Subscribe
Share
Share
Embed
Following the successful first and second series of Unlocking the SDGs – A Blueprint for the Future, Professor Monica Lakhanpaul and Professor Priti Parikh are back with a deep dive into the UN SDGs. Over five episodes, the series considers issues including the role of AI and education in the SDGs and what other countries are doing to achieve the goals. Listen as academics from across UCL’s faculties and beyond bring new perspectives and understanding to this complex global issue.
00:00:12 Speaker 1
Welcome back to the third series of unlocking the SDK's a blueprint for the future. In this podcast, we explore the UN Sustainable Development Goals, or SDG's, and what they mean for society. I'm Professor Priti Parikh, professor of infrastructure engineering and international development at the Bartlett UCL School of Sustainable Construction.
00:00:34 Speaker 2
And I'm professor Monika Lakhanpal, professor of integrated community child health in the UCL Great Ormond St Institute for Child Health. In this episode, we're taking a look at the role of technology and artificial intelligence, or some will call it AI, in supporting the implementation of the goal as we know, AI is a rapidly evolving field, but really, what is its role in sustainable development? How can it really enable progress to a fairer world? Or will it prove more of a hindrance?
00:01:02 Speaker 2
What are the practical and ethical issues that researchers and policymakers need to take into account?
00:01:13 Speaker 1
Our guests today are Dr Maria Perez Ortiz from the UCL Department of Computer Science, and Professor Jack Stilgoe from the UCL Department of Science and Technology Studies. A warm welcome to both of.
00:01:25 Speaker 3
You thanks for having us.
00:01:27 Speaker 1
Thank you, doctor. Parachutists, thank you for joining us today.
00:01:31 Speaker 1
You've been involved with this issue for nearly a decade now. Can you tell us a little bit more about your?
00:01:36 Speaker 4
Background, of course, and thank you very much for the kind invitation to this podcast. I'm a computer scientist by training, but I've always really enjoyed working on projects with social and environmental impact. So during the last 13 years, I moved between domains from AI and biomedicine.
00:01:50 Speaker 4
Climate change, sustainable agriculture, neuroscience, education and more.
00:01:55 Speaker 4
In all of this, I experience the opportunities of AI, but also importantly the risks and the trade-offs and some of these left a big mark on me, especially the risk for which I realised that my background in computer science never actually prepared me for it. So it was never part of the process of technological design at the time to think of the risks of technology, something that I've always wanted to change.
00:02:16 Speaker 4
So all of these started to crystallise in the last few years in the creation of an MSC programme here at UCL on AI for sustainable development.
00:02:24 Speaker 4
Allowing students not only to learn about AI, but also the impact that these technologies are having on the SDG's. This is because when you think of the impact that these technologies are having and the future that we actually want to build, you realise that there's a huge.
00:02:38 Speaker 4
Disconnect. We are not building technology for the future that we want. So part of the programme is about incorporating this in the design of the technology, teaching students about AI and sustainable development so that they can innovate in this space, but also, very importantly, innovating responsibly, which is something that I think we'll.
00:02:56 Speaker 4
About today and this is because technology can both enable progress towards sustainable development but also inhibit it, so the programme try to enhance critical thinking well, to give engineers of the future more tools to mitigate any unintended risks that might.
00:03:11 Speaker 1
Arise and it's great that you use the term responsible AI and risk because it's important to kind of acknowledge.
00:03:18 Speaker 1
The future risks that AI can pose to our cities to achieving sustainable development goals.
00:03:25 Speaker 2
And just to take this forward, really because in our last series we spoke with Doctor Jessica Asby. I don't know if you know her from the University of Bristol.
00:03:32 Speaker 2
She spoke to us about the challenge of reaching consensus when developing the goals, and I imagine that similar to your experience, you're now in academia looking at how AI can enable overseas development. Maria, can you give us some examples of how this technology can help to address the SDK's like climate change and also a little bit of an add-on question you talked.
00:03:53 Speaker 2
About students and, you know, teaching students about I I was never taught about AI as a student. I'm just wondering, do you think I should become part of every curriculum across the whole of?
