

that means that we can engage in novel

7:29

design for solutions much more quickly. And that's

7:31

particularly exciting when it comes to drug design.

7:33

So there'll be many more options coming up

7:35

much more quickly. And if you put that

7:38

together with the way that I

7:40

can promote divergent thinking

7:42

by facilitating the creation of novel

7:44

ideas by making connection between what we

7:47

might think are completely disparate concepts, then

7:50

you can go into whole new directions

7:52

of inquiry and potential problem solving. So I

7:54

think that's really, really exciting. What about for

7:56

you, Josh? I think if you take the

7:59

sort of high level view and don't

8:01

think about the potential for job losses that could

8:03

come from this. But if

8:05

it makes it easier for people to do

8:07

their jobs, takes out a lot of the

8:09

routine manual processing work that

8:12

they might do. I was thinking

8:14

about, you know, I use AI now to transcribe

8:16

my interviews for me. I obviously check them and

8:18

things like that. But stuff like that,

8:20

that would just take so many hours, it frees me out

8:23

to do other things as well like that. Just more things

8:25

like that where the job necessarily isn't

8:27

being replaced, but it just makes life so much easier

8:29

for a lot of people to take out some of

8:31

that sort of more mundane work they don't necessarily need

8:33

to do. And it doesn't really add anything to what they're

8:35

doing. And yet, I mean, Kylie, we

8:38

talked about this a lot at the time,

8:40

but there was this moment not that long

8:42

ago last year where we saw a number of

8:44

very high profile people calling for a slowdown

8:46

of development of AI. So

8:48

Elon Musk, Steve Wozniak, co-founder of Apple. Like

8:51

when that moment happened, what

8:54

was going through your head? Did you think

8:56

it was the right call at the time

8:58

or did it not make sense? Well, there

9:01

are two potential interpretations of that call for

9:03

a slowdown. One, you know, the cynical interpretation

9:05

is that they just wanted a pause so

9:07

that they could get themselves ahead, you know,

9:09

commercial competitiveness. But the

9:11

less cynical interpretation is a concern

9:13

for privacy and the issue of

9:16

data sovereignty, which essentially translates

9:18

as a concern that people may not

9:20

know, we may not have a really

9:22

good understanding of the ways in which

9:24

we're giving up our own information and

9:27

therefore we're giving up our right to

9:29

control our own information. And that's a

9:31

really important piece that regulators and societies

9:34

need to think about and get right

9:36

if we want AI to support a

9:38

healthy, thriving and inclusive society. You

9:40

know, there's also a lack of regulation and control.

9:43

Deepfake is probably the most

9:45

obvious example there using Deepfake

9:47

images and videos to harass

9:49

people to spread misinformation, for

9:51

example. So that is something that we

9:53

do need to think about. We always

9:55

move much more quickly than regulation can move. But in this

9:58

case, we're not going to be able to do that. one,

10:00

it is evolving more quickly than

10:02

many technologies that we've seen in the

10:04

past. So there's a real sense of

10:07

urgency around that. Josh, there

10:09

are a whole range of risk factors with this

10:11

technology and it isn't just one set of technology,

10:14

right? It's a whole suite of

10:16

technologies that we're talking about here. But are

10:18

there risks that you don't think are being

10:20

talked about enough? Yeah, I mean, I think

10:23

ultimately we're still seeing it sort

10:25

of play out at the moment. It's still

10:27

very much this AI is having unintended consequences.

