Malaria bites back, and the lunar base race

25 Jul 2025 · 38 min · 16 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

The episode covers three science stories: rising malaria risk in Africa and its knock-on effects, whether baby sex is truly random, and South Korea’s plan for a lunar base (plus a later segment on “ethical black boxes” for AI).

Guests

Jane Carlton, director of the Malaria Research Institute at Johns Hopkins; Jorge Chavarro, Harvard School of Public Health; Richard Hollingham of the Space Muffins podcast; Alan Winfield, Bristol Robotics Lab robot ethics researcher; Renita Bardom and Bill Sutherland from Cambridge (Question of the week).

Key claims

malaria cases are increasing due to US funding cuts (e.g., USAID/President’s Malaria Initiative) and insecticide resistance plus the spread of Anopheles stephensi from Asia into Africa; locally acquired malaria in the US could rise if infected travelers bring parasites home. Baby sex is “not random”: analysis of 60,000+ women suggests a weighted coin (about 60/40) influenced by maternal age and some genetic signals. Lunar base: South Korea aims for a base by 2045, likely using lava tubes and lunar-dust bricks; it’s uncertain how fast it will happen due to geopolitical “blocks.”

Notable examples

Zimbabwe’s tripling infections; locally acquired US cases in 2023 (Florida, Maryland); EBB simulations (robot dog distress, assisted-living robot failing to alert).

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Introduction to Malaria Issues

0:45 to 1:42

Overview of the current malaria situation and its causes.

“Without further ado, this is The Naked Scientist.”

Malaria Transmission and Resistance

1:42 to 2:18

Discussion on mosquito species and resistance issues affecting malaria control.

“bites of infected anopheline mosquitoes, mostly at night and mostly in sub-Saharan Africa.”

Impact of US Funding Cuts

2:18 to 4:30

Exploring how reductions in US funding affect malaria control in Africa.

“Another factor is that a new species of mosquito vector has stepped into the ring recently.”

Global Repercussions of Malaria

4:30 to 6:53

Analyzing the potential spread of malaria cases beyond Africa.

“what the future is going to hold it's very unfortunate the way that this was done as well It wasn't a decrease over time.”

Political Perspectives on Malaria Funding

6:53 to 8:00

Discussion on the political implications of malaria funding cuts.

“but of course infectious diseases and the vectors the mosquitoes that carry them don't know boundaries at all.”

Local Responses to Malaria Control

8:00 to 9:04

Strategies African countries are implementing to manage malaria internally.

“Well, I know many African countries are now looking internally to see whether they themselves can take over their procurement of bed nets and other control methods.”

Study on Baby Gender Chances

9:12 to 10:10

Exploring research on the factors influencing baby gender outcomes.

“For years, scientists assumed that it was and that the odds of having a boy or a girl were 50-50.”

Behavioral Factors Affecting Gender

10:10 to 11:34

Investigating how human behavior influences baby gender distribution.

“If sex is supposed to be determined by the sperm adversization, where you have either a Y chromosome or an X chromosome, you would expect that the distribution of sex would be essentially random.”

Maternal Age and Gender Outcomes

11:34 to 14:01

The impact of maternal age on the likelihood of having children of one gender.

“That is very useful because it allows you to look not only at the overall distribution overall, meaning, okay, how many boys, how many girls in general, but it can also get at behavioral factors.”

Exploring Genetic Signals in Child Gender

14:01 to 16:51

Learn about research findings on the correlation between parental age and the sex of children.

“And what we found was that women who started having children later in life, over 28 years, compared to having your first child before age 23.”
Show all 16 chapters

South Korea's Lunar Base Aspirations

17:43 to 17:55

Examine South Korea's plans for a lunar base and the implications for space exploration.

“Find out how Spitfire can empower your company at spitfire.co.uk.”

South Korea's Lunar Base Aspirations

17:59 to 22:30

Examine South Korea's plans for a lunar base and the implications for space exploration.

“Perfect music for audio and video productions.”

Ethical Black Boxes for AI and Robots

22:31 to 28:00

Discover the concept of ethical black boxes for understanding AI decision-making.