00:04:04 Speaker 4
UCLA, those are great questions. Let me start with the with the 1st.
00:04:09 Speaker 4
So we have now many examples of how AI can help us do more with less. So here I'm thinking about the carbonization, the materialisation and so on, but also to help us understand the effects of climate change, the most common use case of of a high in climate change, which we are seeing around the world, is already for climate change adaptation.
00:04:29 Speaker 4
For example, building early warning systems for.
00:04:33 Speaker 4
But I'm more fascinated by climate change mitigation because I think those examples are something that we don't talk about a lot. We usually talk about the carbon footprint of AI, but I think there's a lot of opportunities as well in terms of reducing carbon footprint. So AI, for example, can help the integration of renewables into the energy grid. And this is very important because usually we need to keep gas.
00:04:54 Speaker 4
Reserves running just because both renewable energies are very variable, so it might allow us, for example, to adopt A more complex form of the grids where houses not only consume but also produce.
00:05:06 Speaker 4
Energy. But also there's many cases in which, for example, we can use the so-called digital twins, which is something that is used now a lot as a term 2. As I said, get more with less. So there's a lot of computational methods that we can use to get for example 2 or 3% more energy out of a wind farm.
00:05:28 Speaker 4
Or we can use those methods to optimise much more. How a cement or a steel?
00:05:33 Speaker 4
And manufacturing plant works so that we can get around a 10 to 20% decrease in carbon footprint of these industries. So these solutions are very easy to implement. The question is are they being implemented? And the answer is generally no. Whereas AI is currently used a lot, for example for targeted advertising.
00:05:54 Speaker 4
Which increases consumption and obviously many of the models that are being used, like large language models, have a huge carbon footprint. So while we have opportunities, they are not a reality yet. And I think as you very well said, it's going to be key to bring these topics to every single student just because.
00:06:14 Speaker 4
Not only is going to be useful in the usage of these technologies, but it's also going to help a lot to build these technologies much more responsibly, so we cannot have just AI engineers building these these models. We need people in philosophy, we need social scientists, we need economies.
00:06:34 Speaker 4
Thinking about the impact that this can have also.
00:06:37 Speaker 4
In the job market and many other topics that are really so important. So I do think that we need to include AI in universities.
00:06:46 Speaker 2
So maybe UCL, being an interdisciplinary university, this is the time that we bring that to the fore and ensure one of our action points from this podcast could be checking if AI is in.
00:06:57 Speaker 2
Every course that we teach at UCL.
00:06:58 Speaker 4
I hope so. In the same way that we need to include sustainability as.
00:07:02 Speaker 1
Well, and you see, it's not only interdisciplinary, it's also global university.
00:07:07 Speaker 1
So if we think about international partnerships, doing good internationally and I work in International Development, I can see the huge potential of AI and how that can improve infrastructure. You mentioned digital twins and how that can really improve efficiency of infrastructure buildings that we may use.
00:07:27 Speaker 1
But.
00:07:28 Speaker 1
Could you unpack for me a little bit more what's the role of academic institutions and researchers in integrating AI into international aid work and how can we strengthen our partnerships with governments and NGO's and use research to inform policy makers?
00:07:42 Speaker 4
So in my opinion, we need to understand first where we've got with AI so far to understand what's left to do for it to have an impact in International Development. So I think we have enough methodological bases in AI now for building many of these prototypes. So I think these models are powerful enough.
00:08:02 Speaker 4
So we don't really need to keep working on making them even more powerful.
00:08:06 Speaker 4
Another challenge is really how do we make them safe and how do we actually use them around the world for solving some of these challenges that humanity is facing. So the incentives in this case are already starting to emerge. I think universities are now being ranked according to their impact on sustainability and their research and funding bodies are starting to create more opportunities in.
00:08:28 Speaker 4
Applied AI work rather than just foundational work, and this is really important for International Development.
