Episode 4: ‘The road ahead for driverless cars’ with Professor Michael Milford

00:10 – 00:23
Raph Goldenberg: Hello and welcome to Legally Consumed, a consumer products podcast by a consumer products law firm in Australia. I’m Raph Goldenberg and I’m joined by co-hosts Kaye Ho and Will McMinn, who are members of the team at CIE Legal.

00:23 – 00:37
Kaye Ho: We chat with executives of consumer products companies, their legal teams, and industry experts who give us a peek into their journeys as people. We explore industry-changing ideas and even share tips and tricks on how to navigate the consumer product space.

00:38:00 – 01:02
Will McMinn: This season we’ll feature guests from the automotive, retail, advertising and insights, food, beverage, and primary production and franchising sectors.

~ Music Interlude ~

01:04 – 01:39
Raph Goldenberg: Hi everyone. Welcome to Legally Consumed, a CIE Legal Podcast about consumer products by the consumer product lawyers. Today we have a very special guest, Professor Michael Milford. Michael is a professor at the Queensland University of Technology and is an expert in robotics, as well as driverless cars or autonomous cars. And that’s all I’m going to say about Michael. Michael, maybe you could tell us a little bit about yourself, and if I could start by asking you, what was your first job?

01:41 – 01:47
Professor Michael Milford: My first job was very decidedly not technological. I worked at Donut King making donuts.

01:47 – 01:54
Raph Goldenberg: Right. Very good. And what did you do after that? And how did you end up where you are today?

01:55 – 02:15
Professor Michael Milford: Well, not to give you my whole life story, I guess a couple of highlights were that I was fascinated by science fiction when I was a kid. I loved science fiction books and I loved science fiction movies like The Terminator, fabulous movies to grow up with. Another formative experience that put me on my current career journey was I was a victim of hacking early on.

02:16 – 02:16
Raph Goldenberg: Oh!

02:16 – 02:24
Professor Michael Milford: And revenge motivated my desire to learn how to code and program. And then I never looked back from that point onwards.

02:24 – 02:30
Raph Goldenberg: Right. And so what did you do before you started teaching people? What did you study?

02:31 – 02:47
Professor Michael Milford: I studied mechanical and space engineering. The other thing that fascinated me was rockets. But rockets are a long payoff thing. You have to work on them for decades sometimes and sometimes they blow up on the launch pad on you’re back to score a zero. So then I pivoted into artificial intelligence and robotics, and that’s where I am today.

02:48 – 02:59
Raph Goldenberg: Fantastic. So you’re at Queensland Uni of Technology. How long have you been there and sort of what was the path to… because you look too young to be a professor.

03:00 – 03:26
Professor Michael Milford: Oh, thank you. Academia is a youthful game. Kids are very young on the outside. Definitely not young on the inside. I did a PhD with an amazing supervisor once again in the robotics space. I did a bunch of sort of research roles at a fairly junior level, got a, what you call a faculty job, at QUT in 2011. So that’s a whole 11 years ago. Sort of rose up through the ranks, became a professor, proudly joint director of the QUT Center for Robotics.

03:27 – 03:44
Professor Michael Milford: First part of my career was very much more on the sort of basic research side of things. But the last sort of… most of the last decade, we’ve been doing a lot more applied work with companies, government sort of translating a lot of the technology that we create into potential commercial applications and also a lot of outreach and engagement.

03:44 – 03:58
Professor Michael Milford: So making sure everyone understands what’s really behind these technologies that they’re not to be scared of, that they might not be able to do everything that people say they can do because an informed population and informed set of stakeholders is a beautiful thing to have.

03:59 – 04:21
Raph Goldenberg: Absolutely. So, I mean, I think one of the things that interests me is sort of this connection between artificial intelligence, human psychology, robotics, and this whole kind of… have you had to study how the brain works to kind of become an expert in artificial intelligence and robotics?

04:22 – 04:47
Professor Michael Milford: I’ve never formally studied the brain, but I was thrown into the deep end, basically from my first steps as a Ph.D. student all the way to today. I was really fortunate to be on some projects with some of the world’s most amazing neuroscientists and psychologists for example. And so through them, I informally learned a lot, just enough to be dangerous, I guess insight about neuroscience and cognition in the brain.

