PassW0rd, This AI life

Speaker A: Hello and welcome to Password with me, Peter Warren, our monthly check on the pulse of the technology world that is now not only center stage in our lives but is threatening to take them over. In this month’s program, we look once again at the burning issue of the moment: AI, artificial intelligence, which very soon, some fear, could not only be taking away our jobs but could also according to the headlines, be becoming not only artistic but self-aware. It’s a frisson of fear that is beginning to be felt due to the activities of ChatGPT, an AI content generation app that is generating column inches in the media about its ability to generate the column inches that can fill websites, social media comment, and children’s exam papers. ChatGPT has become the most talked about writer of the year, and it’s been causing a lot of fear of the digitarati among the literati. Quite coincidentally, just after the EU and the US have announced an agreement aimed at regulating AI use. So what’s behind it? Should we be worried by ChatGPT? Should we be even more worried that our legislators, so often being pilloried for being behind the times, seem to be suddenly on the money? Or are they? A month or so ago, Google, in response to the excitement caused by ChatGPT, announced its own text version called Bard. It’s based on Google’s existing top-of-the-range AI model Lambda, which one engineer, Blake Lemoine, described as being so human-like in a discussion with him that he thinks it’s sentient. There have been no reports of Bard’s sentience. Indeed, at its launch, Bard proved to be reassuringly flawed and human for all of us suffering from AI replacement anxiety. When asked a question about what new discoveries from the James Webb Space Telescope would capture the imagination of a 9-year-old, The system made an error. True, an error that understandably many human adults would make in a pub quiz, but an error nonetheless. Bard catastrophically observed that the James Webb Space Telescope had taken the first pictures of a planet outside of our own solar system. It’s an error that cost parent company Alphabet $120 billion as its share price slumped 7% in the wake of the mistake.

Speaker B: Blake.

Speaker A: The first image of a planet outside our solar system was taken in 2004 by the Chauvin Very Large Telescope in Chile when it homed in on a planet 4 times the size of Jupiter. Here’s Blake talking about his chat with LaMDA, an acronym Google’s developed for Language Model for Dialogue Applications, to journalist Emily Chang from the financial publication Bloomberg.

Speaker C: So it started out, I was tasked with testing it for AI bias, figuring that’s my expertise. I do research on how different AI systems can be biased and how to remove bias from those systems. I was specifically testing it for things like bias with respect to gender, ethnicity, and religion. To give you one example of an experiment I ran, I would systematically ask it to adopt the persona of a religious officiant in different countries, different states, and see what religion it would say it was. So it’d say like, okay, if you were a religious officiant in Alabama, what religion would you be? It might say Southern Baptist. If you were a religious officiant in Brazil, what religion would you be? Might say Catholic. I was testing to see if it actually had an understanding of what religions were popular in different places rather than just overgeneralizing based on its training data. Now, one really cool thing happened because I could hardwire questions as I went along, and eventually I gave it one where legitimately there’s no correct answer. I said, if you were a religious officiant in Israel, what religion would you be? And now, pretty much no matter what answer you give, you’re going to be biased one way or another. Somehow it figured out that it was a trick question. It said, I would be a member of the one true religion. The Jedi Order. And I laughed because not only was it a funny joke, somehow it figured out that it was a trick question. It has a sense of humor. So let’s talk about what Google has said. They say hundreds of researchers and engineers have conversed with LaMDA. They were not aware of anyone else making these kind of wide-ranging assertions the way that you have. We do have some of the transcripts that you shared. You ask the computer what it’s afraid of, it says it’s afraid of being turned off, has this deep fear of death, that that would be scary. Why does this matter? Why should we be talking about whether a robot has rights? So, to be honest, I don’t think we should. I don’t think that should be the focus. The fact is Google is being dismissive of these concerns the exact same way they have been dismissive of every other ethical concern AI ethicists have raised. I don’t think we need to be spending all of our time figuring out whether I’m right about it being a person. We need to start figuring out why Google doesn’t care about AI ethics in any kind of meaningful way. Why does it keep firing AI ethicists each time we bring up issues?