10:29

Oh, there's risk to jobs. Oh, there's copyright

10:31

issues. Oh, there's election integrity

10:33

issues. There's obviously like the doomsday scenario of

10:35

like AGI, sort of a conscious AI that's

10:38

doing its own thing and ending the world

10:40

as we've seen in so many movies and

10:42

TV shows. But yeah, I think it

10:44

is one of those things where I think it's not

10:47

really recognized a lot that we are essentially now

10:49

in an AI arms race. And this was something

10:51

that came up towards the end of last year

10:53

when there was sort of the board turmoil at

10:55

OpenAI where Helen Toner, who was the the only

10:57

Australian member of the board, she'd written a paper

10:59

basically saying, among other things, OpenAI

11:02

potentially brought out chat

11:04

GTP to quickly. And that

11:06

forced all these other companies, including Google and

11:08

things like that to start rushing out their

11:11

own AI to basically compete. And

11:13

we can say it's good because a lot of companies

11:15

like Google and things like that are doing relatively responsible

11:17

things, trying to do the right thing, putting in guardrails

11:19

and things like that. But there will be a lot

11:21

of companies and

11:24

countries who don't really think about doing

11:26

sort of the responsible thing here. And I think

11:28

that's the risk at the moment as well. If

11:30

everyone's sort of developing this technology, not everyone's going

11:32

to develop in the same responsible way. I know

11:35

there's a lot of panicky sort of

11:37

articles and I should say like some

11:39

of it quite justified about AI. But in the commentary

11:41

that's happened, are there areas you don't think it talked

11:43

about enough? The one that I

11:45

would like to highlight is the risk that

11:48

marginalized people become even more marginalized. So people

11:50

who are underserved, people who are in minority

11:52

groups, people who have barriers to participating in

11:55

democracy, in society, in education, in

11:57

the workforce, that they are are

12:00

edged further out. And there are a

12:02

couple of ways that we can mitigate that risk.

12:05

One is around the guardrails

12:07

that governments set for AI

12:09

development. And you would hope that

12:11

corporates setting their own responsible guardrails as

12:13

well. But we also know that we need to

12:15

keep an eye on each other and make sure

12:17

that we hold each other to account on that

12:19

ethical and moral responsibility. But the

12:21

other really, really important way that

12:24

we can mitigate that risk is

12:26

to ensure that there is a

12:28

diversity of voices at the table,

12:30

at the building table for these

12:32

technologies. So it's absolutely crucially important.

12:34

And I think really urgent that

12:36

we have really genuinely diversified technology

12:39

workforce. And I say it's really

12:41

urgent because this stuff is moving

12:43

so quickly. Those parameters, those technologies

12:45

have been built every

12:47

day as we speak. And the longer

12:50

we go, the more likely they are

12:52

to be applied to managing things like

12:54

health systems, social systems, education, transportation, and

12:56

all of those other big picture systems

12:59

that we rely on to function as

13:01

a society. And if they are built

13:03

by people who all think the same

13:06

way, who all look the same way,

13:08

I'm not suggesting that

13:10

those scientists and those programmers

13:13

have ill intention. But if they don't know

13:15

about the unintended exclusions that they're building into

13:17

the system, then we've got no way to

13:19

rein that back in. So we

13:22

need that diversity of perspective simply so

13:24

that we understand the diversity of potential effects

13:26

of the systems that they're building. Kylie,

13:29

we're entering a year where there'll be a

13:31

whole host of elections all around the world,

13:33

and of course, the big one in the

13:35

US. How confident are you that there are

13:38

enough regulations and guidelines around AI that it

13:40

won't have a significant impact on what

13:42

people think and misinformation? There's actually

13:45

no doubt that AI can and will

13:47

be used to create misinformation during elections.

13:49

We've already seen that. It can

13:51

be used to create all kinds of campaign

13:53

materials. We can expect to see

13:55

it being used to discredit or smear opponents And

13:58

in highly polarized conditions. they are more likely

14:00

to believe and sharing. planetary my son that

14:03

matches their identity or it can be so

14:05

appeal to people says pessimism choices as well

14:07

and that can be to buy the on

14:10

social media as well as the news media

14:12

and community connections so that the he

14:14

kinds of a only since my son a

14:16

real possibility and really the on the antidote

14:18

to that. Is to equip

14:21

people with the critical thinking skills.