00:08:35 Speaker 4
Also, specific calls for the use of this in lmis, so now it's even mandatory that many of these big AI conferences in their papers have a section on risks and ethical dilemmas. I think all of this is helping a lot already for this progress because I don't think we need to put the weight of this work on individual researchers.
00:08:58 Speaker 4
But rather, I think we need to change a bit the research system so that it allows us to to support this work on AI for International Development.
00:09:07 Speaker 1
And ethics of AI is really key, especially if you think about internal.
00:09:12 Speaker 1
Regional work where we engage with marginalised or vulnerable communities. The way we manage to that data, use that data. I think ethics is a key point that you've raised here.
00:09:22 Speaker 2
So building on that ethics, we have Professor Jack still go with us here today. So turning to you, Jack, you've spoken in the past about the ethical considerations of AI.
00:09:33 Speaker 2
And the importance of public trust in technology and innovation? Somebody like me is always sceptical about AI and and really doesn't understand it completely and is really interested in the.
00:09:44 Speaker 2
Unintentional consequences of it, but also the positives it can bring to us as well, so I struggle sometimes with with the concept of AI. So what from your perspective, are some of the key considerations when we're thinking about the integration of AI into sustainable development?
00:09:59 Speaker 3
Well, I think the sort of struggle that you describe is an entirely natural response and I feel it too.
00:10:05 Speaker 3
You and if you talk to the people involved in developing AI, they would feel it as well. I mean, we are at this weird stage where almost everybody has heard the phrase and knows that it's being talked about, but we as a society are sort of struggling to to make sense of it. There's a huge amount of hype surrounding the.
00:10:26 Speaker 3
The technology I mean, I'm I'm a social scientist, so I I look at these things as social phenomena and I'm really interested in how scientists and innovators.
00:10:36 Speaker 3
Talk about the potential benefits and potential risks because ultimately, you know, governments will have to make decisions about these things and they need to make decisions based on the the the quality of information that they're getting from uh from experts. But we're at this stage now where almost everybody has an opinion about AI, but it's still very hard to.
00:10:57 Speaker 3
To make sense of it, we're in this situation. I think Maria described it really well.
00:11:02 Speaker 3
You know the the technology, we recognise that it's sort of powerful, but it's still in most cases lacking purpose that we don't know what it's for. In many cases it might be a sort of you know a solution in search of a problem and an answer in search of a of a of a question that becomes really important when we think about.
00:11:22 Speaker 3
Something like the Sustainable Development Goals, because ultimately we're trying to connect technological potential.
00:11:29 Speaker 3
Social to societal needs. And as Maria pointed out, there is that sense of a of a sort of disconnect. One of the things I'm working on at the moment is a big UK research programme that's that's called responsible AI that tries to bring together researchers across a range of disciplines.
00:11:49 Speaker 3
There might be AI, there might be roboticists, there might be.
00:11:53 Speaker 3
Ethicists. They might be social scientists like myself, economists, historians, to help society make sense of of this, because ultimately it's really hard to answer the question of what AI can do for us if we are surrounded by this extraordinary hype about the technology and lots of talk about.
00:12:13 Speaker 3
You know, is it or isn't? Isn't it intelligent, you know, will it or won't it take your job? And this very sort of speculative conversation when ultimately, you know, if it's going to make a difference to the problems that the that the world faces, we really need to.
00:12:28 Speaker 3
Of a a more detailed conversation about about what it can and can't do.
00:12:32 Speaker 2
And do you think again just because my naivety around AI, you know, it seems to be a word that's bandied around everywhere now. Like you said, you know the solution to everything type thing, but do you actually think it could actually increase inequalities around the world?
00:12:45 Speaker 3
I mean my starting point as somebody who'd you know, I mean my my research and teaching at at UCLA is about the sociology of technology. And my starting point, I'm afraid, which rather runs against some of the slightly utopian story about technologies that technologies do exacerbate inequalities. And we quite often kid ourselves that they don't, but they tend.
