0:00

ABC Listen, podcasts,

0:02

radio,

0:03

news, music and more. Is

0:12

it time to start regulating AI?

0:15

Well, the US President Joe Biden certainly

0:17

thinks so. Yes, this week on Download

0:19

This Show, the US has made

0:22

some pretty big demands of the artificial intelligence

0:24

industry. Exactly what is that going to mean for

0:26

the future? We will find out. So

0:29

Meta is introducing an ad-free

0:31

subscription tier for things like Facebook, and

0:33

we look at why one robot taxi company

0:36

has kind of failed to launch. All of that and much

0:38

more coming up. This is your guide to the week

0:41

in media, technology

0:42

and culture. My name

0:44

is Mark Finnell, and welcome to

0:46

Download This Show.

0:57

Yes, indeed, it is a brand new episode of Download

1:00

This Show and our guest this week, the head of content

1:02

at ByteSide, Seamus Byrne. Welcome back.

1:05

Good to be back, Mark. And Dr. Erica

1:07

Mealy, lecturer in computer science at the University

1:10

of Sunshine Coast. Definitely not USC

1:12

though, right, Erica? No, no,

1:14

no, not USC,

1:15

UniSC. Come

1:17

on, get with the program. UniSC.

1:20

So Seamus, let's start with you. There has

1:22

been some big announcements about AI coming

1:24

out of the US. Walk me through it.

1:27

Yeah, President Biden has signed a

1:29

big fancy executive

1:31

order, basically sending a signal

1:33

to the artificial intelligence industry,

1:37

which of course has exploded over the last

1:39

year, that the US government

1:41

is watching closely about what's happening right

1:43

now. And he's getting a whole range

1:45

of different departments to actually start doing

1:47

some really specific investigations into

1:50

everything from what kinds of

1:52

security risks could be attached through

1:55

to consumer privacy implications,

1:58

all this sort of stuff, really trying

1:59

to now push forward with that

2:02

idea of saying we need to get on the front foot

2:04

as a regulatory industry

2:07

and actually govern this stuff. Is there

2:09

anything Seamus in the remit

2:11

that you think they've missed that they should be putting on

2:13

the table? Look,

2:16

I think that's actually a good question and

2:18

I think that they will start to kind of get these different

2:20

departments to bring in some of

2:22

that extra expert help

2:24

rather than just trying to make it up themselves

2:27

based on not really understanding

2:29

it. Erica, your thoughts?

2:31

Well, I find it really interesting that some of the commentators

2:34

have been saying that this executive order doesn't

2:36

have enough teeth because

2:39

AI pictures can really have too many

2:41

teeth. And so the fact that they

2:43

picked the word teeth... Not enough fingers.

2:45

Too many teeth and not enough... No, not

2:47

enough fingers, too many teeth. Yeah. So

2:50

I don't see that in their executive order. We want humans

2:52

that look like real humans and

2:54

the pleasing part of it was alongside

2:57

this executive order, they've also come out and talked

2:59

about setting up a safety in

3:02

AI and an ethics in AI committee

3:04

through their commerce department. So I think

3:06

they really are going to try and put some

3:09

strength behind it and hopefully eventually some

3:11

regulation. But there is questions

3:13

if it's just a PR exercise to kind of

3:15

remind everyone that US are still

3:18

a big player while the UK summit is going

3:20

on at the same time, while the EU are

3:22

looking at their act, China has already beat them out

3:24

with regulations. So there's

3:26

a few people that are a little bit cynical, but I think overall,

3:29

hopefully it's a move in the right direction. What

3:31

actually can be done in terms of regulating it at the

3:33

moment, Erica?

3:34

Because it's so worldwide

3:36

and spread, I think part of the

3:38

problem is we don't know what we don't

3:40

know. There's no sort of

3:43

understanding of what is the data

3:45

being used for? Where is the data going? How

3:48

are they going to then use it further?