“Of course, the US used to collaborate and work with Russia.”

Exploring Ethical Black Boxes in Robotics

28:00 to 32:56

Learn about the importance of ethical black boxes in preventing robotic accidents.

“and because of this particular input of information here, we can see that it made that and that needs changing.”

AI's Role in Environmental Sustainability

32:56 to 38:22

Discover how AI is being leveraged to improve environmental outcomes.

“Now, sticking with AI, it's time for Question of the week and James Titko has been looking for green shoots thanks to this.”

Listener Questions and Show Wrap-Up

38:22 to 39:20

Engage with listener questions and prepare for the next episode's topic.

“The Naked Scientist comes to you from the University of Cambridge.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00Hello, I'm Monica Reinagle, host of the Nutrition Diva podcast. Summer is here and with it the hot and sticky weather. A morning working in the garden or maybe on the golf course can leave you more depleted than you realize. And plain water isn't always the most efficient way to restore the balance. That's where Liquid IV comes in. Its science-backed formula is clinically proven to hydrate faster than water alone. Whether you're exercising, playing outside, or simply surviving a summer heat wave, it's an easy way to stay hydrated. Shop now at liquidiv.com. Liquid IV, hydration that goes wherever life takes you.

0:40All engine running. Absolute genius. Get this. Welcome. Welcome. This is the show where we bring you science. What that essentially means is... Discovery is... Advances. Research. Technology. Unbelievable. Without further ado, this is The Naked Scientist. Hello, welcome to The Naked Scientist podcast. This is the programme that strives to bring you the biggest breakthroughs and talk to the major movers and shakers in the worlds of science, technology and medicine. I'm Chris Smith and coming up this week, malaria is on the move. Cases are up across Africa and a cut to US funding is to blame, some are saying.

1:15Also, while the sex of your baby is not potluck, it turns out, genetic factors can load the dice. And we ask whether artificial intelligence should be more accountable and how. From Cambridge University's Institute of Continuing Education, this is The Naked Scientists.

1:41Up first this week, malaria is a deadly disease that's caused by parasites spread through the bites of infected anopheline mosquitoes, mostly at night and mostly in sub-Saharan Africa. cases are both preventable and treatable but only with sustained funding for things like mosquito bed nets insecticides testing and treatment but the numbers are rising fast and Zimbabwe has seen for instance a tripling of infections just within the past year and other parts of the African continent are also showing a big uptick one contributing factor is funding cuts from the United States Washington has long been one of the biggest backers of global malaria programs that that support has been curtailed.

2:24Another factor is that a new species of mosquito vector has stepped into the ring recently. Jane Carlton is the director of the Malaria Research Institute at Johns Hopkins University. It's a species called Anopheles stephensi and it's usually found in cities and urban areas in Asia and it's moved from Asia into the African Rift Valley countries in particular Ethiopia. And this was first found a couple of years ago or so. In fact, we think the invasion happened perhaps even earlier than that. And so this is very worrisome because it could potentially lead to an increase in malaria cases. Do we know why those mosquitoes are able to do what they do?

3:07The Anopheles species of mosquito which transmit malaria are very prone to becoming resistant to the insecticides that are used either on bed nets, which is one of the main control methods in many African countries, but also what's called indoor residual spraying, which is spraying of insecticide either indoors or outdoors on particular structures such as houses as well. So it's the fact that the mosquito is becoming resistant very easily to these formulations of insecticide, which is very worrisome. Obviously, a lot of research that goes into this gets funded from the US. How has the withdrawal of funding or a lot of the funding been chopped recently?

3:52How has that impacted? Terribly. It's awful. So the USAID, which is the major program that the US funded, for example, in 2024, provided$270 million for Zimbabwe, just one country, for their health and agricultural programs. and that has pretty much been disbanded now and in particular the president's malaria initiative which distributed bed nets anti-malarial drugs methods for detecting the parasite rapid diagnostic tests and all of that has now stopped and so many of these African countries are very worried about what the future is going to hold it's very unfortunate the way that this was done as well It wasn't a decrease over time.