00:13:04 Speaker 3
To give power and wealth to the people who already have power and wealth, which doesn't mean that they don't offer benefits for the poorest and most vulnerable within society. And it doesn't mean that there aren't technologies that sometimes have extraordinarily emancipatory benefits and are able to spread those benefits among the poorest in society. And we can look at things like vaccines.
00:13:29 Speaker 3
You know, as an example of a technology that does that, but.
00:13:33 Speaker 3
Those things are the exception, right, and in most cases, if we leave technology and technologists to their own devices, those inequalities will widen and it's up to society and societies elected representatives to make good decisions about that technology and say, OK, well, how can we then steer technology?
00:13:53 Speaker 3
Towards the world's real problems.
00:13:56 Speaker 1
So Jack, I really like the point you're making about AI with purpose.
00:14:01 Speaker 1
Especially for addressing societal needs and challenges, because that's where I think AI can add value or has the potential to add value.
00:14:08 Speaker 1
Now let's look at the practical implications of AI. So goal #11 focuses on making towns and cities safer, resilient, sustainable, and includes issues around sustainable transport and planning for the future transport needs.
00:14:24 Speaker 1
You've written a fair amount about some of the practical implications of self driving vehicles and the balance of public and private interests. Do you think that the targets in the goals are keeping pace with the rate of technological change?
00:14:37 Speaker 3
So let me just, yeah, zero in on on self driving cars. Not not because I think there are particularly interesting or exciting technology for the SG's and I'll explain why I think you know there's a lot of questions still to be still to be answered there, but they they might tell us something about where the promise of AI.
00:14:57 Speaker 3
Meets the the reality. So I mean self driving vehicles. I originally got really interested in because they seemed to be a real world case of AI operating in the wild, raising all sorts of questions for society and it seemed to be one of one of these examples of a solution in search of a problem.
00:15:18 Speaker 3
Right, that if you're an engineer, absolutely the challenge of getting a car to drive itself is just totally fascinating. And you can understand exactly the excitement around that and that the progress has been extraordinary than in some parts of the world.
00:15:31 Speaker 3
You can get in a car with nobody sat behind the driver's seat and it can drive you safely from one place to another, which is, you know, an amazing achievement for AI and the and the the people behind it. But does it answer?
00:15:47 Speaker 3
The real questions that the world has in terms of, for example, safe city, so the promise of the self driving car is often that it will remove human error, which is, according to some calculations, responsible for the majority.
00:16:01 Speaker 3
Of what is an absolute public health disaster, right, and more than a million people get killed in crashes on the road every year. That's just an extraordinary public health problem.
00:16:13 Speaker 3
But if we.
00:16:13 Speaker 3
Drill down and ask well will self driving cars help tackle that that problem we we say well OK, where does that problem happen and why does it happen and why are people?
00:16:23 Speaker 3
In danger on the roads and what do the risks look like? And we can look at the mismatch between that problem where you know in in lots of developing countries Rd safety is is a real issue and the answer to that is probably not.
00:16:39 Speaker 3
Put robots on the road, right? The answer to that is probably fairly boring things to do with enforcing rules. Speed limits, the safety of vehicles, avoiding corruption, things like that. And you look at where the technology is being developed in places like Phoenix, AZ, and you would say, well, that is that really tackling the problem. So you see a sort of mismatch between.
00:16:59 Speaker 3
The trajectory of AI and the problems facing the world, and I think the more experience we get with the technology, even if the technology.
00:17:07 Speaker 3
Is accelerating and in a really impressive way. We see that you know, it probably isn't gonna make a dent on on that issue. We might see some sort of compromise that if we want to realise the benefits of AI in those circumstances, we might have to adapt.
00:17:27 Speaker 3
Infrastructures to suit the technology and and make those sorts of societal changes that will actually lead to improvements in in people's lives. But that is not just about AI, right? That is also about the sort of investment in in public infrastructure that makes a real difference. It's not just.
00:17:47 Speaker 3
You know, imagining this sort of plug and play technology where you can get a computer driver and drop it in the middle of Delhi, say and expect it to drive safely from one side of the city to the other.
