Podcast: Linking Australian bushfires to climate change, and Asimov's robot ethics

Host: Benjamin Thompson

Welcome back to the Nature Podcast. This week, establishing climate change’s role in the ongoing Australian bushfires…

Host: Nick Howe

And Isaac Asimov’s impact on robot ethics. I’m Nick Howe.

Host: Benjamin Thompson

And I’m Benjamin Thompson.

[Jingle]

Interviewer: Benjamin Thompson

Large parts of Australia have been ravaged by bushfires that have been raging across the country for months. These fires have been, for the want of a better word, devastating, and researchers are trying to work out what contribution climate change has played in them. One of these researchers is Sophie Lewis from the University of New South Wales, Canberra. Sophie lives in a Canberra, Australia’s capital city, located in the southeast of the country. For weeks at the back end of last year, the city was being choked by smoke from nearby fires, but over the holiday period, things escalated.

Interviewee: Sophie Lewis

So, on New Year’s Eve, really, really thick smoke started pouring into Canberra and settling over the city, so on New Year’s Day, it felt apocalyptic. There was a really thick orange haze. Visibility was terrible. There was very little going on in terms of people, animals. Usually, in my suburb, we have kangaroos hopping around happily, we have birdlife, and it was desolate. There was nothing here.

Interviewer: Benjamin Thompson

For many people in Australia, this has been a troubling time.

Interviewee: Sophie Lewis

It was scary. I was really scared and so was my partner, and we made a pretty rapid decision that one of us, at least, was going to get our young daughter out of the situation. So, I flew out to Hobart to spend some time with family there and my partner stayed in Canberra to ready our house for the catastrophic fire weather that was being forecast for the days after New Year’s.

Interviewer: Benjamin Thompson

Thankfully, Sophie’s home was safe.

Interviewee: Sophie Lewis

To date we haven’t lost any houses in Canberra. We have some really fantastic work going on here by our emergency service agencies, and they’re doing a great job of getting people to get their bushfire plans ready, providing information about where is safe to be. We’re very, very lucky that we haven’t been impacted by those fires. We know lots of people on the south coast of New South Wales around that time and they spent time in evacuation or relief centres. They faced really scary situations. So, we’re lucky in that respect, but it’s just been exhausting. Yesterday I spent the afternoon trying to work while listening to emergency service announcements and looking at what was going on with the wind and the smoke to see if we needed to start preparing to leave our home.

Interviewer: Benjamin Thompson

Australia experiences seasonal fires every summer and there have been huge ones in the past, but Sophie describes this season’s fires as unprecedented.

Interviewee: Sophie Lewis

We have had big bushfires before, in the past, but none that have really occurred on the scale we’re talking about. We’re talking about fires over huge areas and for many months. They haven’t gone out yet. There has been quite favourable weather in the last week or two, and our fire services have done a great job of using that period to try and contain some of these fires, but what we’ve experienced has never occurred before in what’s been observed for Australia.

Interviewer: Benjamin Thompson

The current and ongoing fire season began in September 2019 and has seen more than 10 million hectares of land burnt across the country. To try and put that into context, it’s an area over twice the size of Switzerland. So far, over 30 people are known to have lost their lives, thousands of homes have been lost, and some estimates suggest that a billion animals have been killed. At a global level, there are a number of process which have combined to contribute to the severity of this year’s fire season. Here’s Nicky Phillips, Nature’s Asia-Pacific Bureau Chief.

Interviewee: Nicky Phillips

There are very natural processes that drive Australia’s bushfire risk. We’re in the middle of a drought, and drought is usually a factor in how severe our bushfire seasons are. Another natural factor that we had against us this year was this thing called the Indian Ocean Dipole and that’s kind of basically a difference in sea surface temperatures in the ocean, and when it’s in its positive phase, it reduces the amount of rainfall over southern and northern Australia. And then as if we didn’t need it, this year also there’s a phenomenon above Antarctica where these polar winds were kind of affected and changed direction, which, in the end, resulted in more kind of hot and dry weather over Australia. So, we had all these natural processes going on, but on top of all that, we’ve also got climate change.

Interviewer: Benjamin Thompson

Last year was Australia’s driest and hottest on record and, understandably, people want to know what role climate change is playing in the fires. Linking extreme events to climate change is known as attribution science – something we’ve covered on the pod before. By modelling multiple factors, researchers are able to offer an idea of whether an event was made more likely or more serious by climate change. Take the extreme heatwave in western Europe last year – researchers suggested that this spell was made at least ten times more likely by human-induced climate change. Sophie, who you heard from earlier, works on attribution science. Last year, she published a paper linking 2018 fires in northeastern Australia to climate change. Unlike heatwaves, fires are difficult to perform attribution studies on, partly due to their complexity. Here’s Sophie again.