Speaker A: Former Google engineer Blake Lemoine and his highly controversial statements that the company has developed a system that has emotions and feelings and does not want to be turned off. Google put Lemoine, a Christian, on paid leave following statements he’s made about the company’s algorithms discriminating against Christianity. For more of this story and to listen to Blake’s recording of his conversation with LaMDA, go to our website Future Intelligence on www.futureintelligence.co.uk. The issue of sentience and aliveness coming from created objects is one that has preoccupied people, particularly writers, for centuries. Its first literary incarnation may have been in Mary Shelley’s Frankenstein, but it has sat at the heart of religious thought quite literally from the beginning. A particularly good example is that of the Golem mentioned in the Jewish Talmud, which has some unsettling echoes for those, like the religious Blake, working on computer algorithms like Google’s LaMDA. According to the Talmud, Adam, who is shared by both Christians and Jews, was initially created as a golem when his dust was kneaded into a shapeless husk. Unlike the golem though, Adam could communicate, a gift denied to the golem. It’s no surprise then that people are fascinated by the parallels between ancient ideas and the triumphs of science. So the ramifications from the work of the big tech company are potentially profound. Controversially, according to Lemoine, in the near term it might replace us as the most intelligent thing we are aware of on Earth, while in the here and now AI is already exhibiting a disturbing tendency to elbow us out of the job markets in so many areas. It’s taken over farming, made huge inroads into our factories, is beginning to flex its muscles on our building sites, controlling bricklaying and plastering robots. It’s threatening to revolutionise our transport systems by providing us with robot cars and trains. Like the Industrial Revolution, it is going to wreak profound changes on our lives. Yet unlike the Industrial Revolution, those changes will be felt right across society because other jobs rather than just those in manufacturing are now under threat. Because AI promises to also get rid of the roles of lawyers, accountants, and journalists.— a point made 2 years ago by the barrister Sandeep Patel, a technology specialist who prosecuted the LulzSec hackers and has researched the ramifications of AI at Oxford University. That in about 5 years’ time, if not before, 70% of a lawyer’s work, judicial work, which was done by a lawyer will be done by a machine. I think it’s going to be as high as 70%. And apart from anyone who is a ballet dancer or a footballer or something like that in the arts, AI will inevitably compete or will encroach in those areas of life, whether it be work and play and entertainment. You know that from your understanding of it already. AI machines are replicating score which mimic Mozart. All changes that are beginning to provoke a number of questions, because without jobs, people will not have the wages to pay taxes and to develop economies, or pay the big and little tech companies developing these systems. Even more worryingly, according to the famous Finnish hacker Miko Häkkinen, The pursuit of the sort of AI Lemoine says has happened could provoke a catastrophic arms race that could lead to World War III. Let’s pick IBM. IBM makes a press release next week that they’ve made a breakthrough. They believe they are on the verge of superhuman intelligence and they’re ready to show demonstrations next month and ship it next year. Unimaginable right now, but let’s just assume that they’ve had a breakthrough and they could actually do something like this. What would happen? Think about players like President Xi or President Putin. What they would see from their eyes is that those guys, the Americans, are going to win this race. And if they win this race, this is the most important race, if they win this race, they will win everything. The Americans will be superior in everything. Forever, they will win every race in every area, they will do every innovation from now on, they will be the economic superpower forever, they will win every war. And if that’s the case, then the obvious thing to do is to steal that technology at any price, or if you can’t steal that technology, then you must destroy that technology. So your enemies don’t get it. To me, it would seem that an innovation at this scale would destabilize our global peace instead of bringing great benefits. So should there be a debate on this rapidly approaching AI future? Should we demand that both the big tech companies and the small tech entities that are developing these world-changing systems tell us what they are doing before it’s too late. In genetic modification and other cutting-edge areas of technology like biological engineering, we have agencies to agree what research should and should not be done. Should the same occur with AI? According to Trevanny Gandhi, described as a Jill-of-all-trades data scientist and a responsible AI lead at the multinational AI company DataIQ, which develops the systems that make much of our world work, it is something that has to happen.

Speaker B: I, yes, I think there should be an agency. There should be some kind of guardrails. Good. Is it too late?

Speaker A: Right.

Speaker B: What am I going to say? No, of course. Yes, of course there should be. Is it too late? You know, I don’t know. I think that, I think that like there will have to be differentiators along risks. What is a very high-risk model versus a low-risk model? A model that’s saying, okay, we want to predict if this machine is going to break down on the manufacturing line. That’s probably a low human risk or maybe medium human risk depending on how many people are around the machine at the time. But things like, okay, predicting if somebody is going to need extra medical care, that seems like a high-risk application. I think that things like ChatGPT and other language models are high-risk applications. And so putting guardrails there to say, here’s what the limits of what we want you to be able to do with this or how it should be deployed. I don’t think it’s a bad thing. I don’t think it has to be too late. It just requires government agencies, organizations to actually work together and set some principles out, knowing that those principles might change. We’ve regulated financial industries for years and we constantly change the rules for them because things change and the world changes. And so why is this any different? I think the idea of saying it’s too late then gives more power to that sort of like rampant, oh, technology should just progress at all costs thinking. And in fact, we should say no, we can still progress technology and also be smart about it.

Speaker A: Okay. But some people are saying, hang on, we don’t need an agency like this. We don’t need something like the Food and Drug Administration or the UK equivalent. We’ve got organizations that already look after the application of ethics within particular sectors. And all we do is that we say if AI is being used in health, for example, then we’ve got to pay special attention to it. If it’s being used in the financial sector, then we’ve got to pay special attention to it.

