PassWord, Who owns you

Speaker A: Hello, happy New Year and welcome to our first Password program of 2024 on Resonance FM. It probably won’t come as any surprise to you that this program is about the ramifications of artificial intelligence, or now as it’s known AI, the biggest story of last year and a story that will increasingly hit the headlines this year as the technology begins to build up a head of steam and starts to change our lives forever. Despite the fact that interest in AI has apparently declined by 50% in the last month, we hope to rekindle that interest because in this month’s program we concentrate on the one thing that has caused the AI phenomenon Data. And we ask the question that Big Tech would prefer we avoided: who owns our data? Without huge amounts of data, AI would not have developed at the rate it has because AI needs data to understand people and what they do. It’s for that reason that the emergence of the internet and of cloud storage has been fundamental to AI. Because it provides it with more and more data. But AI has to be able to filter through that data, identifying types of information and finding patterns in it, because in that way it can learn how to carry out particular tasks and perform them more efficiently than we can. It can find similar people, examine what they have been doing, and generate alerts. In short, it watches and learns, which is the reason why the huge number of AI applications that we are now encountering on the internet are so helpful, offering to accompany us on our searches and to help us refine what we are looking for, helping us to make pictures using DALL-E 2 or Hotpot. We are training them. What this means is that it can sort good from bad, right from wrong, identify individuals and personality types, and start looking for deeper patterns within the data. Here’s Warwick University’s Professor Irene Ng, an expert on data ownership, on how AI systems use data.

Speaker B: The problem we have when we think about data is that we don’t really have a full grasp of what data really is. And most people jump straightaway into some notion of information, when actually data is in its rawest form, it’s just bit strings of 1s and 0s. You could argue that when you push your money from one bank to another bank, you’re pushing data. We think of that as money, but it’s actually data about something you have rights over, which is a currency. So when somebody says data, people sort of splinter into millions of perspectives as to knowledge and information and the bit strings and the small bits and the large bits. And then for some reason, everybody seems to believe that there is a definition of data that we all understand, and perhaps we do, at least colloquially. So I think there is a colloquial understanding of data that means something more than just raw bit strings, and that it has some form of meaning attached to those bit strings.

Speaker A: This, according to Professor Ng, is where the understanding of data parts company with what the ownership of data is about. About. In a court case, the natural starting point for evidence is to discuss ownership. But what does that mean in an AI world? We certainly do not own the information about ourselves, but we can become the victims of its use or misuse. So if we can become damaged by the use of our data, who is accountable?

Speaker B: The problem is when you then collect it you store it, or when you process it, now this is where AI comes in because AI comes in, in its processing, we then make all these assumptions about where it was collected, where is it stored, how were they all coming together, and how are they being processed. We seem to think there’s this big data lake sloshing about with lots and lots of data and then somehow emerges some intelligence that really needs to be policed. Okay, that is one form but there’s some data that could have analyzed how much rain fell on your windscreen and made your windscreen wiper go faster. So I think if you think about data, it can be small, it can be huge, it can be lumpy, it can be very variable. When you think of its processing, it can also be at the edge, at that little bit, but it can also be done in a big data lake. And when they say, well, we should— we must police AI, and machine learning. Could we just set some boundaries, please, on what we’re talking about? This is where I have a problem in terms of trying to get people to a common set of words that actually confines and scope what the conversation is about.

Speaker A: Because very often we have some notion of the colloquialism of what we mean by data, So, according to Professor Ng, not only do we need to define data, we also need to examine what the data is being used for, how it is being used, what permissions were attached to the data, who is doing what, what their relationship to our data is, and we also need to establish what is done at what point. Only then, says Professor Ng, can we start to examine the issues that AI raises for us via our data.

Speaker B: We have a logical jump into the AI space and then all the fears of what AI can be, could be, must be, shouldn’t be, comes into that space and say, “Ah, but you see, look at the harm it has done.” Then the jump is from processing to impact and now we must do something about it. I’ll give you an example again. If you take the last number of your passport, and you change it digitally, you’re a completely different person. That’s your personal data. If you take the last number of your Fitbit steps today and change it, no one gives a toss. And those are both personal data. The thing about data is it has a few dimensions. One, on its own, It can be very potent, but it also can be very inert and very harmless. When combined, inert data that’s personal can become very potent. Potent data, when combined, could lose its meaning in some ways. So we are now confusing We confuse the medium and the message, and we then say they’re all the same thing, but they are not.

Speaker A: Though this changeability of data can confer great power on it in the blink of an eye, changing it from an inert state into a very potent force— the Chinese state, for example, is accused of running a tremendous AI-based surveillance system. It could combine your Fitbit steps and your passport number and know exactly where you’ve gone. So how do we protect against that?

