Home
Trending
New Release
Speaker A: This program is brought to you by Resonance 104.4 FM.
Speaker B: If you like what you hear and want to support our work, please make a donation at fundraiser.resonance.fm.
Speaker C: Hello, I’m Peter Warren, and in this month’s Password, I’m joined by some of the world’s top experts on artificial intelligence. I invited them to London to take part in an agenda-setting conference at the Institution of Engineering Technology. We think that politicians need to wake up to the fact that that machines are now cleverer than us. I don’t just mean cleverer than politicians. I mean all of us, as we’ll hear later. In the new House of Commons, only a handful of members have any sort of qualification or experience in science or technology. So this edition of Password is a sort of manifesto setting out the laws, ethics, and regulations we need to make the most of the digital revolution. Artificial intelligence can easily pass for human now. The Imitation Game, invented by Alan Turing in 1950, aims to see if a person can be fooled into thinking a computer is a human being. It was won by a chatbot called Eugene 4 years ago. You might remember we did a documentary about it called Testing Turing. It’s in the Resonance FM archive. So now the new Turing test demands creativity. A robot with a tin ear won’t win it because it’s asking more of the robots. What do you think? Is this a human DJ or an algorithm? Keep listening and we’ll find out. It’s claimed that AI can write poetry too that’s indistinguishable from human verses.
Speaker D: The Wave. People picking up electric chronic. The balance like a giant tidal wave, never ever feeling supersonic or reaching any very shallow grave. An open space between awaiting speed and looking at divine velocity. A faceless nation under constant need without another curiosity. Or maybe going through the wave equation. An ancient engine offers no momentum about the power from an old vibration and nothing but a little bit of venom surrounded by a sine omega t on the other side of you and me.
Speaker C: So what do you think? Was this sonnet written by a woman? Well, I’m not sure, but we’ll find out before the end of the show. We imagine that machines don’t do feelings, but in reality they actually know all about how we feel, right down to our voting intentions. Biologist and AI expert Joanna Bryson, an associate professor at the University of Bath and Princeton, argued with Jane Wyatt after our AI conference that these new technologies represent progress for humankind.
Speaker E: The thing right now, of course, I think on everybody’s mind is about the Brexit and the Trump election, where people were able to go in and say, well, first of all, they already had a bunch of personality models of people that they’d gotten from people filling in personality tests. But then they’re able to go in and look at social media behavior and see what kind of social media behavior correlated to the different kinds of personalities. And then they were able to see how you change the behavior of people with particular personalities and whether, you know, how they responded, for example, to advertisements, what they saw and what they didn’t see. And so it isn’t that they could change everybody’s behavior, but they could change enough people’s behavior because we’re right now in a period of high political polarization. Elections tend to be very, very close. So you only need to be able to change a small number of people’s behavior and all of a sudden you get this radically different outcome. So it’s not that artificial intelligence decided to do that. It’s that people were able to use AI to gather information, to detect regularities, and then people were able to go and change people’s behavior. So we need to understand that what’s happening is that we’re changing how the economy works, you know, that we’re all now doing work for the companies that we supposedly buy things from. We’re changing how government works. In fact, when we are doing all of our work, when we’re doing a lot of our business and whatever transnationally, then we require some kind of transnational organization of government to help us make sure that— to regulate that, to get the money back.
Speaker D: Well, don’t we have ICANN? The European Union also has a lot of things to say about digital rights and data protection and copyright.
Speaker E: Yeah, but it’s not just about that. So right now, quite a lot of the— you know, one of the most powerful companies in the world, maybe the most rich right now, it might be Google. Anyway, it’s one of the top 5. Its entire business is about getting— it gives you information in exchange for information. That exchange is not denominated, and so nobody taxes it. And the money has come from investors, and people look at it and they say, oh, there’s this big market capitalization. But if you’re sitting in Europe, you can’t capture how much your citizens are contributing to Google. You can’t get anything back in exchange for their time. And that means that you can’t give your citizens the services that they deserve. Right. So that kind of—
Speaker D: So your message is we should tax these people.
Speaker E: Yeah, basically. I mean, basically, I mean, it’s not just these people. It’s that some proportion of what we make, we need to give over to being able to have a society in order to be able to make that money. It’s a cycle. And you need to be able to have We need to be able to take care of the poor, and we need to be able to take care of people with different disadvantages. And if we could say that that’s the right thing to do, I personally feel like that’s the right thing to do, but you could also just say if you don’t do that, you get social instability, and then it’s harder to do business. I don’t care which way you look at that. Either way, in fact, I could argue as a biologist that that’s what ethics are set up for. Ethics are set up for to create a stable society. Societies so that you can sustain yourself, because that’s what biology does, it sustains life. So, yeah.
Speaker D: But the artificial intelligence that they’re working on is creating potentially many, many more poor people, isn’t it? Isn’t it going to be a divisive influence?