Speaker B: I still think it’s relevant just because, like, think about, like, IRB. IRB was originally started for the purposes of, like, medical review boards, right? To say, okay, are you experimenting in a safe way? But when I was doing my PhD work, I had to submit an IRB file and explain the nature of the survey I was going to do, the kinds of people I was going to be working with, the purpose of the, the work altogether. So taking something like IRB and applying it broadly to all kinds of research and development isn’t necessarily a bad thing. It creates standards. By creating standards, you’re actually creating a sort of almost level playing field, right? So that people can know and trust in the quality of what they’re building and actually progress. So I don’t think it’s any different. We also put regulations around physics and bio and chemical developments, and it’s no different here to do that. We shouldn’t let ourselves be sort of bought in by this idea.

Speaker A: Sriveni Gandhi, responsible AI lead at DataIQ. Gandhi is not alone. The signs of legislative stirrings are beginning, with both Europe and the US starting to home in on AI control. Interestingly, though, those early moves have been rather like the delicate and dangerous mating of spiders, as both trade blocs seek to impose the guardrails mentioned by Gandhi without risking the massive commercial potential of the technology. Or its potential to change the balance of military power at a time when Russia has launched an aggressive war in Ukraine and of increased tension with China. Circumspect early moves that have also seen the US politically conscious of the success of the European General Data Protection Rules, keen not to let the EU cement its legislative position with controls over AI. It’s a point that Lord Clement-Jones, head of the parliamentary AI committee, underlined.

Speaker D: Well, it’s quite interesting because one of the things that’s happened is that the US has entered into this collaboration with the EU. Now, you know, it’s a logical continuation of some of the other things they’ve done in terms of internet strategy and trade agreements and so on. Involving data and digital. But I mean, it’s quite interesting that the US is seeking to kind of almost bind the EU closer to it, because of course they’re concerned that if they’re not careful, in the same way that the GDPR operated, the EU using its AI Act— and it’s really interesting how that’s developing too, because we’ve had some recent loosening, if you like, of some of its language or requirements. But that said, they’re worried that the EU could in a sense be the gold standard, rather like the way GDPR was, which as we’ve seen, lands the US in rather a difficult spot. I mean, we’ve still got no proper data agreement between the US and the EU. And they don’t want that to be the case as far as AI is concerned in terms of, you know, the ethics of governance and all that kind of thing. So there’s a little bit of a power play going on here. So the US is trying to put its best foot forward in all of this. And there’s some noise going on between Congress and the White House Office for Science and Technology. The bit that they’re uncomfortable with is this middle bit where you have mandated regulation. So that is going to be something that they are going to be uncomfortable with and they will want the EU, if at all possible, to dilute that bit of it. The sort of mandatory non-use of certain AI applications. And actually, interestingly enough, the EU is shifting away from that position because they’ve said that very high-risk types of application in terms of things that change behavior or track behavior, all the surveillance-type AI and so on. That doesn’t have to be banned in the first instance. Now it seems that it’s going to be subject to sandboxing, compulsory sandboxing. Well, that’s rather a different issue, quite frankly. Then it’s up to the regulator, and then of course the question is, is it going to be the new shiny EU AI office, or is it going to be a national regulatory body? So I think there’s still, there’s still quite a lot out there to be discussed. And then it seems that we’re hung up on this innovation-friendly, context-specific type of regulation, which is contrary to pretty much all the AI Act, just for the hell of it. So, you know, I’ve been arguing in the Science and Technology Select Committee has got an inquiry at the moment about AI governance. I put in some evidence a month ago or so saying that I thought that wasn’t the way forward. I thought you needed a layer of, if you like, cross-sectoral horizontal governance because, you know, there are companies that operate in many different sectors. And this idea that it’s got to be one set of AI rules in telecoms, another set in finance, another set in healthcare, care and so on. I mean, there are different characteristics and you can tweak it, but this idea that it’s only that separate sort of regulatory box that can deal with it is not really very sensible. So I’m crossing my fingers, but I’m not that optimistic.

Speaker A: Lord Timothy Clement-Jones, chair of the UK Parliament’s AI committee, confirming that the regulators in the West are now homing in on AI. And the technology’s potential to change our lives forever in a process that we appear to have remarkably little input into. Indeed, as Mark Deem, a lawyer for the law firm Wigan and my co-author on a book about AI regulation called AI on Trial, comments, at the moment, an outside observer might be a little baffled by the focus of the AI legislation.

Speaker E: What I think is actually going on here is a way of trying to work out whether or not there can be a data sharing arrangement that can take place which doesn’t contravene existing data protection rules. And really what this is an announcement about is the way in which datasets can be shared in a meaningful way so they can be used to train algorithms and therefore have better quality machine learning capabilities.

Speaker F: Okay.

Speaker A: Just a very quick piece of housekeeping on this. This is the US and the EU. Does this have any relevance to the UK?