04:47 – 05:09
Professor Michael Milford: And people talk about how modern artificial intelligence and software is based on how we think the brain works. And while that’s not to say true at a very detailed level, it definitely draws a lot of its inspiration. And we’re only just starting to scratch the surface. We don’t really understand everything about how the brain works. So there’s a lot of really exciting work to be done over the coming years.

05:09 – 05:38
Raph Goldenberg: Which is a nice segway. I mean, we spoke earlier in the driverless cars space about the idea of deep learning. And one of the things that I found incredible you spoke about was, in effect, some of these AI technology in the deep learning sort of framework is based on the brain. Yeah, in the sense of all these different connections that are making it in the brain with a neural network.

05:38 – 05:50
Raph Goldenberg: Can you speak a little bit about that because it just makes me sort of think, gosh, the brain is such an amazing thing in itself that computers are now trying to copy it.

05:51 – 06:14
Professor Michael Milford: Yeah. And the brain is the most complex, sophisticated thing. As humans, we know it’s 80 or 90 billion neurons and a much larger number of connections between those neurons or cells in the brain. So we’re really just scratching the surface. So at a very high level, these artificial neural networks that people program into software that power robots power the product recommendations that you get when you shop online.

06:15 – 06:35
Professor Michael Milford: That’s all driven by this approach called, I guess, modern AI. And a lot of that is based on these very large artificial neural networks which have been put into software and then deployed. Another key aspect of it is they’re trained on very large amounts of data. So just like as a child, you were taught about the world by interacting with it at a very high level.

06:35 – 06:39
Professor Michael Milford: We trained these artificial intelligence systems in a broadly similar way.

06:40 – 06:58
Raph Goldenberg: Right now, I want to talk a little bit about driverless cars. Firstly, what are they? I’ve heard people call them autonomous vehicles. Is there a difference? Can you tell us a little bit about driverless cars at a higher level?

06:58 – 07:23
Professor Michael Milford: So the terminology varies as autonomous vehicles, driverless, and self-driving cars, they are probably the three most popular terms. Yeah. The good thing about this is it’s much easier to define than a term like artificial intelligence, which no one has really agreed upon for a definition. So an autonomous vehicle is essentially some sort of vehicle. It could be any size that to a large degree is able to drive itself in whatever its operational environment is.

07:24 – 07:39
Professor Michael Milford: That’s obviously a fairly loose definition. And so you can have a vehicle that moves passengers around an airport or you could have this sort of the widespread robotaxi dream of just having autonomous vehicles everywhere in all cities, which are the basic way that people get around.

07:40 – 08:02
Raph Goldenberg: Yeah, and one of the things I don’t think I necessarily realised until I spoke to you was that I’d almost always imagined driverless cars driving people around on the road. But they’re also used not to take passengers. So there are some autonomous vehicles out there in some applications, is that right? In that space where, you know, you don’t have a passenger in the car.

08:03 – 08:23
Professor Michael Milford: Yeah. And the passenger carrying robot taxi, so to speak, was the big sort of driver. That’s where all these trillion dollar market size valuations come from. But as people become a bit more realistic about what’s possible in the near-term, companies are starting to pivot and look at other business models. One of the most promising is autonomous delivery vehicles.

08:24 – 08:46
Professor Michael Milford: So this might be a vehicle that delivers your groceries just like you currently get your groceries delivered, except it’s not driven by a person. Few technical advantages to that. You don’t really care that much whether your groceries take 15 minutes or three or 4 hours to get to you. As long as the price is right, you can air condition or keep the groceries refrigerated in the vehicle.

08:46 – 09:06
Professor Michael Milford: And it’s also easier from a safety perspective because you’re not trying to protect any passengers in the vehicle. The problem is, of course, that there are already very well-oiled, optimised delivery business models, which are quite cheap for the consumer. And so how do you make an entry into that market where the market’s already quite saturated?

09:06 – 09:17
Raph Goldenberg: Yeah. Can you talk a little bit about how driverless cars work and sort of the technology behind them?