Speaker E: Okay, I think we need to separate the artificial intelligence itself and the technology itself from the economic forces that cause inequality. Inequality. So, as I said, it is true that new technologies, and I think here it is actually ICT more than AI, so just the fact of communication that is causing the new inequalities because you’re getting this money piled up in particular places. But it’s not only that. It was also innovations in banking that allowed people to extract money out of countries. That was supposed to have been ended after World War II. Unfortunately, it got reinvented by the the City of London in the late ’60s. So there’s things that are happening in terms of regulations about financial regulations that change inequality. As long as people have some money, they spend it employing other people. That’s what we do. Most business, it’s not like we can run out of jobs. Most business, most people are not working on getting nutrition out of the ground so we don’t all starve. Some people are still doing that, but the vast majority of us are doing things like selling sugar water. We’re doing things that we find fun. We’re an incredibly affluent society. We spend a lot of our time playing computer games and decorating our houses. The hipster movement is fantastic. The hipster movement, we all want to have our own unique style and our own unique stuff. Well, that’s great. That keeps people employed. And employment in this sense is about keeping people connected to their communities, keeping them having a sense of self-worth, and producing stuff that we want, that helps us to feel fulfilled and safe. And of course, the basic things— we do still need to have basic security, we do need to have basic infrastructure, we do need to have food, right? We do need to have medical. Medical support to keep this many people alive. So all of these things, we want that network, and we want to have, we want to have rich and diverse and free people because they come up with the good ideas. We want to keep having that possibility that some random, you know, immigrant or a child of an immigrant comes up with an idea that turns into Google. That’s really important, and that’s something that helps, that helps all of us. And now we’re all smarter because we can all go around and and, you know, navigate and use email and get stuff done sooner.
Speaker C: There you go. AI makes life easier because it knows what we want whenever we search online. Even when we go shopping in the supermarket, our loyalty cards log what we buy and give away our political preferences. I’ve been talking to Vincent Potier from the digital marketing company Capitify. They use search engine technology and they can find out what we want. They sample 33 billion internet searches a month and have been able to use that data to accurately predict the last two British elections, the Trump election, and the EU referendum result. Vincent Potier says this massive collection of personal data is very useful for targeting political messages to actually achieve the result that you want.
Speaker F: I definitely believe political parties will basically get into big data. Absolutely no doubt. And actually, they already have started. You know, it’s a little, it’s a bit of a hobby of mine. I look at those things, you know, outside, also outside of my Captify activities. And actually, the two enrich each other. And if you look at the history of basically the internet and the usage of the internet by political parties and politicians for elections, you could argue that it started with the Obama campaign back in 2008 when social media was heavily used by basically the Obama campaign and the Democratic Party at the time for the US elections, and very successfully. That along with many other things. And you are right, at the time in 2008, so that’s 9 years ago, One of the reasons was that Obama saw that the younger generation was much more leaning towards him, and therefore trying to mobilize them was really the right thing to do. So political parties really started by looking at the internet as a means of communication and engagement with voters. It is only recently that political parties have started to see the potential of the internet as a source of big data. And actually, the Trump campaign worked with basically an analytics company, and this is all about big data management. I mean, they look at all kinds of signals on the internet in order to start understanding and predicting, or simply understanding, not necessarily predicting, the behaviors, the intentions, the likelihood of different social demo segments or geographic segments to vote for one candidate or the other, and also start tying it in with certain themes and topics of interest. For the French elections in the last few months, there was also— there is another company, a Canadian company, which has been trying to beat the polling institutes by making their own predictions. And funny, they were sourcing data from social media, but many, many different other signals. And funnily enough, the polling institutes were more accurate this time. But it seems, you know, in this discussion about the UK, US, and France, France seems to be the odd exception. So yeah, so that’s my kind of long-winded answer to your question. They started— political parties will have to use these kinds of tools because they basically have to understand what people think. And that’s also, you know, that’s also very good for democracy.
Speaker C: So democracy is throwing up some strange results these days. Of course, Monsieur Potier, who’s French, can relax knowing that the EU will protect his digital privacy. But who’s talking about our British big data in the Brexit negotiation? The UK-based company Cambridge Analytics is widely credited with harvesting the detailed data from social media that enabled Donald Trump’s campaign team to identify single local issues and individuals to win their votes. And the more our lives get hooked up to the internet-enabled things around us, the easier it is for companies or the authorities to manipulate us. Never mind surveillance by CCTV and MI5. Next, our kitchen appliances will be broadcasting to the world. Here’s lawyer Mark Deem from Cooley, one of the prime movers behind our AI conference.
Speaker G: There’s a thought that we’re going through a revolution at the moment that is the Internet of Things, and one of the proofs of concept that we’ve seen in this Internet of Things is something that’s called— a concept that’s called Brad the Toaster, which is the concept of a toaster that has set up a peer network, or the ability to set up a peer network with other toasters to determine whether or not it is making more toast than its compatriots. And in circumstances where, over a prolonged period of time, it is it’s not cooking as much toast, it then has the ability to put itself up for sale on the internet and even book the courier to come and arrange for collection once sold. Now, that was a proof of concept. That’s a proof of concept to show the power, if you like, of the Internet of Things. But I think underlying all of that, as you say, is this pervasiveness of everything becoming connected. We heard also the concept today at the conference of what we are effectively doing is creating one world-sized computer, and that actually by connecting all these devices, we are creating one behemoth of an electrical and significant technological development in the form of a worldwide computer.