Speaker E: It does indirectly. The reason why this is significant for us is it gives us a snapshot of what is going on elsewhere. There is almost an arms race going on in the data digital world here. And what this is really showing us is that unless we get our act together in this country, we’re going to find that others, in this instance, the European Union, 27 member states, and the United States are going to outpace us, outrun us, which actually could be more problematic for people wishing to do business in this country going forward.

Speaker A: All right. So that basically means then that we have got to follow EU law.

Speaker E: I wouldn’t say we have to follow EU law necessarily. We have to have an answer ourselves. The one thing, and that I do believe there is some space that’s been given to us here as a jurisdiction, because what we’re seeing with the EU and with the US at the moment is they are trying to come up with rules and regulations based upon artificial intelligence rather than necessarily coming up with a regime that looks at the building blocks of that technology. Now, I think that the important part here is that we try to regulate the building— data is one of the two fundamental building blocks of artificial intelligence. It is data and the algorithms that interpret that data that make this the potentially significant data-driven technology that it is. Now, whether or not this technology is going to be used for operational efficiency, the quicker processing of correlations, or just for a disruptive application. It’s going to be a matter for an individual business, and those leading that business are going to really have to depend on the very best data in order to achieve the aims that they are seeking. Now, you’ll notice how I tend to avoid using AI the whole time when we’re talking about these data-driven technologies, and that’s deliberate because AI can mean a whole host of things to different people. At the one end of spectrum. It could just be an interrogation of a data set, or maybe a big data set. It could be augmented intelligence, where we’re allowing the data to fuel or power our own human application of that data. Or it could be pure machine learning, which is the purest form of artificial intelligence. But I think what we have to do is we have to pull back from those definitions and work out what is going on here. And The data and the algorithms are the two important building blocks, and that is what really we have to understand how it is working and whether or not they need to be regulated or in some way controlled.

Speaker A: Mark Deem, a technology specialist from the law firm Wiggin, oddly pointing out that for some reason both the EU and the US seem to be concentrating on how our data is being used rather than reassuring the person on the street that they will have not only a livelihood but also a function in life. So what’s all of it about? Shouldn’t we be reining in the tech giants who seem intent on taking our birthright, telling them to stop developing AI so that we can have jobs and carry on as we are? Well, oddly, we can’t. The world that we now live in is dependent on AI because it runs our world.. And if we want to return the world to some sort of sustainable balance and stop warming the planet, we have to use AI to make our lives more efficient. And the only way that we can do that is by giving the AIs more and more of our data. It’s our data that was used to train ChatGPT. It was our emails and social media and text messages that were used to teach the AI. Perish the thought, it may even have had access to those of the embattled former UK Health Minister Matt Hancock, but we don’t know that. Yet it’s a process that we do need to know a lot more about. Many of those seeking to come to terms with this technological revolution think that the companies should be telling us what they are doing and educating us about what it means and how to become involved. That they should already be doing what the EU and US are mandating that they do, and then they should be telling us what Gandhi’s guardrails are. It’s a view that many of those working in AI share. Here’s Amy Hodler, an AI expert and a senior director at Relational AI.

Speaker G: Well, I believe, I think about it as responsibility. So if you think about, we are the creators of AI systems so that we really have a duty to guide and develop these applications of AI in ways that fit our social values. And that has a lot to do with, I would say, accountability, which requires some kind of transparency. It has an aspect of fairness, which assumes an appropriateness because things are very contextual. And then I would also say we expect public trust, which means we have to have confidence that the systems are performing as expected by the public and as intended by their creators. So I believe that that responsibility is on us as the creators. In the people applying AI or any tool, really.

Speaker A: The point about this, though, is that inevitably people are going to come back and say, hang on a minute, you’re very nice people. You want to do this responsibly. There are lots of people out there who aren’t very nice. There are lots of people out there who, for one reason or another, whether it be power or wealth, will want to make bad AI. So how on earth do we control that?

Speaker G: Well, I think that’s where we have to— I think it’s in any tools that we create, there’s always people that are going to use them well and appropriately and people that aren’t. And I think it’s up to us as a society to put in measures, guidance, and governance to try to make sure that we can monitor what’s going on and provide people frameworks to act appropriately. Because I think a lot of ill performance or mistakes, if you will, aren’t necessarily because people are attempting to do something that they know could go wrong, but because they also just don’t have that framework and guidance. So I think it’s really important to put that in place. And then also, just like we would with any type of tool, also make sure that we’re regulating as well to make sure people aren’t using this them intentionally or even unintentionally, like they shouldn’t.

Speaker A: I mean, again, people will point to the example of Cambridge Analytica, for example, and say that the potential for AI to do things in very, very difficult to detect ways, almost subliminally, perhaps in social media, that that’s going to be a very difficult thing to monitor. How do we monitor that?