3:50

What are these training data sets? And how

3:52

are they trying to future

3:55

proof the tools? So we

3:57

actually have some kind of consistency.

4:00

So there's some interesting concerns and

4:03

hard to regulate, but I think until there

4:05

is some kind of law that says you must

4:07

or you shall not, there

4:09

won't be any kind of regulation or any

4:11

kind of holding back on what we're doing.

4:15

What do you think, Seamus?

4:16

Yeah, look, I think regulation is

4:18

also one of those really tricky points right now

4:20

where there are a few companies that have exploded

4:23

into the lead. And they're also

4:25

quite often being proponents of saying, yes,

4:27

this is dangerous and we need to regulate. A

4:29

lot of people do feel like that's partly

4:32

because they kind of feel like it's

4:34

a lot easier for them to manage regulations

4:36

now that they are gigantic

4:39

companies that have swallowed up large

4:41

percentages of the entire internet to feed

4:44

their training data right now. And

4:46

it would be really nice to stop other companies from

4:48

sort of chasing them down. So there

4:51

are a lot of those kinds of aspects

4:53

of regulatory capture that kind of get fed

4:55

into that. And there are elements in

4:57

some of these orders and codes of conduct

4:59

and different things that are coming out where they're actually trying

5:02

to say, if you're below a certain

5:04

scale, we want to let you actually

5:06

continue to run a

5:08

little bit more freely for a while in the name

5:10

of ensuring that we don't just end up with Google,

5:13

with Microsoft, with OpenAI and

5:15

Facebook basically controlling this whole

5:18

next wave of technology after they've just

5:20

controlled the last 20 years. I'm glad you brought

5:22

that up. Erica, how have the big tech companies

5:24

reacted to this announcement out of the White House?

5:26

To be honest, I actually haven't seen a great deal

5:29

of their reaction. But I think along the lines

5:31

of OpenAI and many of the others

5:33

have said, yes, we think it is very dangerous.

5:36

So I would imagine that they're going to be

5:38

supporting this, probably up until

5:40

the point where someone says, actually,

5:42

you've taken a whole pile of stuff you weren't supposed

5:45

to take it out. Because the way

5:47

the AI works, we don't know which

5:49

decisions have come from where. So take

5:52

it out means start from scratch. And I don't think

5:54

they'll support that. But I think on the

5:56

whole, they're keen to be able to be within

5:58

regulation. Like Sheamus was saying,

6:00

it's definitely something where now

6:03

that they're the big fish, they get to control

6:05

it a little bit more. On the whole, they're

6:07

keen to be part of the discussion and

6:09

maybe lead it into a direction that helps them

6:12

in the long term. Was that your experience,

6:14

Sheamus? Do you think that they are going to react that

6:16

way?

6:17

Yeah, look, I think that's kind of a

6:19

good fit for what's going on. And again,

6:21

especially when we're talking about different

6:24

jurisdictions as well right now, like literally

6:27

in the same week as this announcement, we've had the

6:30

UK held a big AI summit

6:32

with lots of countries involved and 20 something

6:35

countries all signed

6:37

an agreement as part of that. But of

6:39

course, the EU is always expected to

6:41

become in a lot stronger with these sorts of

6:43

laws. And so I think right now

6:46

the American companies would probably find

6:48

these executive orders quite

6:51

nice in that, again, it's not pushing

6:53

too hard. It's about really kind of

6:55

saying we're going to monitor harder and we're going to start studying

6:59

more of what's going on here. But

7:01

it still gives plenty of runway. Whereas

7:04

I could imagine that in the next few months, we might

7:06

see a lot stronger action

7:09

out of the EU when it comes to actually what are

7:11

you doing with people's data online.

7:14

And yeah, that could kind of change the game in a

7:17

lot of ways. Download the show is what you're

7:19

listening to. It is your guide to the week in media

7:21

technology and culture. I guess this

7:23

week, Seamus Byrne, head of content

7:25

at Biteside and Dr. Erica Milley, a

7:28

lecturer in computer science at the University

7:30

of Sunshine Coast. Mark Fennell is my name.