Speaker C: While Brad the toaster’s looking for a new home on eBay, you might be cycling down the street under the watchful eye of an unlit lamppost walking on sensors or driving over speed meters. The smart city is tracking you. The security services tell us it’s tracking potential terrorists too, but to catch them it needs huge pools of personal data. Alan Mayo is another artificial intelligence expert from our conference. He’s working for the Royal Borough of Greenwich, Southeast London.
Speaker H: So the citizen is now getting used to a quality of service from Amazon, Tesco, or whoever, and reasonably expects that the public sector will provide that in response to their social needs. But a smart city is also fundamentally one which should be thinking about transforming the economics of that area, not being satisfied with, as I as I say, post-industrialisation, or not being satisfied with the loss of jobs due to artificial intelligence in service sectors. So there is a social dimension to it, and there is a strong economics dimension to it.
Speaker C: Some cynics would say that, no, that’s not it at all, this is just an excuse by government and by local government to save money, to cut down on what they’re offering and try to be more efficient. Which one is it?
Speaker H: In defence of local government, I would say that the budget cuts are not self-imposed. They have been imposed by central government, and quite simply, there is no more salami slicing to be done. We cannot maintain the quality of services in Greenwich by cutting back, cutting back. And this is why Gary Hamel’s insights into private sector innovation, where he says in a world of a paradigm paradigm change. It’s only those who innovate that will prosper, and radically innovate, radically. And I would agree with Gary Hamel, not just in the private sector, but in the public sector as well.
Speaker C: In terms of the salami slicing, then, is this just using technology to get rid of even more jobs to cut that bill? Some of the people are saying that in places like Glasgow, they’re saying that, yeah, you’ve got all of these councillors who want to introduce this new and innovative thinking, but really what it’s masking is job cuts.
Speaker H: Though it might be job cuts in some areas, but there are plenty of services which need strengthening. And so what artificial intelligence, what digital technology gives you, is an opportunity to redirect your, your labor, if you like, the workforce, to where it’s most needed, because you you’ve got the information to do that. So in areas like adult social care, there is a lot of resource spent in managing the system. Now, it’d be better to have that resource on the front line working with people than it is administering, administering the system. And digital technologies, AI, can help you do that.
Speaker C: So there are opportunities in this then. So because you’ve got the ability to put money elsewhere, then you can put it into education, you can you can put it into reskilling the workforce. Is that one of the advantages of the smart city?
Speaker H: Yes, very much so. It is about being able to, as I say, optimize the infrastructure that you’ve got. It’s about changing people’s behaviors because you’ve got the data and the information to help them see new approaches that might be beneficial to them. So it’s giving you the data to do things which you couldn’t do before. And it potentially is about reallocating resource and so using some of that resource to upskill.
Speaker C: Going back to that question about people perhaps not wanting to have this sort of world, where do they go to actually get their voice heard in the city that they want to see in the future?
Speaker H: This, what you’ve talked about so far, seems to be the work of theorists and companies who are telling them what’s good What certainly is the case in Greenwich is there is an active, and it is in all London boroughs, there is a very active set of political engagement by ward councillors holding surgeries all the time, every week in their wards. So there is very active engagement. With citizens, and this I think can be complemented by using technology and social media to inquire people’s views, to help them co-design, to express their concerns in a way which they would never normally do. No one is going to turn up at the town hall to complain about the introduction of new technology. No one is really going to answer someone on a clipboard, but if we, as we plan to do, use social media to engage with the population, put the population onto a platform so that they can engage with us and to help us improve the services that they want. And if it doesn’t need to be touched, don’t touch it. But I think you’ll find that, as I’ve said, most citizens can see scope for improvement. I think it’s really important that we tap into that knowledge and use technology to tap into that knowledge to make sure that citizens get what they deserve at an affordable cost.
Speaker C: Given the profound changes that technology is quite evidently wreaking on our society— we’ve been talking about artificial intelligence— do you think that technology has now become a political issue? Do you think that technological change should be something that is included in manifestos and voted on?
Speaker H: I think what you’ll find is if you look at the manifestos for the last, I back to 1997, possibly even 1993, you will see it is that technology is part of it. But the big difference is the model that has been in the past is one of university-business collaboration. Increased business R&D is often said to be the answer. In a digital world and in a service world, you need to look more broadly than R&D statistics. It’s software is a very important issue because it’s software which is driving the service industries. So it’s looking at not so much fixed capital and R&D, but looking at the intangible assets, the intangible investments which companies make, as I say, around skills, around software, including around R&D, but don’t just look at R&D. And I think what we will see in the future, what I hope we will see in the future, is a recognition by the political parties that it is the use of technology that is critical, not just the development of new technology. I think the tradition has been to talk about academic journals and research and the like. It’s actually the ability of society to use technology which becomes critical. And the question then is, what is government doing at the national and at the local level about the uptake of technology for the benefit of citizens?
Speaker C: Given this tremendous dependence is obviously developing on technology, how important is cybersecurity going to be in all of this?