4:41It was just very much a sudden death of these programs. And now that the malaria season in many countries, many African countries, is sort of partway through or starting because of the rainy seasons, then we're really starting to see an increase in cases, in particular as in Zimbabwe, as you mentioned. Do you anticipate, though, that this won't just be a problem for, say, Africa? Because the thing about infectious diseases is that while they may have a hotbed in one geography, they don't have to stay there. And very often they will spread. And there are repercussions for other countries, including those that would regard themselves as not really at risk to start with, like America.

5:21Oh, yes, that's very true. And in fact, two years ago, I think it was in 2023, we had 10 cases of what's called locally acquired malaria in the United States, several in Florida, and one in my state, Maryland, too. And that was unprecedented. We've not had malaria, locally acquired malaria cases in the US for the past 20 years. And that is because we do have the right mosquitoes, Anopheles mosquitoes, to be able to transmit malaria parasites. And so US citizens, when they travel abroad, perhaps back to their original home countries or perhaps working or even, you know, our military, for example, will pick up those malaria parasites and then bring them back to the US.

6:10And if a mosquito then bites them, picks up the parasites, that can be then transmitted onto somebody who's never even left the country. So if the number of malaria cases starts to increase in African countries, then there's a possibility that the number of people bringing back those malaria parasites into the US will increase. And so that will potentially lead to more malaria cases. particularly with a more mobile population because not only are there more people on earth now they are traveling in unprecedented numbers i mean the figures that i've seen quite regularly popping up are something like a million and a half people airborne around the planet on aircraft at any moment in time i think globalization as we know it also includes increase in travel which actually i think in the long term is very much a good thing but of course infectious diseases and the vectors the mosquitoes that carry them don't know boundaries at all.

7:04So you're absolutely right. And what are politicians when you put this to them and say, look, this can be regarded as almost like homeland defence. It's almost like a security blanket. Chopping this off is punching a gaping hole in that sort of defence, not just for America, but for the entire world. So why are they persisting in this course of action? It's so short sighted. It really is. I mean, in the long run, it's just going to, I think, turn around and bite us on the backside. I don't understand it. I don't really want to comment upon why I think it's being done. I also know that the budget, the international aid budget that the US provides is really very small.

7:46And to target that and to remove that is incredibly short-sighted to me, I feel. So what can we do? I mean, we are where we are. America is where it is. What sorts of things can be done to mitigate the impact of this? Well, I know many African countries are now looking internally to see whether they themselves can take over their procurement of bed nets and other control methods. But many of these countries, their economies are pretty, you know, pushed to the limit already. I know Nigeria is another example that has increased its budget this year in order to be able to provide bed nets and drugs and other methods of malaria control.

8:33Apart from that, I'm not sure what else can be done, really. Many of us here in the United States are, you know, continuing our research programs as much as we can. Obviously, there are some issues with the National Institutes of Health and the funding of those international projects, but we're just really trying to push through and see what we can do over the next few years to continue to support the work on malaria control in endemic countries. So things are not altogether great at the moment, are they? Jane Carlton there at the Malaria Research Institute at Johns Hopkins University. Now, is the sex of your baby really just down to chance?

9:17For years, scientists assumed that it was and that the odds of having a boy or a girl were 50-50. But a new study suggests that some parents may be more likely to have a slew of children of the same sex. Now we've all come across those families with a long line of boys who, because they were trying for a girl, kept going and vice versa. But are these the exception or are they actually the rule? Well, after analysing data from over 60 ,000 women, researchers have found the odds are indeed not random. Instead, there appear to be two things going on. One, biology appears to delicately tip the balance slightly in favour of boys or girls for some mothers, but then human nature kicks in and people alter their family planning choices to compensate.

10:06Jorge Chavarro is at Harvard School of Public Health. If sex is supposed to be determined by the sperm adversization, where you have either a Y chromosome or an X chromosome, you would expect that the distribution of sex would be essentially random. But the fact that we see these families that have only boys or only girls so often make you wonder to what extent this phenomenon is random or whether or not there are factors within a couple or within a woman that make it more likely that a couple is going to be having either more boys or more girls from an evolutionary and biological point of view though it makes sense that we ought to have even stevens or thereabouts because we want to have a balanced representation of both sexes in a population ideally don't we correct and although there there are correct me if i'm wrong but there are slightly more boys conceived initially but at all stages of the life course the mortality of males is slightly higher than femur so it does end up at roughly even Stevens when you look at a population level.