7:33

And we move now to Canada. Canada

7:35

has announced it will ban the hugely

7:37

popular app WeChat on government

7:40

services, Erica. But

7:42

why?

7:42

Well, this is an interesting

7:45

one, because not only have they come out

7:47

and said we're banning WeChat, which

7:49

a lot of countries are seeing very similar to TikTok

7:52

because they're Chinese owned. So

7:54

they're seeing it as a security

7:56

risk that we need to actually control.

7:58

But they've also come out and made. Kaspersky,

8:01

which is a Russian cybersecurity company

8:04

as well. So they're not just targeting

8:06

China. They're really starting to look at,

8:08

well, what is available? What

8:11

is potentially exposing this information?

8:14

And how do we actually deal with

8:16

it? And interestingly enough,

8:19

Australia seems to be a little

8:21

bit behind, as we were the last

8:23

one of the sort of five eyes that we talk

8:25

about, to Ben TikTok. And

8:27

despite the fact that our Senate committee actually recommended

8:30

back in August that we should also ban

8:32

WeChat, we haven't done

8:33

it yet. Seamus, good

8:35

call or overreach?

8:38

Look, I think one of the easiest answers

8:40

about this whole thing, right, is that when

8:42

it comes to, you know, what is an app for?

8:45

And then how does its security

8:47

system work related to that task? WeChat

8:50

is fundamentally a messaging app that

8:52

does not support end-to-end encryption.

8:55

It is genuinely set

8:57

up in a way that means it is

9:00

able to be comfortably monitored by the

9:02

Chinese government. And so using

9:05

it for general chats and

9:07

whatever you might sort of think, I'm just going to connect

9:09

with other people through this app, it

9:11

makes total sense to not let this be on

9:13

government devices. You know, I looked

9:16

up sort of a quick background on the safety

9:18

implications of the application. And,

9:20

you know, it is noted for the fact that

9:23

if people in China send messages to each

9:25

other on WeChat and somebody says something that's negative

9:27

to the government, that it will be removed

9:30

from the app during that chat. Like

9:32

it is being directly monitored in those kinds

9:35

of ways within that Chinese environment.

9:37

Now, you know, they might sort

9:39

of claim that it isn't monitored in the

9:41

same way as all around the world, but it ultimately

9:44

flows through centralized WeChat

9:46

servers and doesn't have any encryption

9:48

attached at all. So it makes total

9:50

sense that this is not an app that

9:53

you should be using when it comes to thinking

9:55

you're having a private conversation with anybody. Are

9:57

there other countries around the world that have banned WeChat?

10:00

I know we've talked about TikTok, but are there other

10:02

countries that have banned or throttled WeChat

10:04

around the world? Does anyone know?

10:06

Yeah, I'm not sure, to be honest. I know TikTok has got

10:08

a lot of news around its bans,

10:10

and it's surprising that WeChat has been so

10:13

far behind TikTok in

10:15

that. They've really flown under that radar.

10:17

Why do you think that is, by the way? I'm

10:18

not really sure, but the thing that gets me is that Elon

10:21

Musk has actually said that he wants

10:24

X slash Twitter to become

10:26

like WeChat. He wants it to be the everything

10:29

app. And so, whether

10:31

it's just because it's been flying

10:33

under the radar, whether that perhaps people

10:36

in those environments, I mean, perhaps people

10:38

are less likely to put it on their work devices.

10:41

I mean, I just say, why would you do that? If

10:44

you're working in a security-intense environment,

10:46

why would you put TikTok or WeChat

10:49

or Facebook or Instagram, any of those onto

10:51

your feed? I don't understand that at all. Do

10:54

you think WeChat cares about this move,

10:57

Seamus?