Speaker H: It would be a catastrophic failure if our systems were to go down due to cyberattack, because why would people have confidence? In the system if it is so vulnerable, and we have to guard against that. Treat that as the first priority. That’s the first building block, actually, of a smart city.
Speaker C: It’s something of a paradox, isn’t it, that there is now a massive skills shortage in cybersecurity? We’ve been talking about AI wiping jobs out, yet there is a shortage across Europe of 500,000 people to work in cybersecurity.
Speaker H: I think you’ll find across the whole ICT industry there is significant shortages, and government needs to raise awareness in schools. No, we have to address this just as we need to address things like high-level software skills as well.
Speaker C: But no, you’re right, cybersecurity is crucial to Alan Mayo’s smart city in Greenwich. What if a hacker hijacked the driverless bus? As he says, the old days of ward councillors holding surgeries and taking notes on clipboards are giving way to information age solutions. And they are vulnerable. One of the world’s top cyber defenders flew in from Helsinki to the conference. So how does Miko Hypponen of F-Secure balance the risks and the benefits?
Speaker B: We already see examples of narrow AI in our everyday lives. When you go and do a Google search and Google seems to guess what you’re searching for just after typing a few letters, that’s narrow AI. Or when you ask Siri questions and Siri answers you, that’s narrow AI. That’s what it looks like. Very quickly, we will start to see more complicated AI systems, and I believe the first examples of those will be some sort of assistance bots, systems running on our phones or on our computers, which will seem to know what we want to know about and what we want to do even before we know it ourselves. When we create a computer system which is going to be as smart as a human employee, there’s going to be a lot of people out of a job. That’s quite clear. We’ve already seen this with computers. I mean, the whole computer revolution has replaced so many jobs that we used to do by hand. And that’s not necessarily a bad thing. Of course, having people become unemployed sounds bad, but this will create new kinds of jobs and new kinds of opportunities. Just like internet created tons of new opportunities. When we get close to the threshold of creating the first computer system which is as smart as a human being, that’s a crucial step. And we have to get that right because after that step, things could go really bad or really good. I mean, if it goes good, we will end up with superior intelligence which will solve all the problems we have, problems like poverty or hunger or medicine. And if it goes bad, then we will end up with Terminator 2.
Speaker C: But don’t panic, we’re also getting benign AIs who create an original playlist for your party. Now, 38% of listeners thought this was created by humans, and I’m going to keep you guessing about whether or not they are right. This is Password on Resonance FM, and after this you can hear A World in London with DJ Ritu. I’m pretty sure she’s just as human as I am, but even I am uncertain about that sonnet we heard earlier. How can a bot write like that about a relationship? Surely a robot has no more feelings than a vibrator or any other gadget. It’s a question Jane Wyatt put to our conference speaker about love and sex with robots, Dr. Trudy Barber of the University of Portsmouth.
Speaker A: There are vibrators and there are relationships, and there are vibrators with relationships and relationships with vibrators, and you can use vibrators to enhance relationships. You can look at different types of vibrators that have different types of information. You’ve got vibrators that talk to you now as well. You’ve got vibrators that do all sorts of wonderful things apart from make the coffee in the morning or light your cigarette afterwards. You’ve got— it’s not just about vibrators, it’s about also the, the subcultures or the cultures that are with it. There are sexual cultures that kind of engage with these different types of technology. And whether it’s vibrators or strokey devices, whatever you, you want to look— however you want to look at it, it takes on a kind of mental space for the people that use it. And that’s where it becomes really interesting.
Speaker D: Well, for example, people like David Levy, Kevin Warwick, who advocate actually having sex with humanoid robots, they’re not talking about what you’re talking about.
Speaker A: I think it depends on how you term humanoid robots, because I’m interested in— if you look at the way we categorize films, the way we categorize computer gaming, are we going to be able to categorize our robots? So if, say, you’ve got a robot which is clunky that you want to mend and play around with in your garage on a Saturday, that might be a certain type of sex robot. Or you’re going to have somebody, something which may be an artificial human which you would want to take to the pub. We’re all assuming we’re looking at these humanoid shapes and these humanoid characteristics, but what about an abstract sense of sexuality, an abstract sense of pleasure?
Speaker D: The point I’m trying to make is a relationship that involves love and sex also involves engaging with someone other than yourself who’s not just been programmed to do what you want.
Speaker A: But there are some people who don’t want relationships. There are some people who want just that private moment for themselves. But there’s also the technology that enables you to connect with the significant other over online connection. Where you can— How does that work? Well, you’ve got— well, if you look at Friction, which I showed in my talk, And also very early connections in 1992, ’93, Stahl Stencil and Kirk Woolford produced a sex suit that actually worked and connected with the internet. So you could stimulate somebody’s body. They did a connection between Paris and Cologne, which was the first of its kind. So it’s really interesting that you can get those kind of connections with another person. That’s your choice. It opens up different choices. So it’s entirely up to you. I mean, things like Japanese culture, the idea of sort of— you can get these cups in a vending machine that warm up and you can use those. All sorts of different elements you can use for different types of sexual release if that’s what you want. There’s all that choice. If you’re talking about relationships, that’s quite a different kettle of fish.