09:17 – 09:36
Professor Michael Milford: So at a broad level, there’s really two key components of how a driverless car works. There’s all the technical hardware on the car and that really comes down to the computers that it carries because most of the computation or A.I. is done on the vehicle itself. It’s not sent off to the cloud. And the other big component of the hardware, the sensors.

09:36 – 09:59
Professor Michael Milford: So cars need to see and understand what’s happening in the world around them. So they typically have a combination of radar sensors, camera sensors, a type of range detection sensor called LiDAR, perhaps sonar and a bunch of other types of sensors. And it combines all of this information together and then feeds it into the second key component of what powers the driverless car, which is the on board artificial intelligence.

10:00 – 10:12
Professor Michael Milford: And these are the systems that process what the car is seeing, make sense of what pedestrians and other vehicles are around the car and then work out how to drive safely and effectively through that environment.

10:12 – 10:29
Raph Goldenberg: Yeah, right. So we mentioned that at the moment they’re not necessarily connected to the cloud. Does that mean that some of the issues around security and hacking into an autonomous car, aren’t sort of, at the risk that you might imagine?

10:30 – 10:49
Professor Michael Milford: There’s definitely a risk in terms of hacking and they most definitely are connected to the cloud, but they’re not making their second-to-second decisions based on information coming down from the cloud. But they’re uploading the data. They’re getting data down from the cloud. Hacking is a concern. The good news there is this isn’t a brand new problem for autonomous vehicles.

10:49 – 11:06
Professor Michael Milford: We’ve had the same concerns or planes for smartphones and lots of other technical devices in our life. There are some variations. Obviously, a vehicle is more dangerous if it gets out of control than a smartphone potentially. But when I was starting from zero and there’s been a lot of good work there already.

11:06 – 11:29
Raph Goldenberg: And I think, gosh, if we can manage it with a plane, you just made me freak out about my next trip. The idea that someone might be able to hack into a plane, but then hopefully we can manage that in autonomous cars. So you gave us a bit of a snapshot about the delivery vehicles. Can you give us a bit of a feel as to how the technology is being used currently?

11:29 – 11:47
Raph Goldenberg: When I say, you know, I mean, I think there’s different types. We have obviously some autonomous systems where the car reacts to certain things on the road. I’m talking about, you know, actually lack of driver intervention. You know, is there much going on around the world at the moment on the roads?

11:48 – 12:10
Professor Michael Milford: In terms of trials where there isn’t necessarily a safety driver in the car, there are a number of trials. There’s trials by a few companies going on in San Francisco. If you’ve been to San Francisco? That’s a challenging environment to drive in apart from anything else. It’s very hilly. There are also trials in Phoenix and probably a number of trials happening in other countries like China as well.

12:10 – 12:12
Raph Goldenberg: And you’ve been in some of those cars?

12:12 – 12:33
Professor Michael Milford: Yeah, I’ve been joyriding in autonomous vehicles for the last 10 years. The first autonomous vehicle ride I ever got was the Google car in 2012. And for about five or 10 minutes, it was the most remarkable experience. And then I instantly forgot I was in a car that didn’t have a driver. And ultimately that’s a good sign.

12:33 – 12:39
Raph Goldenberg: Yeah, technology. I was going to say, was it a scary experience? Just that lack of control?

12:39 – 12:42
Professor Michael Milford: I’m a geek and this is my area, so it’s very exciting.

12:42 – 12:43
Raph Goldenberg: Yeah.

12:43 – 12:50
Professor Michael Milford: But then it rapidly became very mundane, except when the vehicle did something different to what you might expect a human driver to do.

12:50 – 13:21
Raph Goldenberg: Yeah. I’m also interested in how autonomous vehicles might change if, you know, assuming there’ll be a point at which they become adopted and become more mainstream. But how they might change our environment that we live in just outside? I mean, aside from the fact that, you know, people won’t be actually controlling the cars, a computer will. Can you talk a little bit about that and whether or not there’s been some thought of planning into how that will look?

13:22 – 13:49
Professor Michael Milford: There’s a lot of futurists around the world earning their living by talking about all of these possibilities, some backed up by evidence, some more speculative. If we take the middle scenario where autonomous vehicles are not perfect but they are good, one of the things you might see is a lot less people would own cars, there’d be a lot fewer vehicles, full stop on the roads and the layout of the city and how people commute or don’t commute to work and where they live could all change substantially.