10:58

Look, Yeah, WeChat is owned

11:00

by Tencent, and they are one of the biggest, certainly

11:02

like game developers in the world. But

11:04

as Erica was saying, that this is within

11:06

its core market. It is this everything

11:09

app these days. So I think

11:11

you can send payments to people. You

11:14

can do all sorts of really useful functions

11:16

through that app if you're in the kind of environment

11:19

where it is used by a really

11:22

large percentage of the population.

11:24

So there are lots of reasons why

11:27

it's great

11:28

for its core markets,

11:31

and its core markets happen to be the biggest

11:33

markets in the world. So yeah, I

11:35

think it's a bit of a blip by

11:37

comparison. In fact, probably the biggest issue,

11:40

and again, this is government-focused, so it's kind of not

11:42

as widespread as it might otherwise be. But

11:44

when there's often been talk about banning things like

11:47

this, there is a reasonable

11:49

problem with actually causing

11:51

problems for Chinese folks who

11:54

live outside of China and might need

11:56

to use it to continue staying

11:59

in touch with their families. family back home and all

12:01

that sort of stuff. So I think widespread

12:04

bans, not necessarily all that functional

12:08

or it becomes a problem for people. Whereas

12:11

saying that people who work in government

12:14

can't use it, that really does sort

12:16

of feel like a far more reasonable idea

12:18

when it comes to risk management. Download the show

12:21

is the name of the program. It is your guide to the

12:23

weekend media, technology and culture. And

12:25

interestingly is that Facebook and Instagram

12:28

are launching subscriptions in most of Europe. That

12:30

will remove adverts

12:33

from the platform. Now, hypothetically,

12:36

Erica, if you had an option to have

12:38

an ad free experience of

12:40

Facebook, would you take it?

12:42

Interesting question. Honestly,

12:45

I'm a bit of a cheapskate. So I probably just live

12:48

with it. Also, I like

12:50

a number of my friends do tend to

12:52

use pseudonyms on our Facebook accounts. So

12:54

what it's tracking may not have any

12:57

bearing on the reality. And so

12:59

in that case, they can serve me whatever they like.

13:01

But I have

13:03

a fundamental problem with you

13:06

pay to have no ads or you leave

13:08

the platform. Honestly,

13:11

they're not great options. That's not what the

13:13

EU consent laws are really about. They're

13:15

really about do you choose to be part

13:17

of this and do you want to continue

13:19

to be part of it? But it is

13:22

interesting that they've come up with this idea. Seamus,

13:24

do you agree?

13:26

I mean, I feel like I would try it for at least

13:28

one month. That would be my take. Just

13:30

at least get a good look at it and see how

13:32

different does it feel. I know I'm one of those people

13:34

who pays for for YouTube premium

13:36

because it really does make a difference in

13:38

your life if you're not seeing a million Facebook

13:41

ads. I'm sorry. Yeah, YouTube ads.

13:44

But yeah, I think with this one, the

13:46

really sort of big thing here is exactly

13:48

kind of like what Erica was mentioning, it's

13:51

going to be so important for Facebook

13:53

in its battle with how is

13:55

it using the data of European

13:57

citizens that they're

14:00

trying to claim, well, if we give them the option to

14:02

pay, that is effectively the opt-in

14:05

or opt-out when it comes to being tracked

14:07

for advertising. And I think we're

14:09

pretty quickly seeing that the

14:12

EU itself is looking

14:14

to respond to that and basically say, that is

14:17

not the option at all. That is not the either-or.

14:20

And actually, people should be able to make a

14:22

much more informed choice about saying,

14:25

I don't think I should have to be tracked in order

14:27

to stay in touch with all these

14:29

other people who happen to use this same platform. So

14:32

if you are in the EU, you'll end up paying around 10

14:34

euros per month for this

14:36

ad for experience. Erica, how do you think

14:39

their price point is going? Is it too expensive or not expensive

14:41

enough?

14:42

Well, it seems to

14:44

be less than X and Twitter.