00:17:57 Speaker 2
And that's interesting because with my health hat on, my cousin was in a car, a self driving car in China, and he actually said that actually he got very frightened by it. And he in the end asked the driver to take over. So something that came to my mind was.
00:18:12 Speaker 2
It may seem a benefit to one person, but what are the unintended consequences? Again for somebody else? So who has anybody? Maybe they've done this measured the anxiety levels, the fear levels, the blood pressure, the heart rate. When somebody's actually in a self driving car or in the passenger seat.
00:18:28 Speaker 3
So a lot of people have have have started dealing with, including some some of our UCL colleagues actually looking.
00:18:36 Speaker 3
That people's responses to this very peculiar set of circumstances, you know, for any technology to be successful and to solve our problems, it needs to become normalised in some way in, in a profound way. It sort of becomes invisible that we stop thinking about it. And it's exactly the same for a self driving car that if it works.
00:18:56 Speaker 3
It becomes really, really boring. You want it to be really, really boring. So the manufacturers of a self driving car would hope that any anxiety lasts a very short time and people eventually just shrug and say ohh this is, this is safer.
00:19:09 Speaker 3
The alternative, if it carries on being worrying, then the technology is definitely not ready for prime time.
00:19:17 Speaker 1
And that's an important point because if we had cities where all the cars were driverless, then it would be the new normal. But where we have a hybrid system, then we have human behaviour and then we have cars with robots or not. And the interaction of that is very difficult to predict.
00:19:34 Speaker 3
Yeah. I mean it's, you know it's it's it's the sort of engineers paradise that if you took out all the people that get in the way you know this is almost exactly a quote from Kurt Vonnegut, right. You take out the people and everything suddenly looks really easy.
00:19:47 Speaker 3
But the thing I like to remind transport people of is that you know, the road is a shared space and we still haven't fully transitioned away from the horse, right? We still have horses legally allowed to travel on our roads. So this is one of the the societal choices that you can imagine. Yes, you know, take self driving cars, have self driving car.
00:20:07 Speaker 3
Only lanes that starts to look really, really safe really reliable. It also starts to look very, very boring.
00:20:14 Speaker 3
Massively expensive in infrastructure terms, and then you've ended up with something like well, in London here, the Docklands like railway like which we've had for 50 years.
00:20:22 Speaker 1
Yes, and they're parts of the world where people lack access to very basic infrastructure, including decent Rd networks to get access to medical supplies and food supplies. So there are parts of the world where we still grappling with issues of basic infrastructure needs.
00:20:37 Speaker 2
And Maria, just just on that. I'm just wondering, thinking about this, maybe all your students in AI.
00:20:42 Speaker 2
Have some psychological tests and we should do some experiments with them in different scenarios and see how how they fare. Say that's something we could introduce, get the doctors and get the scientists and social scientists and AI technologists together in one room around the table to to think about all of these ethical.
00:20:57 Speaker 4
Issues I can tell you that many of them really struggle with the risks it's it's quite depressing.
00:21:03 Speaker 3
When you say that they struggle with, do.
00:21:04 Speaker 3
Do you mean that they they?
00:21:05 Speaker 4
The struggle being.
00:21:06 Speaker 3
They're anxious about the risk, so they struggle to understand the risk.
00:21:07
Yes.
00:21:09 Speaker 4
They they they are anxious about the risks once they see them. It's very hard not to, not to Unsee. Right. So you see them everywhere and it can be quite disheartening, I think just because.
00:21:11
Yeah, yeah.
00:21:15
Yeah.
00:21:20 Speaker 4
This you've been working hard for for something that somehow there was this technological positive mindset of how technology always helps the world and then you realise that the risks are something that you have never seen. So it's it's hard for them.
00:21:38 Speaker 3
It's really profound. I mean, I'm I'm acutely aware of it. I'm acutely aware of in these sorts of conversations often being because I ask societal questions and policy questions and risk and ethics questions. I'm acutely aware of being a massive downer on these debates, and maybe maybe it's existential for some of those researchers who are thinking about.
00:22:00 Speaker 3
You know their life.