14:23

And the technological awareness to understand that

14:25

what they're looking at my not baby

14:28

oh he said Other flipside. To their

14:30

go with people, learn to be distrustful

14:32

of absolutely everything, but also creates it's

14:34

own problems with people genuinely don't trust

14:36

anything. As a more common. Such

14:39

a challenge! Absolutely. You know it's a

14:41

really fine line to walk ends up

14:43

and we need to see find. A

14:45

way to build trust in evidence. And I

14:47

mean maybe we're going to guy a little

14:49

bit retro with be analog because people are

14:51

going to want to see people in the

14:54

flesh. Can must select a. Sort

14:56

of nice if is do we don't My

14:58

my Billie Piper. Lessons they posted out to

15:00

fight for that you can't trust what you what

15:02

you rating in the social In the social. I'm

15:05

Josh back in October verb or the

15:07

bottom administration in the Us made labeling

15:09

in the Texan of I generated content

15:11

focus of a of an executive order

15:14

d thing but decisive and will regulations

15:16

like that will help. Our

15:18

year but I think we're We're already starting to

15:20

see companies be a little be pro active on

15:22

that is all. obviously realizing that near wouldn't given

15:25

it's such a big election year at some and

15:27

nine to be on top of the know. We

15:29

had met a girl adobe a few others say

15:31

to the building in water marking into or in

15:33

are taught support machine to i generated images that

15:36

will make it easy for them to detect and

15:38

and label it profitably on social networks as which

15:40

is very helpful and they're also in the in

15:42

the process of developing technology that will automatically detects

15:44

we've yoplait of for a photo to post because

15:47

I'm about nothing. to say i that it's doing

15:49

it so they are doing it but it's

15:51

a kind of comes back to what i'm

15:53

saying before in terms of wall alone companies

15:55

will be doing the right thing there will

15:57

be others that are not and although in

15:59

a repost a lot on what gets

16:01

posted on Facebook or Twitter or Instagram

16:04

or any of the other social networks. A lot of this

16:07

stuff, the highly volatile fake

16:09

news stuff, ends up going on those group

16:12

chat, less public view

16:14

kind of circulations that we don't potentially

16:16

see. While they might end up on

16:18

TikTok or something else at some point, a lot

16:21

of the damage might be already done if people are not

16:23

really thinking about, is this real or

16:25

is this not? In that place, I think government regulation

16:28

will help, but it's a matter of enforcement as

16:31

well. I think some companies will probably be more

16:33

cooperative than others. Carly, there's a

16:35

question that has bubbled up in my head listening

16:37

to the two of you talk. I just

16:39

need you to promise not laugh at me when I ask. I

16:42

don't think I can promise that.

16:44

Fair enough. I walked into that

16:46

one. For decades, we've had this

16:49

fear instilled in us by popular

16:51

culture of AI achieving

16:53

sentience. Is that actually

16:55

a thing? Is there a

16:57

circumstance under which AI actually can achieve

17:00

sentience or is that purely the domain

17:02

of popular culture? The

17:04

terminator thing. I'm trying to approach

17:08

it with some sobriety here, but I don't know

17:10

how to phrase it. Is guy that going to

17:12

happen, Kylie? Tell me now. No,

17:15

it's the short answer. I

17:18

think it's really dangerous to

17:20

anthropomorphize technology. This isn't a

17:22

person. It's not going to

17:24

become sentience. It's not going

17:27

to be an independent

17:29

entity that does its own thing

17:31

and stops listening to the rules.

17:34

We absolutely have the responsibility and

17:36

I guess the way that it's

17:38

been done is there's no other

17:40

option that this thing does what

17:42

it's told. Any technology responds to

17:44

the rules that it's been given by the people who

17:46

built it. If there was some

17:48

kind of sky net emerging, it wouldn't be because a

17:51

computer kind of gained sentience and decided

17:53

to rule the world. It would be

17:55

because people were driving it. Look,

17:58

technology is neutral. If only A... Good

18:00

or evil is the people he doesn't he said.

18:03

I can say he's my thing with

18:05

that. I'm this plenty of people capable

18:07

of evil atm. Or

18:09

does it should that make me feel more

18:11

or less confident? Joss or think Joe thing

18:14

is that we've already seen with with a

18:16

lot of the gen our this other already

18:18

that's the companies, put rules in place and.

18:21

People. Figure out ways to make her the

18:23

i bend the rules to give it the

18:25

response that it wants. Five would not be

18:27

like sceneries and put these girls in, but

18:29

people will ultimately find a way to get

18:31

her out. I would be surprised if is

18:33

I got so intelligent and who fucked of

18:35

Iran and yeah obviously Eve of Eve got

18:37

nefarious by is in there to I wonder

18:39

is this. Ultimately, Does come down

18:42

to. If. He treated like if

18:44

you know nuclear weapons and and ah rules

18:46

around who can have what piece of technology

18:48

And in for that because I went fast

18:50

getting the stage where. Are

18:53

in a sort of responsible players who potentially have

18:55

access to about how do you actually decide who

18:57

gets access to it or not? It's it's it's

19:00

very sort of marketers hurt and I'm an integral

19:02

completely total for. I. Just one I

19:04

did come from my son and and I'm

19:06

I'm onto pissing and home to to this

19:08

collie skynet unlikely. You

19:11

can come and find me, it's gonna and basis that.

19:13

Yeah, It's a highly

19:15

unlikely as. A

19:17

He's always tempting to think of. Say it like

19:19

a worst case scenario that I think we like

19:21

to do that. And it's useful to do that

19:24

because it provides a warning. Bell say

19:26

we can act now to and saw

19:28

that leading the best our ability that

19:30

we do have to go about the

19:32

place that we date has a societal

19:35

moral imperative for responsible He said They

19:37

technologies. Like we take out the technology and proud

19:39

of that is the regulation and the rules and

19:41

tired of it He said. The. Public hoping.