14:47

That's always a good thing to be less than Twitter. Less than

14:49

Twitter on every front would be great. And

14:52

TikTok, though, is also

14:54

a little bit on the cheaper side. So it seems

14:56

to be about the middle of the road. Whether

14:59

that's the best price point or not, it will be

15:01

interesting to see. It's also interesting

15:04

that if you pay via the app

15:06

stores, you pay a premium because

15:08

an amount has to go back to Google or to Apple.

15:11

So they're actually saying, please don't pay via the

15:13

app store. Come and pay us directly, which

15:16

is a very interesting

15:18

way to be able to say, oh, yeah, but they're taking their

15:20

slice. It's like, yeah,

15:22

but how many people

15:23

did you pick up by being in that app store? Do

15:26

you think ultimately it will work? Because,

15:29

OK, the reason I asked Seamus is because it

15:31

feels like a thing that's been done in reaction

15:34

to legislation rather than a thing that's

15:36

been done for business reasons. Yeah,

15:39

that's right. And so I think in that sense,

15:42

this is probably framed around the

15:44

idea that they know they're not going to make money

15:46

from it. I think they're very good company at getting

15:48

data from experiments like this.

15:51

They'll probably find some interesting demographic

15:53

data out of who does decide

15:55

to opt in or out of this particular

15:58

system. But I don't expect. they're

16:00

thinking that it's going to replace

16:02

it in any way, particularly with the

16:04

idea that from March next year, they're even making

16:06

it more complicated by saying that if you

16:08

have a business account and your personal

16:11

account, that those accounts will

16:13

have separate fees attached

16:15

so that you'll have that first 10 euros

16:18

a month, and then you'll need to pay

16:20

an extra six euros a month for

16:23

each extra account that you run through

16:25

your account setup. And again,

16:28

that even feels like the kind of additional structures

16:31

that will probably start to make

16:33

the EU authorities even more annoyed

16:36

at the fact that this is not the solution

16:38

they were thinking should be brought

16:40

in for this kind of problem.

16:42

This is really interesting that

16:44

they're looking at doing this because after X

16:46

has added its subscription costs

16:49

alongside all the other crazy

16:51

changes that Elon has made, they've

16:53

actually found that the primary

16:56

user group in Twitter

16:58

is now sports fans. So

17:00

going from a very tech heavy platform,

17:03

suddenly it's the people cheering

17:05

on major league baseball that really

17:07

can't do without Twitter. So

17:10

it would be interesting to see, because they're talking about

17:12

the number of users if Facebook and Instagram

17:14

having increased with the death

17:16

of Twitter, how this actually

17:18

works over on that platform. So that doesn't actually surprise

17:20

me at all, because the last thing,

17:22

the last live thing that people genuinely

17:24

watch on television will be sport.

17:27

So it kind of makes sense that the thing that

17:29

you have to engage with live and in real time,

17:32

that's the last holdout for Twitter. In a sense,

17:34

it doesn't overly surprise me, Erica.

17:37

Seamus, what are your thoughts? Yeah, look,

17:39

I think there's a really important sort of additional

17:41

factor on this Facebook EU

17:46

subscription offer. And that is the fact

17:48

that within days

17:50

of this having been announced, and actually it

17:53

should kick in this week, is

17:56

the fact that actually the European Data Protection

17:58

Board has issued a decision

18:01

that actually says that Facebook and Instagram

18:04

cannot continue to use

18:07

private data to target their

18:09

behavioral advertising. This

18:11

is going to become, I think, a very big

18:14

talking point, I'm sure, in coming weeks, on

18:17

how Meta responds to the fact that the EU

18:19

has now effectively said that you are banned

18:21

from using data

18:23

to actually target behavioral advertising

18:26

on your customers. It could have

18:28

a huge impact on their ability to sell

18:31

ads in Europe.