Interviewee: Sophie Lewis

We can have a really hot day in Australia, a scorcher, and we don’t have a bushfire because there’s so much more that goes into it. We need ignition, so that could be anything from a powerline failure, it could be a lightning strike, but we also need conditions to be primed in the environment for that fire so we need dry conditions, in terms of the fuel load and the fuel moisture content, we need winds to sustain the fire, we need, usually, low humidity, so there’s lots of different factors that go into that event, so it’s not as straightforward as looking at a heatwave, even though there’s also complexity in the weather that we experience with our heatwaves.

Interviewer: Benjamin Thompson

Sophie is now part of a group led by researchers in Europe who have turned their attention to the current bushfire season to examine the role of climate change on the bushfires. The team first need to define the timeframe and location they’re going to focus on, which is tricky given the scale of the fires. Then they’ll then run a huge number of simulations to determine the influence of climate change on individual factors, like temperature or humidity, and its effect on all of these factors combined. Nicky’s been looking into this group’s work for a feature in this week’s Nature and she explained their approach.

Interviewee: Nicky Phillips

The reason they’re doing it that way is because the researchers that I’ve been talking to are fairly confident that they’ll be able to see that climate change influenced the extreme temperatures because we know the globe is warming, but we don’t know whether climate change is influencing the other factors that go into extreme fire danger, like the wind and the humidity and that sort of thing, and so they’ve got to look at them separately and together to see whether they can get a picture of the contribution of climate change to an event like we’ve seen this year.

Interviewer: Benjamin Thompson

The team are hoping to publish their results soon, and Sophie hopes that attribution studies like this will help people plan for future events.

Interviewee: Sophie Lewis

If a study finds that climate change made a bushfire in northern Australia and Queensland more likely, that’s really important information for people who are making planning decisions about our future because places like that aren’t necessarily well set up for these fires. In southern Australia, we have very sophisticated responses in terms of emergency services and management. We have teams of volunteers who are trained and ready to go, but that’s not necessarily the case in places where fires are occurring where they didn’t before. So, the hope is that that information will then be used and available to anyone who’s at the frontline of those events.

Interviewer: Benjamin Thompson

Climate change discussions in Australia are fiercely politicised, and there’s a recent editorial in Nature calling for the country’s leaders to act in the face of overwhelming evidence and public opinion. I asked Nicky about the mood among the climate scientists she spoke to.

Interviewee: Nicky Phillips

At least the ones in Australia feel that our government isn’t doing enough to address the factors that are contributing to climate change. They did warn us. Some of the first studies saying that climate change would make fire seasons worse in Australia came out in 1981. In 2008, a government report said that the fire seasons would start earlier, end later and be more severe, so I think there is an element of ‘you’ve been warned’.

Interviewer: Benjamin Thompson

That was Nicky Phillips. You also heard from Sophie Lewis. You can read Nicky’s feature, which talks more about fire attribution studies and the situation in Australia, over at nature.com/news.

Host: Nick Howe

Later in the show, we’ll have an update on the Wuhan coronavirus story and hear how social scientists are struggling to cope with social media bots – both of those stories are in the News Chat. Now though, it’s time for the Research Highlights, read by Anna Nagle.

[Jingle]

Anna Nagle

If you’ve heard of Pompeii, it’s likely because of the violent eruption of Mount Vesuvius in AD 79 that destroyed the town. Today, tourists visit the site to see casts of the volcano’s victims frozen forever at the moment they were engulfed in volcanic debris. But you may not know that nearby Herculaneum was also hit by Vesuvius. Now, remains from this site have thrown some confusion into theories of how people perished. It’s long been thought that Vesuvius’ victims died instantly – their soft tissue vaporised by the heat. But a new analysis of bones in Herculaneum suggest that those who died suffocated and baked rather than vaporised. That may not be the end of the story though. Separately, brain tissue in the remains of a skull from Herculaneum had turned to glass, suggesting the victim had been subjected to extreme temperatures that would have vaporised human tissue. Vaporised or baked – precisely how Vesuvius takes it victims is clearly still a matter of debate. Find those papers in the New England Journal of Medicine and Antiquity.