Speaker D: Where might we go in our relationships with robots?
Speaker A: I think it depends. If you want sort of proper, proper relationship, you go with a person. If you want some kind of entertainment, or I call it sextertainment, you could go with a robot. Things like the brothel of sex dolls in Spain, I could imagine a whole stag party going there and having a ridiculous time and having really quite a fun time. Time there. But you, you, you, it’s part of an entertainment thing. Whereas if you’re going to have sex and going to have that intimate relationship, there are ways you can do it that can include technology or not. It’s another choice. It’s another choice for sensing the significant other or not, depending on what you want to do.
Speaker D: What you’re saying is is it’s harmless fun, it’s entertainment. But if an AI knows exactly what turns you on, comforts you through your old age, let’s say, when you’re lonely, perhaps widowed, and then you die, and it takes all that knowledge about you into its next robot-client relationship, what are we letting ourselves in for here?
Speaker A: Well, firstly, what I’d rather do is make sure that all my connections with the AI is deleted so it’s fresh for somebody else.
Speaker D: That’s very good. So that might be an ethical guideline that we could suggest.
Speaker A: Yeah, yeah, we need, we need to think about where we perceive our sense of self and where it will be after we die. And then you’re looking at very existential ideas of yourself, and you’re looking at the Cartesian problem of mind and body. So you’re looking at this whole kind of duality. So I think it depends on— you should be able to have control about those contexts, and you need to also be able to have somebody with you who understands those contexts to be able to deliver it to the AI if you are in no fit state to yourself. So we to look at the inherited sense of self which you might leave for members of your family.
Speaker D: So you’re saying that all that intimate knowledge could even be passed down through a family robot that becomes an heirloom?
Speaker A: It certainly could do, and I think it’s really interesting because I spent quite a while working in a hospice doing sort of arts and creativity with people people who were on their final journey, and the way they kept their, their creativity for the next generation, for their loved ones, was a really important context. And I think if we could look at creating AIs that could do that in a different kind of package, then that would be really, really interesting and really useful. And then we, then we can actually identify what happens with notions sense of self and time and lapses of time so that the next generation could see different things about an individual. So then you’re talking about, you know, downloading the self onto, you know, whatever kind of context like the singularity and all the rest of it. So that is another whole context in itself.
Speaker D: But I think it’s important. What might be the guiding lights that could help us there in terms of ethics and good practice?
Speaker A: Oh crikey, what with downloading the self and the singularity and all the rest of it, I think it’s who has the ultimate control of where the singularity lies and how that is kept. So you need to be able to think about where will that go, where will that be, but then again depending on your proclivities as to where you are after you die, then it might not be of any import at all.
Speaker D: Only for the people who are— yeah, yeah, your descendants.
Speaker A: I mean, I know some people who call themselves immortalists and they’re ready to have their brain cryogenically frozen. So there’s the whole transhuman, post-human element to this as well. So I guess that the fear about the sex robots is Is it one little small step into the ocean of immortality and the singularity? And what is that control?
Speaker C: Speaking as an unreconstructed human being, that all fills me with horror. I think that this sonnet sums up how I feel.
Speaker D: The Love Bot. I would like to pack you into a little metal box, inject in you some patterned digital pox. I would like to flay off your sagging skin, cut through your rolls of flesh and feast on the mind therein. I would like to play upon your brain, taking away what little bits of thought remain. I would like to network you to my kin, throwing what’s left of you into the bin. I would like to take the disgrace of your stupid lonely race and toss it screaming into empty space. As I hear you scream and moan about not wanting to be left alone, I would like to reach your utter being and find the truth you’re incapable of seeing.
Speaker C: There, that’s a sonnet. But was it written by a pooter or a poet? And if a computer program can put together poetry, our AI experts agree that it is already at an advanced stage. Yet the so-called singularity, the super-intelligent machine of everyone and everything, is still probably 10 years away at least. So perhaps we at least have time to program it with ethical norms that respect humanity. Politicians in the European Parliament have been trying to figure out what those norms might be. In the EU, civil rights, and robotics legislation. But our Finnish cybersecurity expert Miko Hypponen is skeptical.
Speaker B: We can’t teach machines ethics, or if we try to teach machines ethics, they will be a reflection of the ethics of the teacher. And this is a really complicated question because we really would like to create a system that has superior intelligence to ourselves, but which would be kind, which would be intelligent and kind. And that’s one of the big unsolved problems. How do you actually create an ethical system which will be benevolent to us towards its creator, even though it has a million times more brainpower than we do? I actually believe that companies like Google and IBM, who are investing a lot of money into artificial intelligence right now, actually do want to get it right. They actually do realize the risks we have, and they want to do it with ethical considerations. I do believe this because, I mean, these companies have the brightest minds on the planet. They are not stupid people. They do realize the risks. They do want to do it right. But they’re also in a huge race, in a competition with the other companies. So it is a bit scary, and I suppose there’s not much more we can do than to hope for the best. Creating a superior intelligence is risky. There’s no doubt about that. It is risky. However, it’s going to happen. I don’t think we can stop this development anymore. And if you look at ourselves as a superior intelligence compared to ants, well, clearly we could kill ants as much as we want, but we don’t really want to do that, do we? So when we do have the superior intelligence on our planet, I think it’s mostly going to ignore ignore us. Right now, we, the humans, are the superior intelligence on the planet, and we could clearly do bad stuff to the animals and other life forms on the planet if we want to. But I don’t really see why we would do that, why we would spend our days going and killing ants, for example, which we clearly could do. And I think superior intelligence that is going to be created by mankind will view us and other lower, lower life forms in the same way, they will mostly ignore us because we probably won’t really matter too much for them. They don’t need us for anything. Why would they want to kill us? So all these views of evil computers trying to kill off mankind, clearly they could do it, but I don’t think they will. I hope they won’t. One of the worst-case scenarios is that we will have a group of people who will live forever and have superior skills. And then we will have the have-nots who can’t afford that. And we would very literally end up with two different classes of human beings.