18:32

Download the show is the name of the program. What

18:36

happens when a robot taxi

18:38

or robot taxi

18:40

decides to shut down

18:42

in the name of trust,

18:44

or so it seems? Shamus, talk me through what's

18:46

happened with a US company called Cruise. Most

18:50

of the headline reports have pointed out that

18:53

Cruise is voluntarily pausing

18:56

its autonomous fleet of robot

18:58

taxis, which have heavily

19:00

been driving around San Francisco over

19:02

recent months because

19:04

they want to make sure they want to, quote,

19:07

earn public trust and

19:09

restore some faith in their operations.

19:12

It also just happens that a couple of

19:14

days before they announced that

19:16

they would voluntarily do this, that

19:18

the California DMV suspended their robot

19:21

taxi permit effective immediately. Unrelated.

19:24

Unrelated fact, surely. Yeah. They

19:26

have operations in a couple of other cities,

19:29

I think, that have allowed this sort of testing. So

19:31

they are definitely pausing in some places

19:34

that they were allowed to continue. But

19:37

it's pretty dramatic given that it was just August

19:40

that California expanded their

19:42

allowance to run robot

19:44

taxis in San Francisco. What

19:47

does this mean for the future of, I

19:49

mean, not just driverless cars, but sort of robot taxis

19:51

in particular, because I do think that it's

19:53

one of those ideas that sounds

19:55

great, but actually in practice, there's

19:58

a whole bunch of unexpected kinks in the future.

19:59

the story Erica.

20:01

It's definitely something I was genuinely

20:03

shocked when I first heard about it that San

20:06

Francisco were willingly letting

20:08

their citizens basically

20:11

be crash test dummies for these these vehicles.

20:13

There's you know this poor woman was dragged 20

20:16

feet apparently underneath the robo taxi.

20:19

I mean allegedly it's horrific

20:22

and allegedly though it was caused by

20:24

another human driver but they've

20:26

hit trees, they've hit fire

20:29

engines. There's I think my favorite

20:31

part though of the whole shenanigans

20:34

is that there was a guerrilla protest

20:36

movement that discovered that you could disable

20:39

the taxis by putting a traffic cone

20:41

on their bonnet. So there were people riding around

20:44

San Francisco and putting traffic cones

20:46

on bonnets of cars to try and protest

20:49

the robo taxis. Look I

20:51

really think it's going to put a significant

20:53

dent in the idea and

20:56

I was shocked that they came

20:58

to 24-7 so quickly. The idea of a 3

21:01

a.m. when there's no other cars on the road

21:04

seemed to be quite a sensible idea. Let's

21:06

get drunk drivers out of their cars, let's

21:08

get people into these vehicles and if they

21:11

are the only vehicles around then

21:14

you really are constraining the independent variables.

21:16

You're really locking it down from that computer-centered

21:21

design perspective but I think

21:23

it's going to take a while to be

21:26

able to recover and I don't think anyone

21:29

believes that this is more than

21:31

a PR exercise. Oh we're volunteering

21:33

it for trust.

21:35

No, you were

21:36

banned and this

21:39

is not for trust but good try,

21:41

good try marketing department.

21:42

So what would it actually take?

21:45

If we take it on face value shames, what would it

21:47

actually take to rebuild trust

21:49

in a service like this? Yeah

21:51

there's definitely a bit of a going back

21:53

to square one element here of getting

21:56

back to basics on the

21:58

way in which these things are being tested. the kinds

22:01

of inputs,

22:03

like Eric was saying about controlling your variables,

22:05

it's like, what are the inputs that these are actually responding

22:08

to? Because that idea that there's

22:12

been lots of these second order effects

22:14

when it comes to the way that they've been involved

22:16

in accidents. So that

22:19

fire engine incident, it was the fact that

22:22

it was driving into an intersection where it

22:24

had a green light. So it was making all

22:26

of the normal assumptions, but it somehow

22:28

wasn't noticing or hearing

22:30

that there's a fire engine steaming through

22:33

this intersection. And that's how you know that

22:35

a robo taxi was never raised with Thomas the Tank

22:37

Engine.