[Jingle]

Anna Nagle

We’re about to travel back in time to ancient Egypt. Behold the voice of Nesyamun, scribe and priest at the temple of Karnak. Well, what were you expecting from a 3,000-year-old mummy? The sound you just heard was produced from a 3D-printed vocal tract and electronic larynx. Using CT scans, researchers were able to take precise measurements of the mummified priest’s vocal tract, which was particularly well preserved. The researchers say their work could have applications for how history is presented to the public, allowing us to hear the voices of those long dead. For Nesyamun, his desire was to have his voice heard in the afterlife, in order to live forever. Perhaps science has helped him achieve that. Uncover that research in Scientific Reports.

[Jingle]

Host: Nick Howe

Living in the year 2020 is great. Every morning, my robot butler brings me my replicated coffee and at work, my friendly android colleague and I make the weekly podcast. Well, okay, the year 2020 isn’t exactly how science-fiction writers of the past may have imagined. Artificial intelligence is everywhere. It’s more likely to be found in an algorithm sending me targeted advertising than in a helpful robot. But one thing that’s still important is the question of how machines make decisions, in particular, ethical decisions. One science-fiction writer who’s had a big impact on these kinds of ideas is Isaac Asimov. Now, Asimov’s personal ethics were dubious. He was well known during his lifetime for his unapologetic harassment of women. Despite this, his writing, in particular, stories imagining how robots might be designed to follow simple, ethical rules, continues to inspire debate. This month marks 100 years since Asimov was born, and Nature has published an essay on Asimov’s work by David Leslie, Ethics Fellow at the Alan Turing Institute. Reporter Shamini Bundell set out to talk to David about whether Asimov’s ideas about robotic ethics still apply, as artificial intelligence becomes more prevalent. She found him at the Institute, which is based in London, nestled in the centre of the British Library.

Interviewer: Shamini Bundell

Here we are in the British Library. I can see just huge stacks of antique books here already, and how many Asimov books do you reckon they’ve got in the library here?

Interviewee: David Leslie

Hundreds. I hope that it would be nearly 500 that exist, or around 500.

Interviewer: Shamini Bundell

As well as being an Asimov fan, David has a particular interest in Asimov’s vision of a future where humans live and work alongside robots.

Interviewee: David Leslie

What Asimov did was he took a world in which robots were portrayed as kind of alien monsters, right, and he tried to make the stories more realistic, where they’re exploring the possibilities that are opened up by robotics and what we now call artificial intelligence.

Interviewer: Shamini Bundell

And you’ve got some books on the table here, including I, Robot, which is a number of short stories. Could you give us some idea about the kind of things that came up in these stories that he discusses in these books that were quite new at the time and have had quite a lot of influence since then?

Interviewee: David Leslie

Well, I mean, I think we would have to first talk about the famous three laws of robotics, which actual arise in the I, Robot series, and the three laws are, basically: a robot must not injure a human being or allow, through inaction, a human being to come to harm. That’s the first law. The second law is: a robot must obey orders given by a human being unless that contradicts with the first law. And the third law is: a robot must protect their existence, unless that protection would come into conflict with the first two laws. So, the stories that he wrote in I, Robot and the other robot stories beyond that had to do with the ways in which these three laws play out in real world circumstances. For instance, I’m thinking of one in particular where he has a robot, I think it was Herbie, who is able to read minds and Herbie started to lie. So here it is, a very interesting passage:

“She [Calvin] faced them and spoke sarcastically. “Surely you know the fundamental First Law or Robotics.

The other two nodded together.

“Certainly,” said Bogert, irritably. “A robot may not injure a human being or, through inaction, allow him to come to harm.

“Nicely put,” sneered Calvin. “But what kind of harm?”

“Why – any kind.”

“Exactly! Any kind! But what about hurt feelings? What about deflation of one’s ego? What about the blasting of one’s hope? Is that injury?”

Lanning frowned, “What would a robot know about that” and then he caught himself with a gasp.

“You’ve caught on, have you? This robot reads minds. Do you suppose it doesn’t know everything about mental injury? Do you suppose that if asked a question, it wouldn’t give exactly that answer that one wants to hear?”

Anyway, that’s one of these great passages where the scientists are realising that one can’t programme a notion of injury or harm, simply, formally, into a computer because it requires interpretation.

Interviewer: Shamini Bundell

The three laws seem to be a way of trying to program in ethics – let’s solve all sort of moral conundrums about people tied to tram tracks with some simple rules. How useful are those rules in your discussions of AI ethics?