13:49 – 14:06
Professor Michael Milford: So the notion of central business districts of where people tend to reside, the notion of there even being a rush hour, these are all things that could change substantially if you saw the number of vehicles drop substantially, ownership of vehicles dropped substantially because autonomous vehicles are here.

14:07 – 14:25
Raph Goldenberg: And I know this is in the realm of futurism, but at the moment I think people are so wired to having their own vehicle. But do you see a time in the future where people are happy to kind of share? Because obviously to reduce the number of cars on the road, that’s the only way to do it, isn’t it?

14:26 – 14:45
Professor Michael Milford: Yeah. And where I’ll just remind you, if you look at the precedent of other technologies that have come in, the one I always use at a much more minor scale is streaming media. So if you queried people before streaming and media became a thing, the idea of no longer owning you know, movies and music was really abhorrent to a lot of people.

14:45 – 15:09
Professor Michael Milford: But the younger generation who grew up with it, where the tech pretty much worked relatively cheap, they largely didn’t even think about it. And the same could happen with autonomous vehicles. You see the age at which people get their driver’s licences already with ride sharing vehicles being available, dropping in certain areas. And so there’s already a precedent that IF the technology works, younger people at least won’t even think much about it.

15:10 – 15:32
Raph Goldenberg: Yeah, right. And what do you… I mean, in terms of the impact on industries once this starts? Because I feel like despite all the naysayers, that this is sort of an inevitable development that’s with no knowledge, but it just seems like it’s heading that way. But the impact on industries and industries likely to be affected.

15:32 – 15:42
Raph Goldenberg: I mean, we do a lot of work in the automotive space. How will it affect the car companies, other industries? Do you have any thoughts on that?

15:42 – 16:05
Professor Michael Milford: Sure. So it depends on the exact scenario. If we go with the scenario where there’s just a lot fewer vehicles and obviously any company that makes its primary revenue from selling vehicles will need to change its business model and the whole sector may change. But there’s opportunities there too. In terms of the vehicle generating ongoing revenue. It can have some sort of shared autonomous vehicle service.

16:05 – 16:10
Professor Michael Milford: So there are new business models which may be just as attractive, but they’d require a major change.

16:10 – 16:22
Raph Goldenberg: Yeah, I guess one of them might be advertising if you start, you know, we see it in the lift as we came up before, there’s advertising everywhere. The more eyeballs you have, you know, sort of a public space.

16:22 – 16:40
Professor Michael Milford: And that has already happened with non-autonomous vehicles. Lots of cab companies around the world already serve you on turn off of all the non mutable ads. And you can imagine in the ultimate capitalist vision that your subscription service would be at different tiers. And the higher the tier, the less ads get forced upon you. Yeah.

16:40 – 16:53
Raph Goldenberg: Okay. Pay for the silence. Yeah, I get. So why are we doing this? Like, what are the benefits of driverless cars? Why is there so much academic and commercial interest in this topic?

16:54 – 17:16
Professor Michael Milford: So there’s a lot of rational motivation and also quite a bit of irrational motivation. So the irrational motivation is it’s an incredibly fascinating technical challenge and that’s what’s drawn a lot of these companies and these technical people to work on it. It’s like the space race for our generation, in some respects. The rational reasons are really economic and health.

17:16 – 17:41
Professor Michael Milford: So economics is if you become a company that’s successful in this space and manage to maintain a good market share, theoretically it might be quite lucrative for you, whether it’s autonomous delivery robots or passenger carrying robot taxis. From a health perspective, the numbers that are always bandied around is sort of the automotive road toll associated with a figure which is over a million globally and a much larger number in terms of injuries.

17:41 – 18:04
Professor Michael Milford: And so if you can make a significant dent in that, you have a huge beneficial impact on quality of life for lots of people that would have otherwise been affected tragically by this. It’s not that simple, though. I feel I should mention that the areas where it’s most economically viable to deploy autonomous vehicles are often the areas where the per capita road toll is the best.

18:04 – 18:10
Professor Michael Milford: And so the areas which would really benefit potentially from this technology might not get it for a lot longer.