11:14Isn't that right? That is correct. You are at approximately 50%. So is the balance achieved then because the vast majority of people do have roughly equal numbers of boys and girls and there are some exceptions who have only one or only the other or is it that people have loads of one sex and other people have loads of another sex and that balances things out? Right. So we use data from two very large studies in the United States, the Nurses Health Study and the Nurses Health Study 2, where we had information not only of whether or not women had been pregnant throughout their life course, but also we knew for each of their pregnancies, whether each pregnancy had been a boy or a girl.

11:54That is very useful because it allows you to look not only at the overall distribution overall, meaning, okay, how many boys, how many girls in general, but it can also get at behavioral factors. So, for example, is it the same to have boy, boy, girl than boy, girl, boy? At the end, you get the same number of males and females, but those two patterns may lead you to have a very different behavior on, for example, trying to decide a fourth child. Or like more obvious, if you have a boy and a boy, are you more or less likely to decide to have a third one to try to balance the family and have a girl?

12:34Or are you more likely to stop? And if things are essentially random, you should have approximately equal numbers of boy-boy, girl-boy, boy-girl, and girl-boy. Those four numbers should be approximately identical. And what we see is that they're not. right and and when we look at different family sizes a family of two a family of three a family of four we see that there's a very clear pattern where people are stopping when they have children of at least two sexes so that's one clue that there may be a strong behavioral component to the sex distribution right so something is potentially biasing things and then human factors kick in and attempt to reset the bias is what you're saying these numbers show us that is exactly what these numbers are showing what do you think the mechanism is then that's driving this obviously there's there's a behavioral aspect to this but as you said there's something loading the coin so there's a bias which is there biologically so what is that how does it work and how can you get at that correct so these data suggest there was definitely a behavioral component but we wanted to understand if there was something other than behavior, which led us to do several types of analysis.

13:54And really the only signal that was quite robust, regardless of what analysis we were doing, was maternal age. And what we found was that women who started having children later in life, over 28 years, compared to having your first child before age 23. But even that small gap was associated with having only children of one sex and the signal was stronger for having only girls. So one possible interpretation is if you start later, you run out of time for having children faster, right? So that's possible. It's possible that that may be what age is capturing, but it is also possible that age is capturing aging-related factors that we do not understand yet.

14:41Other ways that we can get at genetic traits, you can comb through the entire genome looking for markers that crop up more often in groups that do have a trait compared to groups that don't. And it doesn't tell you what the gene is, it tells you where in the genetic code something might be influencing the equation. Have you used a technique like that and looked at this? That is actually one of the things that we did in this case. We did something called a genome-wide association study, where you do exactly what you guys described. We compared over 2 million genomic markers across the entire genome, comparing women who had had either only children of one sex or only boys or only girls to women who had had children of both sexes at different family sizes.

15:29And we did identify that there are a few genetic signals that are associated with having either only boys or only girls. But the few genetic hits that we identified have not been described before as having something to do with like embryo development or with ovulation or OSI development. So it is difficult to assign a specific biological function to the genetic signals that we're identifying. Does this tell us then that couples that are falling victim to gambler's fallacy, where they keep going because it's been a load of boys and the next one's going to be a girl, they're probably victims of that and they should probably quip while they're ahead?

16:19so we actually provide estimates of that in our paper the bottom line is it is not random it is not a coincidence that there are families that happen to have only boys or that happen to have only girls it's not like if you've had two boys you're all but guaranteed to have a third boy but it is definitely the case that the probability is not 50 50 it's more like 60 40 so you should know that you are probably taking your chances with a weighted coin as opposed to with a completely unbiased coin. Jorge Javaro from Harvard School of Public Health there and that study has just come out in the journal Science.