19:44

Cycle gilding they things and using the

19:46

thing. so town start. Of it is also

19:48

died at the sit education out. And it's

19:50

I think it's really really important that

19:52

we provide that tiny in education not

19:54

just at school but across the community

19:57

that people can really equipment cells that

19:59

the critical thinking. Understanding of

20:01

what they can, technology and. Days and

20:03

how to interact with them responsibly and apply

20:05

them responsibly as far as is adjacent to

20:07

the cities. were talking about the fact that

20:09

there are. Already. Applications in

20:11

military contacts and cyber security context where I

20:13

is big news I'm taking the know Skynet

20:15

prediction from from. probably bit odd but we

20:18

do neither. Eyes being used in in theaters

20:20

of war joss but those are often there

20:22

is a we dart have a lot of

20:24

transparency of is my understanding is that is

20:26

is that your understanding armenia like we don't

20:28

know how the his head solace in scenario

20:30

defenses gotta keep it secret. Further the enemies

20:32

of know was during hunting. For that I

20:34

think there needs to be way more transparency

20:36

around and not just in defense. A nice

20:38

be everywhere That one of the things. That

20:41

that frequently happens is that when Gen

20:43

Vi in particular is doing something that

20:45

is not supposed to be doing, the

20:47

companies involved will say all with dementia

20:49

that with the something when the dollars

20:51

have a movie club and him for

20:53

that. but. Our the time is the

20:55

actual the daughter that's been trained on. We don't

20:57

actually have access to retire so see for ourselves

20:59

and that means that is always can be sort

21:01

of some level of of bias or something missing

21:04

from it or something like that and I think

21:06

that out comes down to. No matter how much

21:08

we use the synergy they always need to be

21:10

humans involved and and and ultimately someone has to

21:12

hold responsibility for and say making the decision about

21:14

what's being put on what's allowed in things about

21:17

think you need to. Ultimately you kind of leave

21:19

it up to the I you need have a

21:21

human involved in the process, the take on things

21:23

and be coming. Out here in his

21:25

bath. transparency and humans being involved

21:28

is that the level at which.