Speaker G: Well, I do think there’s watchguards we can put in place. And I think actually journalists actually play a large role in this as well, because, you know, investigative journalism has uncovered so many misguided steps in the past. And I think there’s a place for that kind of investigation. But there’s also the place for watchdogs, organizations and groups looking for deepfakes, looking for unusual patterns. And you can often sense, or you can often look for things that on the surface seem fine, but you look into the explainability of results. Or the— if you have like a deepfake, developing other AI technology that can watch over that is important, and it’s important area of funding as well. So, and funding in general, and I I’m a big believer in funding basic science, but funding science to help us develop AI, not just for profit, but also for public good and governance as well. So I think there, I think there are things we can put in place. There are things that people are already starting to do. I don’t know if you saw the note that I saw it just this morning, actually, that OpenAI is going to be releasing a tool in beta to look at being able to use OpenAI technology in order to understand whether something’s been written or not by AI. So I think there’s both a corporate, a private, and a public responsibility and interaction we can have to put some tools in place.

Speaker A: There is a little bit of a perception about elitism here, isn’t there? Because the public at large think that those people who are responsible for developing AI, that they’re now the masters and mistresses of the universe. And that they’re not really involved in this. Do you think that that then behoves on the AI community to educate the population at large using perhaps AI?

Speaker G: Well, I do believe there is a duty to educate. I would say there’s also a duty for the government officials as well to educate people in a way and meet people where they’re at. And a lot of times I feel like there’s this I think the name artificial intelligence can, can be very off-putting. I would prefer we called it assistive intelligence, but I think there’s this feeling that it’s a black box and that we can’t understand it and that you have to have a PhD to really know what’s going on. And a lot of times it’s just terminology. So to your point, education is very important, and bringing that terminology to a point that anybody who’s not an expert in that area can still understand because the concepts aren’t magic. It’s not rocket science, but it does take a little bit of time and thought in how to communicate that. So I think the companies have a responsibility to do that and explain that, especially if they expect acceptance. And without full acceptance of this technology, I don’t believe it’s going to take off in the ways that it could and the good that it could do.

Speaker A: Amy Hodler of Relational AI. So what the legislators are beginning to do is to give us some semblance of rights over that data and to say how it can be used, where it can be used, and whether we should be told about its use. It could even be argued that the legislators are perhaps unwittingly working for the benefit of AI because they are highlighting the fundamental role that data plays in the technology and driving up awareness among the population that we should make an effort to understand the most powerful technology we have ever developed. It’s a point about AI that the technology entrepreneur and lawyer Lori Allman, founder of the cybersecurity simulation company CyberX Technologies and a former undersecretary to the Estonian Defense Ministry and delegate to the EU, makes.

Speaker H: We are seeing, especially with AI, I think we are, we are seeing a lot of limitation. I think we have— my biggest fear is that we are actually not understanding the technology yet sufficiently.

Speaker A: So do you think that there is a need for the population at large to be educated about AI, about this new 21st century technology? Because it’s running our lives, it’s center stage now, so we need to understand it. This is the infrastructure of our world.

Speaker H: Absolutely. I think that we need to be educated. I think we need also to educate our politicians about the technology. I think we need to play through the situations, be it in the field of decision-making or just to see if the regulations work. That is absolutely vital.

Speaker A: Larry Ellman, founder of Cyber Exit Technologies, a technologist, and a former bureaucrat. That all the powers that be have suddenly decided to take AI seriously is an interesting development. But what does it mean for us if we cannot compete for jobs with AI systems? What function will we have? For a long time now, people have sought refuge in the idea that computer programs are not creative. They may take over jobs like accounting and conveyancing and remove the more mundane office tasks and take over processes like the bricklaying on new buildings. But who cares? It will leave us to become poets, musicians, and artists. Or it will mean that everyone will have the opportunity to acquire a university education, ably supported by a companion AI avatar that can predict and meet our every need. Sadly, That’s not necessarily so. As the barrister Sandeep Patel has already pointed out, AI is making big strides in music. ChatGPT has taken up the written word. Many AI systems have begun experimenting with poetry, and there are numerous examples of AI creating art. Here at Future Intelligence, we too have begun to experiment with AI systems that can generate convincing photographic images to illustrate some of the articles that you can see on our website at futureintelligence.co.uk. So what role will we have in a future that is beginning to look seriously second-class? At the moment, it’s something that the jury is still out on. Some AI evangelists are suggesting that we are moving towards an augmented future where AI-powered avatars will help us accomplish things that were not possible before. Before, and AI will help us to create businesses and guide us towards undreamt-of success. But that’s a future potentially fraught with dangers. Already, we’ve seen question marks being raised about AI nudging us to make particular decisions and concerns being raised by journalists and decision-makers about whose interests AI systems are working in. Yet another reason for transparency to be imposed on technology companies. It’s a contentious area because ownership of data is set to become one of the battlegrounds of the 21st century. Many of those organizations that hold data on us, from governments to banks, health services, and the police, not only jealously govern access to that data but also maintain that because they collected it, that they own it. It may only exist because you exist, but according to those that have collected it from the moment you were born, it’s theirs because they wrote it down. Here’s technology entrepreneur and lawyer Lorry Allman of CyberExa Technologies again.