22:37

That's right. But

22:40

also, is it possible to actually get to that stage

22:42

Erica without having cars on the road? Well,

22:45

there was a talk I went to many, many

22:47

years ago, which the RACQ were discussing

22:50

and they firmly believed that

22:52

the only way to commingle

22:55

autonomous cars and people was to have them

22:57

physically divided. So in sunny

22:59

Queensland, we have our busways that are completely

23:02

removed from the road. And so if

23:04

the autonomous vehicles were able to use some

23:06

of these ones where you're limiting

23:09

the chaos that humans cause,

23:12

then possibly it could happen. But then at

23:14

the same time, if you're making them go on a busway,

23:16

why don't you get on a bus? So

23:19

there's those questions that it

23:21

raises. But you know, do we divide

23:23

the motorway so that there is an autonomous vehicle

23:26

lane and or things like

23:28

that? Take them out of city centers, perhaps

23:30

where there's lots

23:31

of people, lots of cars and lots of unpredictable

23:33

events that are going to happen. But

23:36

yeah, downtown San Francisco

23:37

has a lot to try and grapple

23:40

with to work out if they can make this work. But

23:43

one of the essential pitches of

23:46

driverless cars is that if you have enough driverless

23:48

cars on the road, they will be more efficient

23:50

because they'll understand the roads better and they'll

23:53

be able to produce less traffic in

23:55

time. Like if you start to give them their own lane,

23:58

are you?

24:06

And

26:00

I think even in the context of passengers

26:03

and pedestrians and people, there's

26:05

a lot of the city centres that are moving to no vehicles

26:08

full stop and having those kinds of concession

26:11

charges and congestion

26:13

charges and not letting the cars

26:16

intermix with people, let alone autonomous

26:18

ones. So

26:19

we have to think about that context too. Just

26:22

so we're absolutely clear, Niamh, the producer, shook her head

26:24

so hard at me when I made the wheels reference.

26:27

Seamus, do you think the idea

26:29

of, for lack of a better term, segregating

26:32

driverless cars from driver-full

26:34

cars make sense? Yeah, certainly

26:36

in the way that we do

26:39

have bike paths and bus lanes, all these

26:41

sorts of things, I think that idea of sharing

26:44

bus lanes could be good within a certain volume

26:46

context. But the thing that really does strike

26:48

me about all of this testing that's happened

26:51

is when I think maybe

26:53

there's a lot of lessons that have

26:55

been learned and data that could have been gained,

26:58

but then you have to stop and realise these

27:00

are private companies who are not sharing their data.

27:02

And I feel like that is perhaps

27:04

the thing that we need to revisit, is if we're going

27:07

to let these people test these

27:09

kinds of vehicles in public spaces, then

27:11

they should be doing it in a way that actually elevates

27:13

the entire autonomous

27:16

vehicle industry and that they have to share their

27:18

data publicly and create

27:20

a much more of a public

27:23

open source of a sense of how

27:26

these things are working and what they've

27:28

been learning through these processes, and

27:30

especially when there's been any kind of an incident

27:32

at all, because that should be the kind of process

27:34

here. If we're going to let them do this, then they

27:37

should be getting better much faster. But

27:39

if it's just private companies being allowed

27:41

to do this and keep that data to themselves,

27:44

then actually, yeah, we will ban

27:46

them in this case and then none

27:49

of us will be better for it. And with that,

27:51

we are out of time. Huge thank you to

27:53

our panelists this week. Seamus Byrne, head of content

27:55

at Bikeside. Thanks for joining us in the show.

27:57

Thank you. And Dr. Eric Amelie, let's...

28:00

in Computer Science at the University. Sunshine

28:03

Coast, thank you so much. Thank you,

28:05

pleasure as always. And with that, I shall

28:07

leave you. My name is Mark Fennell, and I'll catch

28:09

you next week for another episode

28:11

of Brown notation. Thank you.