18:10 – 18:22
Raph Goldenberg: Can you drill down a little, get into why we think that there will be a net reduction in road accidents as a result of autonomous vehicles?

18:23 – 18:48
Speaker 4
Humans are incredibly capable at what they do. The fact that we can drive probably shouldn’t be, but people drive while they’re tired, while distracted and still generally don’t crash is a testament to how incredible humans are at what they do. But ultimately humans are fallible and humans are generally getting better. Yeah, or not substantially better. Most of the road toll improvements have been around much better vehicle technologies.

18:48 – 19:03
Professor Michael Milford: It’s much safer to be in a crash now than even probably ten years ago. So the allure of autonomous vehicles or even assisted autonomous driving is that at least theoretically you can keep improving it, keep improving it until it’s about as close to perfect as you could imagine.

19:04 – 19:31
Raph Goldenberg: Yeah. And I think that’s one of the questions that in my mind when I’ve thought about this and heard you speak about it, it’s the thing that feels like it’s holding up adoption is the corner case, you know. Yes we might if you look at it objectively we might reduce car accidents and injuries and have millions of people saved as a result of the technology.

19:31 – 20:08
Raph Goldenberg: But there’ll be a certain number of unpredictable accidents. And my sense is that people just aren’t willing to let that be decided by a computer, that somehow a person should control that. What are the, I guess that’s more my personal little soapbox view of it, but for you, what are the risks that keep you up at night when you think about driverless cars or are you kind of more comfortable with the whole thing?

20:09 – 20:36
Professor Michael Milford: On a sort of practical near-term basis. The thing that did concern me a lot years ago, but isn’t as much of a concern now, was that these big, very powerful corporate players who are driving this technology race would come in with this product and sort of overwhelm local and federal governments. And as a result of it not being fully thought out in terms of their introduction, that essentially the public might get screwed.

20:36 – 20:56
Professor Michael Milford: I think I’m nowhere near as worried about that because now in all my interactions with government and bureaucrats, they are an order of magnitude more informed than they were probably even five or six years ago. And that’s a very healthy situation to be in because they’re going to run a pretty sceptical eye over any company trying to introduce this tech here, which is great.

20:56 – 21:17
Raph Goldenberg: That’s great. I mean, I’ll mention one risk, but what are some of the big issues that are still to be worked through? I guess because it seems like, and tell me if I’m wrong, but it seems like to a degree the technology’s pretty well developed. I mean, we have cars on the road, as you said, in San Francisco, driving around.

21:17 – 21:25
Raph Goldenberg: What are the issues, and think the big issues, that need to be worked through before we see, you know, autonomous cars adopted on a large scale?

21:26 – 21:52
Professor Michael Milford: There’s really two key issues. One is, are they technically capable of doing the job that we need them to do? And the second one is economic chaos. So for simpler autonomous tasks like autonomous delivery, you can argue that they’re probably already sufficiently technologically capable, but the margins aren’t as good economically. And so the economic case may or may not be there in terms of the Holy Grail, so to speak, of widespread passenger carrying automotive vehicles everywhere.

21:53 – 22:14
Professor Michael Milford: The economic case, although not as rosy as I think people originally imagined, is probably still there, but we don’t quite know yet technologically whether it’s off to scratch. The good news is from these trials that are being performed, maybe not the public, but some people will start to get some pretty compelling statistics, at scale about the exact viability of the systems.

22:14 – 22:15
Raph Goldenberg: Yeah.

22:15 – 22:41
Raph Goldenberg: I mean, I think we talked before about some of the infrastructure related issues. And I think one of the things that I’ve heard bandied around about autonomous cars is that they can’t really work without massive changes to infrastructure, which I’ve also heard spoken about in the EV space with everything. I think it’s a pretty clear case there.

22:41 – 22:55
Raph Goldenberg: I mean, you know, you do need to build a charging infrastructure. Would you agree with that as a proposition in relation to autonomous cars? Can we start using them without the, you know, the infrastructure and things being detected on the side of the road?

22:56 – 23:22
Professor Michael Milford: From the commercial side of things? Most of the companies would really like the tech, including all those vehicles to require as little modification by their customer. I’d say most of them are trying to design it to not require changes to the infrastructure from the public and government side of things. The potential benefit of autonomous vehicles is probably not great enough to justify investing billions of dollars and investing in modifying infrastructure.