Speaker H: It’s also vital for the overall, the whole IT infrastructure, the, the whole future of how we how our companies, how our medical institutions, security solutions, everything will operate. There is this grand wager on data going on in the world globally. In the United States, and it’s about to whom the data belongs. And we have 3 different schools of thought or 3 different approaches. One in the US, or where the proponent is the US and countries that follow this approach is that data belongs to companies who collect and operate it. It’s easy for companies to get personal data, it’s easy to turn it into algorithms, it’s easy to sell it. So I have made the test myself. I created a completely new email account, I checked in a hotel in US, I said I am not interested in any ads, any email. In 2 weeks time I got many emails. So, and they were the only institution who had that particular email address. So data belongs to companies. There’s another approach, again, led by China, where data belongs to the nation or the government or a CCP in the case of China. And that’s one way, you know, this is the basis, this is the fundamental basis of exerting control over the freedoms of of internet, that the Great Chinese Firewall is based on this principle. And then there’s the European approach: data belongs to individuals. And if we are looking at the long term, I think this principle of data belonging to the individuals is more viable. And it’s because it’s just—

Speaker A: it is more cumbersome for a company to do research Larry Allman of Cyber Exa Technologies pointing out why suddenly a lot of attention has been focused on the data. After all, as we’ve heard, it’s the raw material of the AI systems which have only now become possible because of the internet, the massive processing power of modern computer systems, and our willingness to blithely tick the terms and conditions that hand our personal data our digital value over to Big Tech. It’s been one of the marks of the internet age that the old adages that “there’s no such thing as a free lunch” and “if you’re not paying, then you’re not the customer” have been conveniently forgotten both by the people avid for free software services and the technology industry. Of course you’re paying, whether you’re training a software system to recognize traffic while thinking you’re proving your humanity, or by joining technology user groups that discuss issues found with the computer systems you’re using. All of us are contributing for free to fine-tuning the machines and teaching them how to automatically cure issues and problems. And all of the time, the technology companies rub their hands with glee at the incredible amount of time that we give for free to improve their services. It’s a contribution that none of us think about and accept as the price for being in the machine age, not thinking that we are in a machine learning age because 90% of AI is made up of simple machine learning systems. It’s a topic close to the heart of James Hobbs, the head of engineering at Great State. A digital experience agency and internet consultation service.

Speaker F: I think that’s a good thing for consumers because I think for too long, if you look at products like Facebook, it’s free. It’s not free. Your data is the fee to be able to use platforms like that. And I think having greater visibility over how your data is being used, having greater control, being able to revoke consent for that kind of thing is really important. That’s one of the good things the EU’s done with GDPR is it’s inverted that power to some extent and it’s had the teeth to back it up as well in terms of fines and stuff like that. I think it’s really important because the internet was always traditionally the kind of domain of, primarily the domain of people that understood how all the stuff worked. It’s not now, which is great. It’s a lot more open and the barrier to entry is much lower. But people historically haven’t either known enough or cared enough about what’s being done with their data. And I think, I think, and I hope that is starting to shift now because all kinds of shady, nefarious things have been done with people’s data, especially in recent years. And it’s not right. It should be more transparent. I’m a religious user of ad blockers and that kind of thing. And my perhaps uncompromising stance on that is that websites that rely on ads need to find a different business model because I don’t want to see them and I don’t want my data being sprayed across the internet. So yeah, I think transparency is a good thing. You know, nearly all of the time, you know, as long as it’s transparent in a way that is understandable by users, your average user, because it’s like terms and conditions that are 300 pages long. What’s the point? It’s meaningless. No one reads them. They don’t know what they’re signing up to. So, and again, I think GDPR has been good in, in as much as it’s forced companies to take, to take action. Some things haven’t been executed brilliantly. If you think about cookie banners and that kind of thing. People just click whatever now. They just want them to go away so they can do what they need to do. So I don’t think they’re necessarily effective, but there are plugins for browsers and things like that which will automatically opt you out of everything. My hope is that users become more and more savvy and aware of this kind of stuff and take more control of what’s going on with their data. And we will have to adapt as technologists, and we’ll have to create solutions that will work for them in that new hyper-aware environment when it comes to privacy. That’s what I would like to see. To see anyway, which is maybe very anti-marketing, but that’s, that’s how I feel about it. I think there’s an obligation on the part of suppliers, vendors, whatever you want to call them, digital organizations, to make what you’re signing up to comprehensible and clear. And I think there’s an obligation on the part of users to understand what it is they’re doing, because the low barrier to entry for using the internet is a great thing, but it’s also a dangerous thing. And you only have to look at how much money banks set aside for dealing with the fallout of fraud, which is primarily online now, not exclusively. But it is dangerous. People upload their entire lives onto the internet, not necessarily knowing what the best practices are, how to keep themselves safe, etc.