00:22:01 Speaker 3
Their lifes work. I mean, one of the things that my team and I have introduced at UCLA is is we teach for a lot of the engineering and physical sciences PhD students. We teach courses in responsible innovation where we.
00:22:15 Speaker 3
And when I say encourage, but actually.
00:22:17 Speaker 3
It's more like.
00:22:17 Speaker 3
Force some of these conversations to take place for the PhD students because UCL believes that it's a really important.
00:22:24 Speaker 3
Part of a well-rounded PhD training, but often people in their area don't want to have those those conversations because they can be quite unsettling. They can induce anxiety or they can be seen as as a distraction from the work that these researchers want to want to get on with. On the other hand, I would just say some of them.
00:22:44 Speaker 3
Are totally, totally brilliant and occasionally you have those, those moments of of genuine interdisciplinary magic where some are really thoughtful. Researcher will engage with those questions of of risk and ethics and it.
00:23:00 Speaker 3
Produce a new A a new research question.
00:23:03 Speaker 2
And we do encourage disagreeing well at UCLA, we do encourage our students to be curious and really foster innovation in a responsible way. And I like you just said, that's something you've been working on for quite a while. So how can you engage with policymakers and development?
00:23:20 Speaker 2
Agencies to create an environment that does foster that innovation and responsible AI in achieving the SDG's.
00:23:27 Speaker 2
And what policies or frameworks do you think are needed to support the integration of that AI into sustainable development initiatives?
00:23:34 Speaker 3
So without wishing to get sort of too wonky ish on what those actual rules and policies might look like, I think at the sort of simplest level it's about.
00:23:47 Speaker 3
Trying to put new voices into the discussion, so at the moment the debate about AI is being dominated by a small number of very, very loud voice.
00:23:57 Speaker 3
Is from people who are all also among the world's most powerful and and and richest people from within the biggest companies. And it means that the perspectives of those people who might actually articulate what the SDG's or other you know, big societal questions might be.
00:24:18 Speaker 3
Get drowned out, right? So we're seeing a huge amount of.
00:24:22 Speaker 3
Of hype about the potential of AI, which might, as Maria says you know, be directed towards things that at the moment the technology industry comes naturally to the technology industry. Like how can we get more people to click on ads? Right. That's not a really important problem. So I think it's up to universities.
00:24:42 Speaker 3
To policy.
00:24:43 Speaker 3
Because to look for those perspectives, to try to put more of the needs into that conversation so that we're not just drowned out by the technological hype.
00:24:53 Speaker 2
So do you think we're shouting loud enough as universities in that arena?
00:24:56 Speaker 3
I mean, I would say we, we do our best. One of the interesting things about about the AI conversation, though is that it's very, very quickly being dominated by the by the private sector. So this is something that I don't know if you would agree, Maria, but but for a university computer scientist, it's quite difficult.
00:25:14 Speaker 3
You know a world in which suddenly researchers are going to move into the private sector because that's where the money and the and the social capital is, where private sector interests are, are dominating. Government funders are working out how they operate in this world. And I think for big multi faculty universities.
00:25:34 Speaker 3
Like UCL, working out what are places in that in that conversation and working out how to shout louder is is is really.
00:25:40 Speaker 4
Vital, and it's even very hard right now to do research in AI at universities. Just because you don't have the resources to compete.
00:25:50 Speaker 4
Even dealing with some of these models, trying to audit some of the large language models that are out there and so on, you don't really have the resources to even do that. So it's becoming harder and harder to do this work in universities. So we need to find a way to support more this role of academia.
00:26:10 Speaker 4
In this space.
00:26:10
It's.
00:26:17 Speaker 1
And throughout the series, we've spoken about the importance of community LED action around the goals and it's very clear that action for sustainable development can be seen as something that governments and agencies impose on communities. But there needs to be a strong element of Co creation. So if we take the Co creation piece.
00:26:38 Speaker 1
And earlier you mentioned the power dynamics as a result.