16:57You might want to bear that in mind if you're planning or in the process of having a family.

17:14Select interior paint via rebate. Plus, get two 2.5-quart or 3-quart mums for just$8. A fresh look and savings? Even better. Labor Day deals are on now at Lowe's. In-store and online. Valid through 9-9. While supplies last. Mum sizes vary by location and exclude Alaska and Hawaii. See store at lowes.com for more details.

17:37The Naked Scientist podcast is produced in association with Spitfire. Cost-effective voice, internet and IP engineering services for UK businesses. Find out how Spitfire can empower your company at spitfire.co.uk. Music in the programme is sponsored by Epidemic Sound. Perfect music for audio and video productions. This is the Naked Scientist podcast with me, Chris Smith. Still to come, should AI systems be fitted with black boxes to document their decision making? That's coming up. But first, South Korea has said it wants to establish a lunar base by 2045. The proposal, which is part of a wider roadmap unveiled by Korea Aerospace Administration, will include missions in low Earth orbit, microgravity research and solar science.

18:28With renewed lunar ambitions from the US and China and others, it appears that the new space race is accelerating. But is this a realistic prospect for South Korea? Here's Richard Hollingham from the Space Muffins podcast. South Korea has got a track record in space. In fact, it has a satellite around the moon right now. And it's had one astronaut, but only one astronaut. It doesn't actually have a human space program at the moment. Of course, the US has its plan to go to the moon, but no sort of hard plans to build a moon base yet. There's Russia. I think the most serious contender is probably China.

19:07and then the other nations like India, the Arab nations, and remember they have a huge amount of money, are also keen on exploiting the moon, exploring the moon. So I think this will happen. It's just that I don't think it's going to happen immediately. What is the draw though? What is it about the moon that they are potentially seeking to exploit? I think as much as anything it's a political ambition to go to the moon. It's a political ambition to put the flag there to explore beyond the Earth and to exploit beyond the Earth. It's also a stepping stone. So if we imagine humanity leaving the Earth, heading onto the Moon, then maybe exploiting asteroids.

19:52Now there is money to be made potentially with rare earth minerals and just metal in asteroids. And then the next stepping stone onto Mars. If you can't live on the moon, there's no way we can sensibly look to go to Mars and move on beyond the Earth. New golden age of exploration of, use the word hesitantly, colonisation, and just moving out beyond the Earth into the wider solar system. Have we got the first clue, though, how to actually build a viable, sustainable base on another celestial body like the moon? Sort of, yes. One thing that's been talked about, but we don't know these exist or we don't have exact proof they exist, we haven't seen them, is they're probably lava tubes on the moon.

20:38So we could live in lava tubes while living deep underground on the moon. The European Space Agency actually has done a lot of work on moon bricks, of sintering bricks, essentially making bricks out of lunar dust, very much like you'd make a brick on the earth and using the power of the sun on the moon focusing that and i've actually been to the facility in germany and it does look like a sort of bond villain lair with this giant mirror focusing a beam of light i think it's man with the golden gun actually very similar to that focusing a beam of light so really intense heat and just blending together, sintering they call it, moon dust, into these quite ugly bricks.

21:21And then the idea is you sort of join them together. You have kind of brick layers on the moon, probably robotic brick layers on the moon, that join all these up, perhaps over some sort of airtight inflatable. So an airtight inflatable they live underneath, and then these sort of bricks around the outside, and I mean maybe a window so you can see out and see the view of Earth. So there has been a lot done on this, and it would make sense and it's what they say about you know either living on the moon or living on mars is you've got to live off the land you know otherwise you've got to carry everything from the earth that's going to cost a huge amount of money to take something to the moon whereas if you can live off the land if you can use lunar dust you can use water on the moon to drink and maybe provide oxygen and then recycle as well so you know any bits of stray spacecraft are kicking around you you rip them up you melt them down you make something else out of them so you know that that that is the potential so we sort of do know how to do it i think we could but it's going to take a lot of will and an awful lot of money well on that point most space activity is collaborative because it involves a lot of know-how it involves a lot of risk and people like to share that risk both in terms of sharing the brain power but sharing the cost and the technological demands so is south korea just throwing its cap in the ring and saying we'll team up with people or are they saying we want to go this alone this is our single-minded single country ambition i wonder if it is sort of saying well look we want to be in this as well and they will join with one of these blocks that's emerging so at the moment we've got the artemis block which is NASA's funded project to return to the moon with astronauts.