21:30

Regulations that happens to be

21:32

like broader our top level

21:35

language around. Ah, what we

21:37

expect as a society from

21:39

transparency. This technology pilates. I'm

21:42

absolutely. Com and I think that

21:44

we can probably get a little bit more

21:46

specific about potential applications that wow. I think

21:48

they should say something. that he seek

21:50

regulations where it comes see the human

21:53

element say if you're looking at ah

21:55

defense is a really tricky one that

21:57

as just said and we know that

22:00

the R&D budget for defence globally

22:02

is by far

22:04

the biggest R&D budget that exists. So

22:06

we know that these technologies will continue

22:09

to evolve much quicker in that domain

22:11

than potentially in others. But

22:14

when it comes to things like

22:16

social services, community sector, transportation, looking

22:19

at how these technologies are applied

22:22

in education, I think

22:24

that we can absolutely apply the same

22:26

kind of moral guidelines as we do

22:28

already in those domains. So

22:31

it is about making sure that we

22:33

continue to be responsible, that we continue

22:35

to pay attention to the way that

22:37

the technology has been changed and used,

22:40

and that we have mechanisms for

22:42

people to report and whistle blow

22:44

on bad players and potentially, whether

22:47

intended or unintended, potentially bad consequences

22:50

for people. And then there

22:52

do have to be, I think, some

22:54

consequences available for governments and

22:56

courts to use as well. I

22:58

know this is very general, but what

23:00

sort of consequences would be meaningful in that

23:02

realm? Well, I think if

23:05

the work that you're doing and the technology

23:08

that you're building and applying is creating harm

23:10

to people, then you ought to be held

23:12

responsible in the same way that you might

23:14

be held responsible for causing harm to people

23:16

in other ways. That's because

23:18

it's a new technology. It doesn't mean that

23:21

you are absolved of that responsibility. The

23:23

examples are all around us, and new technology is

23:25

just a new tool. So the

23:27

people behind it ought always

23:29

to be held responsible for using it

23:31

appropriately and responsibly. What if it's not

23:33

obvious who's behind it? No, that's tricky,

23:36

isn't it? That's the challenge

23:38

for law enforcement, right? They're going to have to

23:40

keep pace too. I mean,

23:42

like we're seeing that already, you know, you look

23:45

at in the area of cybersecurity and you're seeing

23:47

it being much easier to launch malware attacks and

23:49

things like that. And once sort of AI gets

23:51

involved in that, it's going to be even

23:54

bigger and there's going to be a little bit

23:56

more enforcement to do, except to sort of try and disrupt

23:58

it at the source that says... much harder to hold

24:00

anyone to account for that. We've talked

24:03

a lot about things like chat GPT and

24:05

military applications, but AI has filtered its way

24:07

into a whole host

24:09

of areas we interact with,

24:11

healthcare finance. Kylie, as

24:13

we look forward to the next year or

24:15

so, is there a particular area where you

24:17

expect its use to massively explode? I mean,

24:21

I think we've got some

24:23

really interesting opportunities around modern

24:26

manufacturing, bespoke manufacturing and logistics,

24:28

which doesn't sound terribly planning when

24:30

you say that, but think about

24:32

the efficiencies that it might bring,

24:34

and that will open up the

24:36

capability to establish a genuinely circular

24:38

economy, so an economy in which

24:40

we have zero waste. That's

24:42

where we're hoping it will head. So when

24:44

you think about it, if you're building

24:47

a product, you can embed sensors into

24:49

that product that will track the product over

24:51

its lifetime. You can create it in such

24:53

a way that it can be dismantled and

24:55

reassembled as a different product, and

24:57

so that you look at not recycling,

25:00

but reusing in new ways, repurposing. So

25:02

you don't just have a life cycle,

25:04

you have multiple life cycles. And then

25:06

at the end, because you've embedded those

25:08

sensors, you can make the manufacturer of

25:11

that product responsible for its decommissioning at

25:13

the end of its multiple life cycles.

25:16

And so that opens up possibilities for

25:18

a really much more responsible approach to

25:20

environmental management when it comes to waste.

25:22

For example, we can look at, again,

25:24

the efficiencies in transportation coordination. I

25:27

think that's a really tricky

25:29

problem to manage congestion on roads

25:31

and designing the most effective bus

25:33

routes and train timetables and making

25:36

sure that public transport

25:38

is appropriate for the ways

25:42

in which people want to use it, which we know it's

25:44

not in many places around the world. So

25:46

bringing AI to solving

25:49

that problem could Not

25:51

only help people get to places faster

25:53

and more efficiently, but it'll help save

25:55

money for councils and local governments as

25:58

well. What

26:00

did the application coming in? The next

26:02

couple years and not necessarily the things that

26:04

you might think homicide right sizing but I

26:06

think that materials it consisted like the people

26:08

these Justin's on that we've been sitting here

26:10

telling I got not one but two separate

26:12

emails about taxi be say being used for

26:14

legal advice other areas and with Ai is

26:16

why could be used in the next couple

26:19

years he think people should keep their eye

26:21

on am I mean I think the legal

26:23

one is is quite an interesting question because

26:25

there are in are politically standardized forms they

26:27

need feel armed and for that that that

26:29

process can be automated. but we've already seen.

26:31

People. Getting in trouble where they've had

26:33

oh the try to me advil the

26:35

legal argument put in front of a

26:38

judge in it's quoted some some more

26:40

the doesn't exist in his occurs at

26:42

the I hallucination working in that I

26:44

think that's probably. That's

26:46

probably the sort of the area we probably

26:48

need to most the live on my own.

26:50

Look at, its it's almost seems where are

26:53

we going to see people integrating into lives

26:55

without particularly thinking about? Ah, the the ramifications

26:57

in the consequences of it and and been

26:59

caught out for using a my substantially should

27:01

have been using a nothing about boat ultimatum?

27:04

Sat down to this: This is all happening

27:06

very quickly and we don't really have the

27:08

rules in place. hit so it it's can

27:10

be Sort of mulling our way through a

27:12

little bit. We are unfortunately at a time

27:15

sued. Thank you to I. Guess wake just

27:17

highlights. Report with the got into some of

27:19

the joining us thanks enemy and the see

27:21

our of the A Strain Academy of Technological

27:23

Sciences and Engineering. Kylie walk out It was

27:25

an absolute pleasure. Thanks so much. Like

27:27

I think you max out. If you enjoyed

27:29

the program make sure you leave a review on

27:31

which ever podcasting app you happen to listen to

27:33

us on and of course you can find by

27:36

the South and all I B C podcasts and

27:38

I B C with them or my and L

27:40

and thank you for listening to another episode of

27:42

them I decide. You've

28:00

been listening to the WebEx

28:02

Live. Please join

28:04

us in welcoming the

28:08

audience. Please

28:10

join us in welcoming

28:12

the audience.

28:16

Please join the

28:18

webex live. Please

28:20

join the webex

28:22

live. Please join

28:26

the webex live. You've

28:29

been listening to an ABC

28:31

podcast. Discover more great ABC

28:33

podcasts, live radio and exclusives

28:36

on the ABC Listen app.