00:26:40 Speaker 1
Of the leadership from private sector. So let's look at trust. Let's look at Co creation. What does Co creation and community involvement look in the context of responsible AI and how do you build trust and public acceptance of new technologies and engage with communities?
00:26:57 Speaker 3
I mean, I think it's absolutely it's an absolutely vital question. I would I would just tweak it to say that rather than seeking to build.
00:27:05 Speaker 3
Trust and acceptance, and presuming that the technology is just sort of one thing and it's up to, it's up to us to sort of impose that technology on people and get them to like it right, the the task should be to build technologies that are more trustworthy and more acceptable, right. So and. And and I don't think that's just sort of splitting hairs.
00:27:26 Speaker 3
It's about.
00:27:26 Speaker 3
Not genuinely saying the technology is not a sort of fixed object, we can steer it in in, in better directions. Now if we're going to do that, we absolutely need to include the perspectives of some of those people recognising that some of those people, when it comes to the sustainable development goals, will have.
00:27:46 Speaker 3
Some of the quietest voices in in global society, right? So those voices need to be need to be.
00:27:52 Speaker 3
Qualified and we particularly in the West in privileged universities need to not impose our framings upon those those people. So there's a huge project to be under, which is sort of behind the SDG's, if you like to understand and be able to articulate genuine societal needs in ways that.
00:28:12 Speaker 3
That make sense to the the people that might.
00:28:16 Speaker 3
Innovate towards them and that does take us to potentially a really radically new version of innovation, which you expressed, I think rightly in terms of Co creation. You know, what would it mean for AI rather than it being something that feels like it's being done to people to be something that's done with them or even buy them?
00:28:36 Speaker 3
Right. Rather than this mode of innovation in which, you know, technologies are sort of imposed upon.
00:28:42 Speaker 3
Poor communities. If we actually understand and you know, you'll know better than I do. There's a history of of debates to do with appropriate technology and inclusive innovation, where these things have been have been tried, but it's often difficult to make those make a difference at the sort of scale that is required. And here in AI, we have a technology where.
00:29:03 Speaker 3
The scale and the concentration and the speed with which the power has been built up and concentrated with a very small number of actors has.
00:29:11 Speaker 3
Been extraordinary. And for policymakers, there's a sort of sense of whiplash. It's like, well, how do you deal with that? What should we do about about that and making the case for the sort of work where we work with, with communities to understand their perspectives, which might actually be slow and thoughtful work rather than, you know, this, this story about AI that we often tell ourselves, which is it's a race, and you have to keep up and you have to get ahead and.
00:29:36 Speaker 3
And all the rest of it, that's that's going to be hard.
00:29:39 Speaker 2
And working with communities as we know from our own work takes time, takes energy takes.
00:29:44 Speaker 2
Courses like you said it needs to be done thoughtfully and people really need to understand again if I'm struggling with what AI is and the pros and cons, we need a lot of work with the communities to really, really embed ourselves in those conversations in a meaningful way. Like you said, not just a top down way which comes to a bit of a challenging question for you both at the close of this really.
00:30:05 Speaker 2
Maybe a bit of an unfair question, but answer it as you like. So is AI a good or a bad thing for addressing SDG's Maria?
00:30:14 Speaker 4
So we've been talking about AI with purpose, and I do think that AI has had.
00:30:20 Speaker 4
A very clearly defined purpose for a long time, and that is called AGI. That acronym is artificial general intelligence. So we've been trying to create agents that are as intelligent as us or even more, but I think it's time to shift the goal regardless of whether the goal is worth it.
00:30:40 Speaker 4
We need to solve first some of the challenges that we are dealing with and we've mentioned many today and I think it's as Jack has very well said, is a matter of resources, the choices that we are making are political. I really like the the case that you mentioned about self driving cars and we see the same in many other.
00:30:59 Speaker 4
Ads like ad tech educational technology, where we are investing so much in creating technology that can almost take over teachers, but we are realising we cannot replace a teacher and because we are not working toward supporting a teacher, the technology is really not delivering. So those resources that.
00:31:17 Speaker 4
We could be.