23:09Europe is part of that. Of course, the US used to collaborate and work with Russia. I mean, it still does on the space station. That's not going to continue with the moon. So you've got China as this other block. And then, you know, where does South Korea go? And you could imagine a third block emerging where you had South Korea, India, African nations and maybe Arab nations with all the money as this third block also going to the moon or maybe South Korea will partner with China or with Artemis so it's really interesting these geopolitical blocks on earth joining together to work together in space.

23:49And on that when they get there who owns the moon do we do we have a sort of legal framework for this or is this literally a race and whoever plants their flag that's their bit yeah it's it's a really interesting question actually a lot of lawyers working on that um when you look at the various agreements there are lunar agreements there's the artemis accords and you look at who who the signatories are and it's always all the signatures apart from the key people are actually doing the things so you know the artemis accords for example which is you know peaceful cooperation in space the way we would operate on the moon, that's not been signed by Russia or China.

24:28So potentially, they're all aiming for the same place. So the Shackleton crater on the south pole of the moon, you could actually imagine a US base with maybe, you know, some Japanese astronauts there, European astronauts there, Canadian astronauts, and then the other side of the crater, or, you know, a few hundred meters away, a Chinese base with Russians and neither talking to each other. I I mean, that's exactly what's going to emerge at the moment. It's an exciting era, though, isn't it? But will it happen in our lifetimes? Watch this space. The Space Puffins, Richard Hollingham there, and you can catch more from Richard at nakedscientist.com forward slash astronomy.

25:09Black box flight recorders are usually associated, in most people's minds at least, with air crashes. But one expert believes something similar could help us to understand what goes wrong when robots malfunction. This is the brainchild of Alan Winfield, who's been developing what he calls an ethical black box, which can log a robot's decisions and actions in real time. As we move into an era where artificial intelligence is increasingly calling the shots, knowing how systems arrive at their decisions will become mission critical. Will Tingle. Welcome to the Naked Scientist Airlines. This is your captain speaking.

25:47We are currently cruising at an altitude of the... Since the 1960s, planes just like this one have been fitted with a black box, a data recorder that takes in flight data like altitude, aircraft speed and direction. In the event of the unthinkable, this data can be salvaged and used to ascertain which parts of the plane or the crew were at fault. Data recorders are now found in machines across the world, from the air to the land. But our machines and robots are getting smarter, so to speak. More of the impetus is being put on the machine to make decisions. decisions which could be life or death but the way that a lot of ai makes decisions is still not well understood some creators of large language models don't know what information their ai pulled to form an answer so going forward will there be a need for an ethical black box a data recorder that looks at the data fed into a smart robot to see what informed the decision a way of getting the relevant decision tree corresponding to the thing you want to interrogate out of the algorithm.

26:45I went to see Professor of Robot Ethics Alan Winfield. Alan has spent decades conducting research on robots and ethics at Bristol Robotics Lab. His work is funded by UK Research and Innovation and I began by asking what kind of information would be fed into an EBB. It's actually conceptually rather simple because it could be a software module unlike a flight data recorder which of course has to be physically very robust. most social robots don't need that so it's essentially a software module. If we took a smart robot in a hospital what kind of information data would you be anticipating is being fed into that black box?

27:25It's essentially everything that we possibly can capture of where the robot is in the world what its movements have been and are what decisions it's making and the interactions that it's having with the human beings around it. The whole point of this ethical black box is therefore to understand why the robot made its decision. Is the point of that then to create almost an accountability to be able to say that this robot made a decision that we do not agree with, that perhaps costed a life or caused an infirmament that was avoidable, and because of this particular input of information here, we can see that it made that and that needs changing.