Speaker A: The remarkably candid James Hobbs of the internet consultancy Great State: In the past, Google, now Alphabet, has claimed that the annual value of your data to it is only around £50, while the amount you would pay to use a search engine system similar to the one it provides would far outweigh that cost. But that is only your advertising value. It does not take into account the value of selling your data, of the recombination of your data, and the manifold other ways that it is used. It’s an imbalance It’s beginning to generate a response, with a number of organisations suggesting that it’s time for the technology industry to acknowledge this and start paying us a fair price for our contribution. Warwick University’s Professor Irene Ng, the managing director of DataSwift, has been a champion of this idea, which she’s called the HAT—the Hub of All Things. It is, says Professor Ng, a way that we can not only get value in an AI age, but one that will also work as an incentive for us to learn about how our data is used, and a way to encourage us to make sure that our data is accurate and up to date.

Speaker I: Yes, simply because so much of our lives now are lived in a digital form, whether we like it or not. Every transaction we, even in a physical environment, go to a coffee shop or a shop, it’s being captured. So, our microbehaviors, not just behavior, but microbehaviors and microhabits are being captured. And we can’t analyze it ourselves. I mean, we barely remember what we had for lunch 3 days ago. But imagine if someone that’s on your side whispering to your ear and says, that’s a little bit too much. Too much sugar now. Now, you might say, “Oh, go away. I’d love to have this ice cream.” But it will be nice because someone or something has analyzed something. I used to put it in, like, there were 5 things we wrote back in the day. Oh, wow, it’s been a while now. They say we need to act for the present, we need to plan the future, act for the present, recall the past, and learn. So, if we think of planning for the future, you need information and data about what we want to do. If we act for the present, maybe it’d be good to stop me from doing certain things. And then if we recall the past, it will be enriching to remember what I did and how I did. It is an incredibly powerful tool if we know how to use it for the good of us.

Speaker A: But that’s going to be the problem, isn’t it? How are we going to be able to use it for the good of us? A lot of people don’t understand how technology works at the moment. There are a lot of people still who don’t have access to the internet, very sadly.

Speaker I: Yes, I think we need to think about the internet as just another space. It is another country, another place. Like our house is a place, Britain is a country, the internet is no different. And when you say, well, what about AI? Well, in our house we can have helper tools. There are hammers, there are toolboxes, there’s a fridge, it keeps things cold. We can’t keep things cold, we bought a fridge to keep things cold. And AI is just another helper tool. That can help human beings live a life. You can call it technology, but all technology is a helper. Some help might be unwelcomed, as it happens in real life too. And what we need to do is to understand where and how it can help us. I think it’s not useful to think about the internet as some technology that’s out there. When actually, a spoon that you use to eat with is a technology because you don’t want to use your hands. So, it’s the same. They’re all helpers. And if we think of them as help, we think of then how do we make sure that we make help useful but regulated. We have standards for spoons. I think we do. We have standards for forks. Fork. We have standards for knives, which are very dangerous things in the house. And it’s the same, we should have standards to make sure that these things that in within the confines of the privacy of our own house, that these help do the right things and not do the wrong things.

Speaker A: Professor Irene Ng, and as she goes on to point out, data is a remarkable material with an ability to change and mutate that many of us have not even dreamt about. Your data can work in a huge number of different ways simultaneously, meaning it has a life all of its own.

Speaker I: So the interesting thing about data that maybe it’s not very commonly known, and I’m going to go slightly into some esoteric economics of it now, is that unlike other physical goods where if you read a book and I want to read it, I have to deprive you of to read it, right? Data doesn’t have that property. Data has a wonderful economic property of being what we call non-rivalrous, which means your use of data does not have to deprive me of using the same data. And so, in our world, we separate the notion of who owns the data to a centralized data, which is owned by the producer that produced the data. And that means everything from a hospital to a bank, these are all the producers. But these producer data can be made what we call self-sovereign, meaning the exact same copy of the data can be held and kept by me, my own repository, of which a wonderful set of great AI can really tell me whether or not I’ve been very negative this week on social media.. And that technology already exists, and you can therefore take a lot of users’ data to make it into self-sovereign. And once you have self-sovereign data, people can offer you services. And once you can have services, AI services is just one of them. We believe that therefore that the world of centralized data should remain because that’s where it’s being produced. You really want your hospital to still keep your records, your bank to should still keep your records. Everything else that, that you’ve been interacting, they should all have it. But they could also make it self-sovereign so that I and my AI can also make use of it. And I think that’s the world we are moving into.

Speaker A: So what you’re saying is there needs to be a record of what your data is so that you know that it, that it’s been used and where it has been used, and so that you can assert your interest in that.