Interviewee: David Leslie

We have to remember, the three laws of robotics were a literary device for him, so I think that a lot of times in our kind of contemporary world, we’re seeking out a moral panacea for the problems that are raised by artificial intelligence, but for Asimov, he really intended the laws to be an occasion for reflection on the human impacts of technology. For me, it sort of casts a floodlight on the need to actually think about automated systems as automated, as following prescribed rules and the limitations of that. In other words, the system might have prescribed rules, but the system won’t be moral in the same way that humans are moral because humans have to interpret what things like ‘harm’ or ‘human’ or ‘humanity’ mean.

Interviewer: Shamini Bundell

And specifically, if you’re trying to program a machine to make a decision which may include ethical components, what are the kind of challenges that people are facing?

Interviewee: David Leslie

Well, I think first off, there’s the challenge of thinking about where the values are going to come from that are going to inform the programing or the behaviour of the instrument. The way that an automated system is designed derives from all of the values, all of the human choices of those who are involved in its design, production and implementation, and so we have a big set of dilemmas here about who’s making the technology. Are the makers of the technology representative of the world that the technology will impact?

Interviewer: Shamini Bundell

The kind of machines and robots that we have today, we don’t have sort of humanoid robots wandering around helping us with everyday tasks, but we do have things that we’re trying to think about like, self-driving cars is the one that always comes up. What kind of ethical problems are we faced right now?

Interviewee: David Leslie

I think we live in a world where we are increasingly subjected to the decisions of automated systems. We live in an increasingly prediction-oriented society where you’ve got a lot of large-scale algorithms that are anticipating or pre-empting bits of our behaviour. Just think about the various social media outlets that use curatorial algorithms and that world is not necessarily a world where the automated systems are our companions, and I think that would have horrified Asimov. I think that for him, when we live in a world that is algorithmically steered, we’ve lost that component of human agency and human freedom that he saw at the very core of what it is to be human and what it is to actually have and use technology.

Interviewer: Shamini Bundell

And what kind of a future do you see? Where do you see this going, in terms of robots being tools or being used to predict and control and influence?

Interviewee: David Leslie

So, one of the interesting problems that comes up across the stories is this notion of a Frankenstein complex, a kind of irrational gut feel, that in a sense these are just kind of monsters that are going to supplant humans and come to rule the world. The creatures come to take over the creators. And for Asimov, one of his bigger picture thoughts was that we need to overcome this kind of fantasy. For him, robots and robotics was just automation. They were tools. And I think what that means is we need to pay attention to what Asimov says and we need to think of the ways in which machines aren’t necessarily going to be monstrous agents of the future. Rather, think of them as allies, automated allies, that can help us as tools to build a better world together.

Host: Benjamin Thompson

That was David Leslie talking to Shamini Bundell at the Alan Turing Institute. You can find his essay on Asimov in Nature’s books and culture section at nature.com/books-culture.

Interviewer: Nick Howe

Last up on the show, it’s time for the News Chat. This week, I have the pleasure of being joined by two guests. I’ve got Heidi Ledford and Ewen Callaway, both senior reporters here at Nature. Hi guys.

Interviewee: Heidi Ledford

Hello.

Interviewee: Ewen Callaway

Hi there.

Interviewer: Nick Howe

Thanks for joining me. So, Ewen, you were actually with us last week, telling us all about what has come to be known as the Wuhan coronavirus, and this is a fast-changing story and when we talked about this last week, a lot of what we had said was already out of date. So, with that in mind, what more do we know about this virus?

Interviewee: Ewen Callaway

Yeah, and probably what I’m saying now is going to be out of date. I think last week there was still some question over whether cases were being driven by spread between animals and humans, and it’s become utterly clear that humans are spreading this virus between themselves. There are upwards of 4,000 cases that have been confirmed. There are probably thousands more in China alone. We’ve got cases in other Asian countries, cases in the US, in Germany and elsewhere, so things are changing really rapidly, to say the least, and this virus has legs.

Interviewer: Nick Howe

Do we have any idea how quickly the virus is spreading between people?

Interviewee: Ewen Callaway

That’s something that epidemiologists are trying to get a hand on. There’s this kind of figure that they talk about called R0, which is kind of an approximate measure of how many people an infected person tends to infect, and if it’s greater than one, you need to do something to control the outbreak. Less than one then infections should eventually fizzle out because you get fewer and fewer people passing it on in each successive generation. And so, people are trying to come up with rough estimates of this value. It’s 2, 3, 4 –something like that. These are very, very rough estimates because the epidemiology is uncertain, but what’s clear is this virus is spreading between people, and I guess one of the big unanswered questions is when does the virus spread between people, like there’s this period of incubation time that the virus has before it causes symptoms and there’s an open question, can you spread the virus during this incubation period? If you can, it’s quite worrying because you have people without symptoms spreading the virus so it’s hard to identify and isolate them. So, there’s really a lot of pressing questions about this and very little answers or, at least, very little public answers right now, I’m afraid.