28:08Yes, that's right. I mean, essentially, the reason for accident investigation, and this is true for aircraft and vehicles and so on, is essentially to ask three questions. What went wrong? Why did it go wrong? And how can we fix the problem to make sure it doesn't go wrong again? So the ethical black box for robots and AIs has exactly the same motivation. It's about the reduction of the likelihood of accidents and the mitigation of those accidents too. So are these ethical black boxes currently being made? A few, yes. So obviously we've designed and built several EBBs in the lab and at least one if not two companies are also building EBBs for their own robots.

29:00And in fact one company, a very highly regarded company called PAL Robotics and they are certainly developing an ethical black box for their robots. To put these EBBs to task, scientists at the robotics lab devised a series of tests to see if data stored in an EBB could inform a fault in a smart robot. If a child was suffering from signs of psychological distress, could it be put down to a smart robot dog that suddenly became aggressive? If an elderly woman was found collapsed, could it be ascertained if her assisted living robot didn't raise the alarm? Could an EBB shed light on why a movement exoskeleton might suddenly stop functioning?

29:42These were real simulations played out by actors, but crucially to keep things unbiased, the investigators had no idea what had actually happened. So were these black boxes able to help suss out what was responsible? Yes, they were. And in fact, it was very clear that they couldn't have actually figured out what was going wrong without the EBB. They would have a fair idea, they would have good guesses, but there would be no way of verifying those, if you like, educated guesses. so the ebb turned out to be absolutely essential we realized very early in the project that the ebb can be used as an explainer not just as a recorder an explainer that can be used more or less in real time so that you could ask the robot robot why did you just do that if for instance the user of the robot is puzzled the robot should be able to explain why it did that thing because it has an ebb and the second thing which is a little bit more tricky is you could say to the robot what would you do if in other words robot what would you do if i fell on the floor or robot what would you do if i forgot to take my medicine these are fantastically important and powerful extensions of the ebb and essentially my colleague has this wonderful project called pacer in the lab which is developing those explainer functions.

Read the full transcript

31:17And this seems particularly pertinent going forward because obviously those simulations that you ran were with hard physical robotics, but AI is taking over the world. There's no real other way to put it. And stuff like large language models, chat GPT, it is still not entirely certain where they're pulling their information from. Do you think going forward that this is essentially a necessity for us to be able to ethically use stuff like this, which has such a large net that it casts that we don't even know where it's coming from. No, I completely agree. I mean, if we are really serious about using AI for consequential applications, what I mean by that is applications on which lives depend.

31:59So a very good example would be diagnostics, medical diagnostics or clinical diagnostic AI. It's incredibly important to the clinician that the diagnostics tool should be able to explain why it reached that particular decision. Similarly, you know, if your welfare payments suddenly get stopped and it was an AI that made that decision, then the AI should be able to explain why. What was the rationale? So, you know, these are really important things. there needs to be transparency of decision making. And the only way you can do that is to have that process, the process that led up to that decision being recorded.

32:48Will Tingle in conversation with Alan Winfield, and that work has been made possible with the support of UK Research and Innovation. Now, sticking with AI, it's time for Question of the week and James Titko has been looking for green shoots thanks to this. I'm Geoff Frid in Toronto, Canada. Are there any ways now or anticipated in which AI will cause or encourage actual reductions in greenhouse gas emissions and or lessening of environmental degradation? or is AI all negative on the environmental front? Thank you. Thanks for sending that in, Jeff. You make a salient point about the environmental impact of AI.

33:35While AI is undoubtedly improving medical science and our productivity, the infrastructure behind it consumes vast resources, especially energy, but also water and raw materials, such as rare earths. There's no getting away from the fact we need to find solutions to decarbonise this booming industry. But following my conversations with some scientists here at Cambridge, I found there is room for optimism as to what AI can do to benefit the environment as opposed to harm it. First, I spoke with Renita Bardom, Professor of Sustainable Built Environment and Health. She's developed an algorithm to reveal the heat loss from almost every home in the UK.