23:23 – 23:36
Professor Michael Milford: And so that’s something that I at least would definitely suggest exercising caution around. You don’t want to prop up a technology that’s not ready for game time with unsustainable investment in things like massive infrastructure change.

23:36 – 23:56
Raph Goldenberg: ich is also something that resonates because I think I was sort of seeing that in Australia with EVs. I mean it just seems like there’s always been this argument and this reason we haven’t been able to, you know, get adoption of EVs in the markets. The OEMs [original equipment manufacturers] aren’t sending the cars here because there’s just not the infrastructure yet.

23:56 – 24:00
Raph Goldenberg: So I think there’s a learning coming out of that for autonomous vehicles.

24:01 – 24:16
Professor Michael Milford: Yeah. And they need their product to be commercially viable in all the other markets apart from Australia. So regardless of what we do, if those other markets require no infrastructure changes are going to have to solve that problem anyway. And yeah, just be an afterthought.

24:16 – 24:42
Raph Goldenberg: Yeah, it makes sense. Look, it is the topic of the podcast and podcast is Legally Consumed and it would be remiss of me not to think about some of the legal issues and talk to you about some of the legal issues associated with autonomous cars. I mean, as a lawyer, it excites me because, you know, as a lawyer you will always think about worst case scenarios.

24:42 – 25:10
Raph Goldenberg: Have you had to think or when you are in your discussions with companies that are involved in this around product liability accidents, insurance, all this sort of stuff, cybersecurity. Do you have any personal thoughts about that? And do you have a sense about how the researchers who are trying to, you know, deploy this technology are tackling those sorts of issues?

25:10 – 25:32
Professor Michael Milford: Yeah, cybersecurity is particularly topical at this moment and obviously that’s a concern. You can weaponise an individual autonomous vehicle and while you can take a lot of measures to protect against that, getting 100% guarantees on being safe is pretty much impossible to achieve. And these sort of systems, especially if it’s a consumer vehicle that’s out and about on streets.

25:33 – 25:55
Speaker 4
So who’s liable in that case? Can you insure against that? Those are all pretty big issues in terms of liability. One of the things I always say is you will be able to understand and dissect exactly what has gone wrong when an autonomous vehicle is involved in an accident. And you’ve seen this play out with some of the high profile accidents overseas.

25:55 – 26:27
Professor Michael Milford: They’ve been able to dig into the detailed records and basically work out exactly what went wrong. And that’s the technical side of things. How do you then attribute who’s responsible for that? My main concern and being very naive in this area is based on my understanding of how big companies have got away with other stuff. My main concern would be there won’t be sufficient liability traced back to the important people in these companies, but at the same time, we don’t want to stifle innovation by being overregulated and stuff as well.

26:27 – 26:29
Professor Michael Milford: That’s a tricky balance to tread.

26:30 – 26:54
Raph Goldenberg: They absolutely do. Where are we at in Australia, compared with other countries in relation to autonomous cars? And I think it’s probably two sides to that question. One is what you do. The research and you know, on the consulting side, the other part of it is the actual use of autonomous vehicles on Australian roads or in the Australian city.

26:55 – 27:14
Professor Michael Milford: So we have a small number of Teslas operating in Australia. So you could argue that some level of autonomous vehicle penetration. Government has come a long way at local, state and federal levels in the last probably five or six years. So they’re much more informed. They’ve run a lot of trials of various subcomponents of autonomous vehicle technology.

27:14 – 27:36
Professor Michael Milford: We’ve thought about how autonomous vehicles would adapt to Australian specific problems, like kangaroos is the one that’s talked about a lot. But also Australian signage, the factory drive on the left side of the road, all these sorts of things. So we’re in a pretty well informed position. The government is in a reasonable state to sort of adopt the technology if it comes.

27:36 – 27:58
Professor Michael Milford: There have been trials in various communities, so the public has probably seen an autonomous shuttle bus at some stage in a city or a sporting stadium. We don’t have a very vibrant technical investment ecosystem. So there aren’t many startups operating or almost none in this space in Australia. They’re all overseas. We shine in the off road autonomous vehicles area.