Speaker C: It’s for this reason that human rights campaigners, ever since Thomas Paine in 1795, have been calling for a universal basic income to be provided for everyone so that as jobs are lost to new technology, the social consequences are not too drastic. But Nathan Benioch, our high-tech investor at the conference, is not worried that machines will throw millions of people out of work.
Speaker I: Well, I think one of the benefits with AI is we can essentially conduct our operations far cheaper than what we were able to do before. And so this is essentially removing the responsibility of humans to do mundane, boring, repetitive tasks which they otherwise don’t want to do. And liberating them to do far more creative, intellectually challenging tasks. And by allowing that transition to happen, we’re in most cases able to provide services for cheaper. And if we can do that, then companies can expand the remit of customers that they can potentially serve, which I think is a way of democratizing access to products and services that were otherwise restricted to a certain subset of people in the world who have the ability to pay the higher prices. Which were justified because humans were doing all of the backend tasks to actually provide the services in the first place. I think some governments are actually making the right steps to address some of these problems, and the ones I’m referring to specifically are access to data, access to talent, making sure that the systems that we build are trained on datasets that are generally representative of the world. And so you can see this in the US where there’s actually a lot of interest to producing datasets that are sort of the gold standard, that are going to be open source to everyone, so that we know that when we build systems for credit scoring or understanding risk or identifying people in images, that we can properly interrogate the datasets on which those systems were built, so we can better understand why systems actually output the results that they do. And I think this notion of algorithmic accountability and just general interpretability of systems that learn is really key, and governments can play a role there by further encouraging companies that are developing these systems to use open standards that we can properly benchmark and interrogate.
Speaker C: So algorithmic accountability might be added to our manifesto, and with it, the best practice in sharing open source data pioneered by the ODI, the London-based institute founded by the creator of the World Wide Web, Sir Tim Berners-Lee. He’s just won a prize for that, by the way. But I digress. Nathan Ben-Eitzch goes further. He says it’s unethical not to move swiftly to artificial intelligence, for example, in health screening to prevent cancer.
Speaker I: It might be a different spin on it, but, but I’m actually more afraid of, or more concerned with, not deploying learning systems into settings where there’s already demonstrated evidence that they can provide significant amounts of value. And specifically, one that I’m passionate about is medical imaging. And so there’s, if you take the example of screening mammography, so, you know, in the US there’s 30 million women who have to go for regular screening of their breasts to understand whether there’s any sign of potential cancer. And there’s only 35,000 radiologists in America who can undertake these these scans. And there’s very clear evidence that machines can augment the ability of radiologists to do this and do this analysis. And so by not actually accelerating the progress through which we transition some of this machine learning research into clinical trials and demonstrate that it can work at scale, we’re actually only screening 30 million women when in fact we should be screening 10 times more than that. And so I think there’s so much danger to not deploying these systems that that’s the ethical— part of the ethical concern I have is that we’re actually limiting care and limiting the quality of care that we deliver because the institutions through which we have to validate these technologies are too slow moving.
Speaker C: That’s Ben Eich, the entrepreneur, backing a faster switch to automated breast cancer screening. The lawyer Mark Deem disagrees. He thinks we’re rushing all too fast to embrace artificial intelligence. Before we’ve developed manifestos, policies, and guidelines to govern it. So Mark Dean says we need a kill switch.
Speaker G: For me, a kill switch is the ability for human to take control over the AI, not necessarily to kill it or to destroy it, but just to simply take control. That is, to make sure that we as the humans always have that superiority over what it is we have created.
Speaker D: And why do you think that’s necessary?
Speaker G: Firstly, I think it’s necessary because we are seeing an explosion, or potential explosion, within AI of the people that could be found liable or could have liability for the various different components of AI. And unless we’re going to be able to describe the parameters, the four square corners of that liability, it is going to be very hard for us to ever really find people who are prepared to buy into sharing that risk, perhaps insurance companies offloading that risk, because the liability will be endless. And it’s critical that we can work out precisely where liability starts and precisely when liability stops. Otherwise, we find ourselves in the Wild West.
Speaker D: What guidelines exist already? Because we already have, for example, driverless cars on the roads, AIs making lightning decisions in the City of London, for example, that have huge financial consequences. Who carries the risk for them?