34:18Trained on satellite images and previous energy audits, the tool is helping identify the homes which could be prioritised to improve their performance. One of our main motive for the project was to look at how can we decarbonise our built environment, which is one of the energy guzzlers in the biggest sense. But this is not a very easy task. You typically ask the person, how much do you think will your house cost to completely decarbonise? people will have absolutely no idea. And it's not just people, but also people who make homes may not have a correct estimation. That's primarily because we do not have a very correct estimation of total amount of heat loss each building is experiencing during the winter months when we are cranking up our heaters to keep our homes warmer.

35:06And so this whole estimation to do it at a large scale faster with more accuracy and more information that actually feeds directly into policy. Only AI methods can help us. It removes the tediousness, it removes the resource requirement that you would actually require if you have to go and do energy audits for every building in the UK. Doing this the old-fashioned way, where a professional auditor comes around to inspect properties, awarding them an energy performance certificate based on features like age of the property and the type of insulation in place, would take thousands of assessors, thousands of hours.

35:45I'll give you a real-time picture. If you want to get your house audited to get an EPC rating, typically it'll take you three days, even taking out the days you would take to plan it and book a rating and all of that. Here we can do it in a couple of minutes. We have generated an interactive map which is accessible through our website and you can actually explore if a particular postcode is more liable to lose more heat than the neighbouring one. And that helps not just policymaking, but also as an individual homeowner, you could make more informed decisions. Renita barred on there. And that's not all.

36:26Aside from attempts to tackle energy concerns, AI is also putting the turbo boosters behind other environmentally conscious initiatives. I next spoke with Professor Bill Sutherland from Cambridge's Department of Zoology. He's the brains behind the online tool Conservation Evidence. Over the past 20 years, an international team have combed through over a million research papers looking into the effectiveness of actions to improve the prospects of hundreds of species. The result is a readily understandable ecological encyclopedia. But now Bill and the team have been leveraging AI to significantly broaden the pool of relevant research.

37:09The interesting thing is that we've repeated a review we did. We did a review of all the possible measures for looking after the lepidoptera, that's the butterflies and moths. And we repeated that using a large language model for the same set of journals. And it found 97 % of the papers, which is very impressive. But the really impressive thing is it found other papers as well, quite a large number of other papers. And whenever people do us what's called a systematic review, this is a standard approach in medicine, the standard approach in everything is that you look at the title. And if the title looks promising, you then look at the abstract, and then you decide whether or not to include it.

37:51And there's actually a lot of research papers where there is a test of an action, but it's not mentioned in the title or the abstract. So there's a lot of literature that's missed that large language models find. Broadening the scope of studies enhances the evidence base for species-saving interventions. Thanks to Bill Sutherland. so jeff while environmental concerns around ai's energy appetite absolutely need to be addressed i hope that's given you an idea as to how scientists are already finding a whole host of applications for using deep learning to look after the planet next time on question of the week hi this is john bondy from vermont in the u.s recently someone said that there are 400 cell types in the human body how are these cell types different at the genetic level thank you and if you know the answer why not jump onto our forum you can have your say there at nakedscientist.com forward slash forum or you can email me and the same goes for if you have any questions of your own chris at thenakedscientist.com we look forward to hearing from you that's all we've got time for this week join us on tuesday when we're going to be probing the changing face of warfare what's science bringing to the battlefield in the 21st century, we're wondering.

39:14The Naked Scientist comes to you from the University of Cambridge. It's supported by Rolls-Royce. I'm Chris Smith. Thanks for listening. And until next time, goodbye.

40:06We'll be right back. details. Soil Offer excludes Alaskan Hawaii.

From the publisher
In the news podcast, what's causing the uptick in malaria cases in Africa? Also, scientists show statistically that the sex of a baby at birth is not random, and South Korea joins the throng in the race for settling on the Moon. Then, we hear how computer scientists are programming ethical AI to explain its decision making, and, sticking with AI, what are some of the environmentally friendly projects seeking to offset machine learning's vast energy consumption? Like this podcast? Please help us by supporting the Naked Scientists

More from The Naked Scientists Podcast

All 127 episodes
Malaria bites back, and the lunar base raceThe Naked Scientists Podcast · 38 min
Listen in VO