27:58 – 28:24
Professor Michael Milford: So we’ve literally led the world in developing autonomous vehicles for mining. And in mining, it’s a great flagship example because they’re not there for novelty value. They’re there because they make economic sense and there are very surprisingly few or maybe not surprising for many, there are very few enduring commercially viable deployments of robots or autonomous vehicles in the globe and autonomous vehicles and mining is one of those.

28:24 – 28:32
Raph Goldenberg: Yeah, I could imagine that a bit. Are there passengers in these mining vehicles?

28:32 – 28:38
Professor Michael Milford: There’s a variety. You’ll have load haul dump trucks operating underground.

28:38 – 28:38
Raph Goldenberg: Yeah.

28:39 – 28:53
Professor Michael Milford: In semi-autonomous mode or autonomous mode. You’ll have trucks above the ground operating autonomously, hauling or around, and then you’ll have remote operators supervising them from an air conditioned control room, sometimes thousands of kilometres away.

28:54 – 29:26
Raph Goldenberg: Yeah. And that’s the one that I’ve heard about. And I think the miners probably have a different appetite for risk. And just how quickly can we get stuff out of the ground? So maybe that’ll be an application that will drive some improvement in technology, you know, more. Do you see that as you know, where the money is, is probably what’s going to force the technology moving forward and the adoption maybe?

29:26 – 29:55
Professor Michael Milford: And it’s not just the money. Like it’s very edifying to look at agriculture and to look at mining and to look at the differences in terms of automation penetration into mining versus agriculture. People have thought a lot about this. People thought that the family based background to agriculture, there’s a different culture around technological innovation compared to mining, where it really is predominantly large and medium sized sort of company and players. Say what you will about mining.

29:55 – 30:06
Professor Michael Milford: They seem very willing to adopt technology that will probably give them another half a percent efficiency gain in their operations because they know exactly how that will translate into their bottom line.

30:06 – 30:07
Raph Goldenberg: Yep.

30:07 – 30:12
Professor Michael Milford: So there’s a direct accountability trial from the innovation to how this is going to affect the economics.

30:13 – 30:13
Raph Goldenberg: Yeah.

30:15 – 30:39
Raph Goldenberg: And can you tell us a little bit… we know there’s also military applications for autonomous vehicles, probably for obvious reasons. There are certain areas where you want to have vehicles going, but you don’t want to have people going in. Can you tell us a little bit about what’s going on in that space? Have you done any work in those industries?

30:39 – 31:04
Professor Michael Milford: Yeah, so we work extensively in aspects of defence. Defence is growing as an R&D field in Australia. It’s been interesting. So the US Defense Forces initially had this nominal goal to have a substantial fraction of the vehicles and platforms automated by some date, which I think has already passed. And then they realised even with their financing, that that was just unrealistic.

31:04 – 31:32
Professor Michael Milford: So they backed down substantially on that goal. But defence is taking a sort of long term view to this. In Australia, it’s not just about removing defence personnel out of harm’s way. People always talk about dull, dirty and dangerous. And dangerous is definitely an issue with defence issues. Part of it is just scale and force protection. So you have a limited size defence force in terms of absolute numbers.

31:32 – 31:47
Professor Michael Milford: In Australia, you have a very large landmass and surrounding sea area. And so it’s not just about removing people from danger. It’s just, you can’t cover the area without some sort of automated or semi-automated technology that massively scales up your reach.

31:47 – 32:09
Raph Goldenberg: Yeah, I mean, one of the things that are bring thinking about when you’ve been talking about the military side and yes, can protect people, but you think about drones and unmanned drones and the sort of the lack of accountability about the able to sort of drop a bomb from a drone which doesn’t without any risk to the you know, to a person flying over.

32:09 – 32:19
Raph Goldenberg: What are some of the ethical issues around autonomous vehicles? Not just in the military but on the road. And, you know, because I’m fascinated by that, that side of it as well.