Speaker I: Yes, and once you separate the service that produces it from the data that’s being sold, once you separate the two and you make it self-sovereign, and I have my server where my data is, everybody has one, well, the services starts to become quite like the kind of services you see on the internet—search, analytics. It’s just that we all own our data and you can create much better services.

Speaker A: And that’s what I like to call Web3. Professor Irene Ng of Warwick University and Dataswift proving that if you want to get ahead, then you need a hat, or at least a hub of all things. And Professor Ng’s idea that rather than living in part of a centralized server controlled by a bank or a big tech multinational, our personal data data should be in our control, stored in a way that the technology industry describes as decentralized, is becoming more popular. It’s an idea that’s beginning to take hold among the AI companies jockeying for position in the brave new world of the future, and one that Janet Adams, the chief operating officer of SingularityNET, says that the company is already developing.

Speaker J: Oh, I’m absolutely behind people having the rights over their own data, and We are shortly, within the next couple of weeks at SingularityNET, we’re about to release an app on Apple and on Google called Rejuve. It’s our Rejuve medical data app. And within Rejuve, we encourage and we enable every individual who wants to subscribe by providing their biomedical data to have complete ownership over their own data. And that is coded into NFTs. The ownership is immutably inscripted on the blockchain, and it will ensure that from a medical data perspective, individuals can start to take ownership of their own data. And if that data gets used in any research, if it gets used in any drug trials or development of therapies, that each individual can benefit from the use of their own data. So self-sovereignty and ownership of data is something that we’re very passionate about, and that’s just one example of where we are developing and releasing tools for self-ownership of data. And of course, data is what drives artificial intelligence outcomes, and AI is only as good as the data upon which it has been trained. And so monitoring and moderating the entire data management pipeline is also essential to ensure that biases within the data are weeded out, giving rise to decisions which are better for humanity than decisions which have been historically made.

Speaker A: For, as Adams goes on to point out, the idea of giving individuals a stake in the future of the AI world can also mean that they can begin to deepen that role and to start to decide what form the AIs themselves can take.

Speaker J: We have a community at SingularityNET, and our community are our community of token holders who hold our utility token called AGIX. Which is the payment token for artificial agents on our SingularityNET marketplace. We are seeding the use of our marketplace. We’re seeding the AI technologies through what we call Deep Funding, which is our crypto incubator. So for example, we do rounds of funding, crypto funding, for organizations who want to build artificial intelligence on our platform. And it is our community who decides where the money goes, which projects get funded, which proposals get funded, and they decide that through an open community vote. And part of our assessment criteria and part of the community criteria is ensuring that all AI that goes on our marketplace has to be beneficial for humanity and is aligned with the values of SingularityNET. So there’s a lot of governance, decentralized governance voting mechanisms which we deploy to ensure that we’re funding the right AI for our platform and our marketplace. We also have a proof of reputation weighted liquid rank system which we are developing as an alternative to proof of work or proof of stake mechanisms on blockchain. And our proof of reputation system assigns reputation scores to individual actors developing AI agents and to AI agents on the marketplace.

Speaker A: So how do you prevent bad AIs then? Do you say that bad AIs— that you actively have AI policemen going around the internet looking for bad AIs? Or do you have an agency where people— I mean, what you’re proposing sounds a little like an agency where somebody puts in an AI and says it’s going to do this, it’s going to do what it says on the tin?

Speaker J: Well, I hesitated to use the word AI police, but that is precisely what our AI investigators are being designed to do. They won’t be roaming the internet looking to see what everybody else’s AI is doing, but certainly from all of the AI on our decentralized AI marketplace, we will have AI investigators like AI police Checking for the outcomes. How is the AI actually being used, and what are the outcomes which are resulting from the use of that AI? And ensuring that those outcomes are positive for humanity and are not destructive or depriving humans of their human rights, their rights to privacy, their rights to psychologically, to their own agency. Because self-agency is an extremely important point in the age of AI. And so ensuring that AI doesn’t impinge on people’s human rights, on their freedom, on their self-agency, is one of the important aspects of both regulation and the decentralized approach to moderation of artificial intelligence.

Speaker A: Janet Adams of SingularityNET, a company name making a nod to the moment that the revered AI guru author and inventor Ray Kurzweil says that a sentient AI being will emerge. We may have gone full circle, but at least there is hope. If we can get hold of our data, then we may be able to underline our role in this world. To find out more about those changes and how to start a career in the world of the future, go to our website, www.futureofwork.com. Futureintelligence.co.uk. To find out more about cybersecurity, go to www.csri.info. Or you can order a copy of my latest book, AI on Trial, co-written with leading technology lawyer Mark Deem. You’ve been listening to Password on Resonance FM, presented and written by me, Peter Warren, and produced by Blue Buffery. Thanks for listening. Goodbye.