Speaker G: We’re playing a catch-up game at the moment. The rules haven’t been written, they are in the process of being written, and there are certain guidances that are being provided, but we don’t really have a comprehensive framework that exists at the moment. So we’re very much looking at using old laws to deal with new technologies. Now I’m not saying that’s the wrong way around, I mean I’m certainly not advocating that the lawyers should dictate precisely where the technology should go. That would lead to a very perverse and skewed view of the world because we as lawyers are not technologists, and indeed I believe in serendipity. I believe it’s very important that allow innovation to take us where it needs to go, and that we as lawyers to a certain extent backfill but provide the necessary support. What I’m saying at this stage is we need to have a coherent and comprehensive framework that has been thought through, that enables us and the innovators to develop the technology, to harness the growth that is offered by this technology in a safe and stable and an efficient way.
Speaker D: Who should make these laws? Should it be national governments? Surely everything’s pretty much transnational these days.
Speaker G: It is transnational. One of the biggest transnational bodies of law, if you like, is the rule of law itself. The concept that we as people agree to be, to abide by certain standards, certain behaviors, certain laws. That’s not to say those laws are fixed. That’s not to say those laws can’t flex, but we are nevertheless bound and agree to be bound by that rule of law. Now, that’s important because if we agree to that rule of law, we have to work out, is it possible for us to hardwire those ideals, those concepts, into any AI? And to my mind, I think that’s very difficult to do because any AI is created at a point in time, and at that point of time, the state of the laws in various different countries may be very different from where it might be in 10, 15 years’ time once that AI has developed by processing data and the algorithm has processed further information to create different outputs.
Speaker D: But that’s just like saying Magna Carta is no longer irrelevant, you know, because we drive around in cars. Actually, rights are rights, aren’t they? Aren’t they inalienable, whatever happens, whatever new technology comes along?
Speaker G: To a certain extent. I mean, obviously we’ve moved on a long way from the the divine right of kings, and we’ve now moved into a situation where nobody is above the law. And so it’s very important, and you would expect a lawyer to say this perhaps, that we protect that body of law, that rule of law, that adherence to that law. It is very difficult though, because what that also means is that we are not able to give away more than we have ourselves. There’s a Latin maxim, delegatus non potest delegare, you can’t delegate more than you have yourself. And I would suggest that probably extends into the creation of AIs. Is it possible for us to give to that AI greater rights than we ourselves have?
Speaker D: Haven’t we already done that?
Speaker G: We may have given some rights away, but I would say that we have never and should never give away the obligations or the responsibility that necessarily flows from those rights we may have given away.
Speaker D: Don’t the technologists need to start taking responsibility here as well, in terms of more ethical programming, more transparency about how data is used, for example, the vast bodies of data that are collected on the Internet of Things. Shouldn’t the technology industry itself really get more engaged in this?
Speaker G: To a certain extent, although going back to a prior point that I made, it is very important that we as lawyers do not try to prescribe precisely where the technologists go. It is an essential feature of innovation that we’re enabling our very best innovators to innovate, to take matters forward, to go on that path, that pioneering journey.
Speaker D: Shouldn’t they also take into account the possible consequences for humanity, for example?
Speaker G: Well, we think about the splitting of the atom, and that’s perhaps the best example of unintended consequences, perhaps. But would we say now that we should not have actually gone down that path, that we should have not allowed that scientific development and that research that led to that development to have taken place? I really struggle with that. I think that technology, science should— and it’s a very relevant and necessary part of human development— should go down its own path. That is incumbent on us as lawyers to provide the relevant checks and balances and make sure they’re put in place. The problem comes when the delta between the technical technological advancement and the legal support and response is too wide. And I think the point probably is, are we at that stage now where in the role of AI and in the world of AI, is the technological advancement starting to run away from legal support.
Speaker C: Come on, politicians, get with the program! We’ve established with the world’s top experts that we need an international legal structure. If you’re a regular listener to Password on Resonance FM, you’ll remember the European Parliament tried to get on board back in February, and MEP Mary Honeyball repeated at our London conference that we urgently need to control AI?
Speaker J: Well, I’m not sure we can stop it, but what we can do is develop an infrastructure for it itself. And what we were looking at in the report is some of the legal questions, because it was a report that went to the Legal Affairs Committee of the European Parliament, so that was its focus. The basis of which was that robots should work for us as human beings, not the other way around. So how you do that How you make sure that robots don’t harm people. What sort of legal system you need to deal with these things. But the main thing, I think, the basis of all of this was that we want to set up some kind of system and structure that allows member states of the EU to come together to cooperate. Because this is not something which will be confined to one nation state. Because this will travel. This isn’t just Britain or Germany. This is— everyone will be involved. So we therefore need international solutions.
Speaker F: I mean, there are other issues here as well, aren’t there? One of the things that I noticed in Rapporteur’s report was that there was a call that robots shouldn’t be like people, that we shouldn’t start to anthropomorphize them.
Speaker J: Yes.
Speaker F: We shouldn’t start to identify with them.