32:20 – 32:53
Professor Michael Milford: So in civilian applications, people always talk about the hypothetical trolley problem whereby if the autonomous vehicle is in a split second accident situation, how does it decide what course of action to take? There’s problems with these hypotheticals because they’re absolute hypotheticals, and all humans know that nothing is ever certain. And so they have trouble reasoning through them. Speaking sort of at a 30,000 foot view, you would hope that all the key stakeholders would have some sort of generally agreed upon framework for minimising harm.

32:54 – 33:14
Professor Michael Milford: There’s no perfect definition of that. And that would be a definition that is constantly reevaluated by all the key stakeholders. And those stakeholders have to include vulnerable members of the public who tend to not have enough say in these technological matters, and that you could continually monitor and refine that process until it’s sort of maximally acceptable to everyone.

33:14 – 33:31
Raph Goldenberg: Yeah. And that trolley situation is, well the trolley example that you provided, how do I decide who to hit kind of, or I have to hit someone. And is that some scenario that people are tackling?

33:31 – 33:55
Professor Michael Milford: Yeah. And there’s infinite scenarios there’s also if you’re after a bit of lighter humour to distract you, there’s some animations online where people animate increasingly farcical, ridiculous trolley problems. It’s fun to go through, but a common example of this would be if you have a passenger inside an autonomous vehicle and there’s a situation where it’s either going to have to hit a brick wall or it’s going to have to hit a pedestrian.

33:56 – 34:16
Professor Michael Milford: And so these are really challenging situations because the passenger in the car has a lot more protection, like cars have awesome safety systems, but hitting a brick wall at 60 kilometres per hour is still going to have a high chance of bad outcomes. The pedestrian has no protection, is extremely vulnerable, so even a glancing blow could be fatal or cause serious injury.

34:16 – 34:30
Professor Michael Milford: And the car can’t model exactly what’s going to happen because there’s a degree of uncertainty. And so coming up with an acceptable framework for decision making, in that situation is challenging.

34:30 – 34:36
Raph Goldenberg: Yeah. And I guess we’ve never had to do that, right, because people have controlled it and they make their own decisions.

34:36 – 34:59
Professor Michael Milford: We haven’t had to do that specifically. But it is important to note that these trade offs are already made in safety decisions that affect thousands of lives. So crash barrier protections, speed limits, they all implicitly have tradeoffs in terms of value of life versus productivity, etc., baked into them. And they are reviewed occasionally. And that’s what should happen with autonomous vehicles.

35:00 – 35:33
Raph Goldenberg: You make it sound simple. Look at I think and I’m not going to ask you about how long it’s going to take before we see autonomous cars on the road. And I don’t think I can expect you to answer that question, but I do want to finish with just a question which touching on again, the legal side and this is a question without notice, but if you were the Prime minister of Australia for the day and you could change a law and it doesn’t have to be about autonomous vehicles, which one would you change?

35:34 – 35:57
Professor Michael Milford: I would make donations much more transparent and accountable. And this relates to big tech and autonomous vehicles. You want to make sure that all the parties that are lobbying for views that everything around finances and other support in the political system is as transparent and obvious to everyone. I think that would go a long way to making our society a lot fairer.

35:58 – 36:07
Raph Goldenberg: Excellent. I didn’t expect you to say that, but I think that’s a great way to finish off. Thank you, Michael. It’s been fascinating talking to you and thanks again.

36:07 – 36:23
Professor Michael Milford: Great. Thank you for having me.

36:29 – 36:47
Raph Goldenberg: Thank you for tuning into Legally Consumed, a consumer products podcast by the consumer products law firm. This was presented by Raph Goldenberg, Kaye Ho and Will McMinn from CIE Legal. Our theme song is by our very own CIE Legal House Band, which is comprised of Will McMinn, Andrew Maher and Andrew Thompson.

36:47 – 36:50
Kaye Ho: This podcast production is in partnership with Social Star.

36:50 – 37:27
Will McMinn: If you’ve enjoyed this episode, please make sure you follow us on Apple Podcasts, Google Play, Spotify, or wherever else you get your podcast from. For more updates and behind-the-scenes footage, head to our Instagram, @cielegal, all one word. You can also check out our website to listen to a range of episodes and find additional free information at www.cielegal.com.au. To get in contact with CIE legal about our consumer products and legal services, reach out to us at info@cielegal.com. Thanks again for listening to Legally Consumed and see you all in the next episode!