00:31:18 Speaker 4
Putting into better education in other ways, we are using it in to build this technology that is not the.
00:31:24 Speaker 4
So right now for me, I think that AI is more inhibiting targets that enabling them, even though I've been speaking a lot about the opportunities, I think the recent matter of resources combined with all the risks, things like truth decay, risk to democracy, new ways of crime and.
00:31:44 Speaker 4
River crime emerging, for example, with boys cloning, political polarisation, even amplification of stereotypes in humans, because we are seeing this AI companions, AI girlfriends, AI boyfriends that the Jews are using and those are increasing stereotypes even in humans.
00:32:00 Speaker 4
All of this is making me think that right now it's inhibiting sustainable development rather than enabling it, but as Jack very well said, we are the ones constructing it. So we can turn this around and we can create technological agendas for the future that we want to see and for the technology that we want to have.
00:32:19 Speaker 4
As for me, the the key here.
00:32:21 Speaker 3
I think just to, I mean I I don't have much to to add, but I'd share the the concern about this idea, this sort of utopia of artificial general intelligence, which does really get in the way of a sensible discussion about.
00:32:35 Speaker 3
Not how technologies can be used for good, right? Because it it basically the the language of artificial general intelligence leads to a a mode of innovation that says we innovate for innovate innovation's sake and leave us alone and get on with it right. And I think if we're going to realise the.
00:32:56 Speaker 3
Opportunities of artificial intelligence. We need to disrupt that, that that conversation.
00:33:01 Speaker 3
I mean one way that we can avoid, I think the sort of fatalism that comes with that is to say that AI is not one thing, right? It might be a sort of suite of opportunities. It might be a, a, a set of of technologies. So, you know, using AI to enable world class education to be accessible.
00:33:22 Speaker 3
Around the world, or to enable agricultural developments to to be spread in, in, in new places might.
00:33:30 Speaker 3
Be very, very.
00:33:30 Speaker 3
Different as a sort of technological project.
00:33:32 Speaker 3
From improving chat bots or the efficiency of of of advertising, and so ultimately I mean it, it comes down to the same thing which is saying, rather than framing the debate in terms that suits the technology industry, we need to frame the debate in ways that start with the problem, turn it on its head and say.
00:33:53 Speaker 3
If you start from the Sustainable development goals rather than starting from the exciting technology and then ask what can technology do for us, then there's the possibility of a of a constructive conversation otherwise.
00:34:05 Speaker 3
Will just fall victim to the same hype that we're that we're seeing around us all.
00:34:09 Speaker 2
Time. Well, I knew that wasn't going to be a quick answer after all, that is all. And that's been curious, isn't it?
00:34:14 Speaker 3
Social scientists will always say it depends. It depends.
00:34:20 Speaker 1
But it sounds like if a is very much going to be part of the future, and I think.
00:34:25 Speaker 1
It will be.
00:34:26 Speaker 1
We need to make sure AI is purposeful and we need to make it work for us and for achieving.
00:34:32 Speaker 1
So it's partly down to us to make sure we realise the potential unlock the good that AI can do, acknowledging all the risks, Maria, that you've identified earlier. Thank you both for joining us today. Welcome listeners find you online.
00:34:49 Speaker 4
They can find me on LinkedIn.
00:34:51 Speaker 3
And they can find me for my sins all over Twitter, which I insist I still call Twitter.
00:34:57 Speaker 1
And not tax. But thank you very.
00:34:58 Speaker 2
Much you've been listening to unlocking the SDG's.
00:35:02 Speaker 2
This episode was presented by me Professor Monica Lakhanpal and me Professor Preeti Parikh and produced by the UCLA SDG's Initiative and edited by Frontier. Our guest today were Professor Maria Perez Ortiz and Professor Jacks Durgo. If you'd like to hear more podcasts from Lucio, subscribe to UCL.
00:35:23 Speaker 1
Wherever you download your podcasts or visit https://www.ucl.ac.uk/sustainable-development-goals/ join us next time on unlocking.