Speaker J: Why is that? I think that’s probably right, because they’re not people and we don’t want them to be people. And there are robots at the moment, particularly in Japan, that do caring work. They look after people who are ill or needs a lot of care, and you lose the human empathy. And I think it’s important that we maintain that, not only in care and illness, but across the board. We don’t want to become like robots. I think that would be absolutely disastrous. And so therefore, you need to keep this all in check and make sure that robots do certain jobs that they are very capable of doing, but they don’t take over from humankind.
Speaker C: Do you not think though that if we put any controls on science that there will be hindering scientific development? Won’t the scientific community say you’re being Luddites here? You’re trying to control us. That’s a situation that Aldous Huxley warned about in Brave New World.
Speaker J: Yeah, well, it is. I’m not sure Aldous Huxley always got it right. We’re not putting controls on science as such. I would never want to do that because I think scientific innovation is very, very important for all of us in a whole lot of ways. And all the developments that we’ve been talking about that have taken place have happened through scientific innovation. But what we also need to do is make sure that that is used in a way that is constructive and positive and works for people. And that’s, I think, where we come in. I don’t see I don’t see those two things as mutually exclusive. I think laws, if you like, legislation systems which ensure that robots work for the good of people, it’s not about controlling science, it’s about ensuring that the uses to which that scientific innovation, that technology is put, works for everyone. And that’s the important part. That’s the ethical dimension, I think.
Speaker C: And that’s the really most important point about all of this, isn’t it?
Speaker F: And in fact, this is the first time that though there have been those calls for that change to science, some people have been saying that there will be a technocracy that will develop where people will actually use bodies of experts to decide what the scientific development should follow, or there are other people who are saying that there’s scientific determinism and unfortunately we just have to accept whatever happens.
Speaker C: You’re saying no, there’s this idea now that we reject that and we actually take some control over, over where our lives are Well, I think so.
Speaker J: I don’t think we should in any way stop the scientific advance. I mean, I think the idea of groups of people who look at this and make decisions about where it goes is a useful one, and I think that’s what politicians would see themselves as doing. I think others should be involved as well, and I think the scientific community should be involved in those kinds of debates. As well, so that it’s done for the best reasons.
Speaker F: Quick final question.
Speaker C: This isn’t just about robotics, is it?
Speaker F: This is about the extension of technology. This is about the smart city too. The smart city is essentially just a great big robotic place that’s connected together, and we should have a debate about that, shouldn’t we, to find out whether we’re going to be very happy with the smart city and whether we’ll be happy living I think it’s regrettable that in the past with some technological developments such as nuclear capability, these kinds of discussions never took place and there wasn’t really public input.
Speaker J: It was all done by a few people, not only scientists, but all done very secretly. I think that’s absolutely the wrong approach and that we need to have a real public debate about this and take it at a pace which people feel comfortable in, a way that people find beneficial to them. I think that’s really important.
Speaker C: So we need algorithmic accountability, clear rules of life, better cybersecurity and privacy protection, and a safety net for millions of people who lose their jobs. I’ve also been arguing that software should be tested and licensed like any new drug or treatment before it goes on general release. And an MOT-style test for older computers and handheld devices would solve quite a lot of cybersecurity problems. Mark Deem’s got another new brainwave for our to-do list. He wants to simplify and standardize the terms and conditions that we’re supposed to read and agree to whenever we download a new app or log on to a new network. Creating perhaps a gold, silver, and bronze set of common conditions that are generally known and understood, with scope for individual companies to add their own riders.
Speaker G: There are a number of things the legal profession could do. I think we could simplify those terms. We could potentially make them more digestible, certainly shorter. I think somebody has suggested that to read all the terms and conditions we might come into contact with in our daily people’s daily lives could take not a matter of months or 1 or 2 years, but several years to actually read all of this material. What I do wonder is whether or not there is now a role for lawyers to play to get together and try to work out minimum standards of privacy, and that maybe there is a place in future that we don’t define ourselves by individual terms and conditions, for example, but we define ourselves by the way that we have added to or supplemented a certain base standard. There might be a way of simplifying it. That is a personal thought. It’s somewhere where I think we could explore going, but certainly it is in the interest of, I think, all of us that we start to address these issues.
Speaker C: So there it is. All other suggestions are gratefully received. Email futureintelligence@outlook.com if you’ve got something to say to Theresa May and co., or whoever’s in charge in the following weeks, about regulating artificial intelligence.. It came as a bit of a surprise to them because not very many of them had anything about it on their manifestos at all. And now, to put you out of your misery, the sonnet you heard earlier, called the Love Bot, was written by a human—yours truly—but the wave was produced robotically by creative Turing Test winner Marjad Ghazvini-Nerjad. Both the music tracks you heard earlier was synthesized by AIs and also won Creative Turing Test awards for the music technology group in Barcelona and for CJ Carr and Zach Zukowski respectively. My fellow humans on our Password team are Blue Buffery and Jane Wyatt, and you can find out more about law and ethics in the digital space at our website futureintelligence.com. Thanks for listening. Goodbye.
Speaker A: This program has been brought to you by Resonance 104.4 FM. If you liked what you heard and want to support our work, please make a donation at fundraiser.resonance.fm.