Interviewer: Nick Howe

Is there anything more that can be done to try and control this or help this issue?

Interviewee: Ewen Callaway

I don’t know. I think that’s the question that epidemiologists around the world are asking, is what can be done to control this? If you have a situation where people without symptoms are spreading the virus for a long period of time, it’s going to be tricky to control. With SARS, one of the reasons that people were able to control SARS when they did was because there were few cases in which people didn’t have symptoms, so it was possible to isolate most people before they became transmissible. If you can’t do that here then it’s unclear what’s going to work. I’m sure public health officials are thinking very hard about all these different scenarios and I think only time will tell, really.

Interviewer: Nick Howe

Well, obviously, we’ll keep an eye on this and we’ll have updates on this in the future. But moving on to our second story, we all know that social media is a battleground, but social scientists have been battling bots. So, Heidi, in the first instance, what is a bot?

Interviewee: Heidi Ledford

A bot is essentially some code that is written to allow a social media account to look as though it’s being run by a person. We hear a lot about bots, I think, when we talk about, for example, the 2016 presidential election in the United States. There’s been some talk about bots in terms of the Brexit vote that year as well and how they could be spreading misinformation in an attempt to sway elections, or public health officials think a lot about bots these days as well because there are some bots out there that seem to be spreading misinformation about various health-related topics like vaccination, e-cigarettes and so forth.

Interviewer: Nick Howe

So, they’re able to be used for sort of nefarious purposes. What’s been the impact on social science?

Interviewee: Heidi Ledford

Yeah, so there’s a whole field of social science these days that really likes to go through and look at data from social media sites. It’s just a playground for them to learn more about how humans interact with one another and how information spreads from one source to another, how it influences behaviour. That list just goes on and on. But the problem is if you’re trying to mine these big data sets and your data is polluted by a bunch of bots that are pretended to be people but are not actually people then it doesn’t really tell you about human behaviour after all.

Interviewer: Nick Howe

And so, are social scientists able to screen for these bots in any way? Is there anything they can do to say, this is a bot and this is a person?

Interviewee: Heidi Ledford

Yeah, so there are programs out there. There are things like Botometer and BotSlayer, these different programs that are meant to weed through and look for certain characteristics of bot-like behaviour and then weed that out of your dataset if you’re a social scientist. There are some people who say that they are not being used often enough, either because awareness is not high enough or because it’s a bit of a new field and people are still kind of working their way around trying to find the data standards and the computational standards for how these sorts of datasets are handled. But the other issue, the sort of more conceptual issue, is that bots are getting more and more sophisticated. They started out being very simple and often just accounts that were trying to drive followers to other accounts and things like that, and now, they’ve become much more sophisticated. They behave more and more like people. Some of them even have human-generated content mixed into the bot-generated content to make it harder to detect, and so on. So, it’s an arms race, basically, between the detectors and the bot developers.

Interviewer: Nick Howe

Other than trying to detect the bots, if it is hard – they’re more human-like, as you say – is there any other strategies that social scientists can use?

Interviewee: Heidi Ledford

As far as cleaning their datasets, I think probably the best thing is to try to get the bot-generated data out of there as best they can. There are some social scientists who have used bots to do studies and some really interesting studies about racism and other aspects of human behaviour online. So, not all bots are nefarious. I subscribe on Twitter to an account that sends me a line from Moby Dick every day – I’m pretty sure that’s a bot, but it makes me very happy. Some social scientists have used these as tools to help their own research, but I did speak to at least one who said he has used them in the past but now that they’ve gotten such a bad reputation, he thinks he may not use them in the future because he doesn’t want people to get angry with him for using a bot.

Interviewer: Nick Howe

Well, I’ll be on the lookout for bots next time I’m on social media. Heidi, Ewen, thanks for joining me, and listeners, for more on those stories, head over to nature.com/news.

Host: Benjamin Thompson

That’s all we’ve got time for this week. If you’d like to follow a Twitter account that’s not run by bots, you can follow us @NaturePodcast. Also, if you’d like more coronavirus info, head over to our YouTube channel where we’ve got a quick explainer video. You can find that and all our other videos over at youtube.com/NatureVideoChannel. I’m Benjamin Thompson.

Host: Nick Howe

And I’m Nick Howe. Thanks for listening.