Speaker B: What you’re saying is we should really try to curtail power. I believe we’ve been trying to do that for thousands of years. You don’t need data. This is basically power. It’s about power, whether it’s from data or from any other means. So you said, all right, we need some regulation, but you went from the Chinese state to— what is it exactly we’re talking about? Google? Amazon? That guy who just created an app that processes 3 bits of your data? Who are we talking about here? What are we talking about exactly that we’re policing? Are we policing data, AI, or power?

Speaker A: So It appears that what we want to do is to restrict the potential for AI to be used in ways that may not be beneficial to people. But then we start running into other issues, not only for the people who are subjected to the AI, but also for the companies that develop the technology, because there is the possibility to always blame the AI system and its developers. There is also the potential to use hindsight and develop rights for people after the event if some accident has occurred.

Speaker B: There is potential for knives to create great harm. There’s potential for guns to create great harm. There are potential for society’s machinations to create great harm. We need to have a sensible discussion on what and where harm is created and how it interacts with power to do harm, where it interacts with the inability for people to exercise their agency to stop certain things, which is related to power. But then again, on the other hand, we have to think about choice and freedom Because very often, as we know very well, even in the world without data, in the world of democracy and rights, you cannot police without destroying some freedoms. We need to talk about these freedoms and what the choices are that you are curtailing when you start to police it. But very often it’s a one-sided conversation.

Speaker A: Warwick University’s Professor Irene Ng on what our data means in an AI world. And AI has a voracious appetite for data. And the development of the metaverse is spawning ever more uses for knowing exactly who we are. Only this week, Meta, in a bid to comply with new regulations coming into force in a number of countries concerning online harms, is seeking to restrict the sort of material that children and minors under the age of 18 can access online, an issue facing many of the big tech companies. But the only way that they can do that is by knowing who you are and how old you are. And the only way that can happen is by modeling and monitoring what you are doing. If you like, creating a passport for you to enter the digital world. This appetite for knowing everything about us is leading to challenges to the technology company that they would rather avoid because they do not want to have AI’s relationship to our data exposed. The most prominent attack has come from the New York Times, which over Christmas launched legal action against OpenAI over its use of the newspaper’s online content to train itself. The New York Times claims that this is an invasion of its copyright, because to train the AI systems, the technology has to copy the database. —something the New York Times claims does not constitute fair use. And the New York Times is not alone. Several UK national newspapers are also rumored to be examining legal action against AI companies for invasion of their copyright. It’s a conflict that Big Tech wants to head off at speed, because it would rather not antagonize extremely powerful and vocal opponents able to shine a light on its training of AI systems. Principally because fears of technology companies have of a process they call disgorgement, making them give back both data and share the value of the use that tech companies have put the data to. Big Tech does not want people to be aware of what data it has been using and how it has been making money from it. Here’s US media lawyer Kevin Cassini on Big Tech’s AI worries.

Speaker C: Yeah, so it will or may come down to the definition of copy. What does it mean to actually make a copy? You and I have very base-level understanding, walking-around knowledge of what a copy is, right? It’s a, it’s a mimeograph, or it’s a reproduction. It’s essentially taking one thing and making another version of it. But when you get into the technical language, as the courts are willing to do, you can have arguments that lead to what seem like nonsensical results, but when looked at in the vacuum of that case, can make sense. And as you pointed out, in Europe, some things are advanced, some cases have advanced further than others and will advance more quickly in some regards, and you may end up with disparate results, which is again very challenging. Article 17 kind of led some of those tech companies into that as well, as far as social media posts and, and what it meant to allow things to exist online. And if they want to adopt that same standard in the States, then they can do so, but they’re not required to do so right now. So they have essentially two systems, and we may end up at that point as well. I’m not sure. The further we get down the line on all these cases without some type of uniformity, the more difficult it will be for us to come out of this on the other side with a coalescence of agreements, right? You need essentially a bunch of different types of licensing and agreements stacked on top of one another that cover everybody’s everything.

Speaker A: It’s interesting too, though, isn’t it? Because this process, as I understand it, there’s a word going around Silicon Valley which is called disgorgement. This is the fact that the AI companies, that the large tech companies tech companies will have to give up some of the ways that they have been using people’s data, because we’re not just talking about large organizations like the New York Times here. In the terms and conditions that everybody lively ticks and says, you know, go for it, includes, oh, they’ve given the right for their text messages to be mined, for their emails to be mined. How far will this go?

Speaker C: As far as disgorgement goes, you’re talking about the remedies that are available at law, at least in the states. The highest form of statutory damages that are available, or disgorgement, which would take away profits that stem from an infringement or infringing use, they’re not going to be available to everybody. In order to file suit and bring claim and receive all of the damages that you think may be available, They’re going to have to demonstrate that there is some type of willful infringement that happened whereby the companies that used their copyrighted material knew full well that what they were doing was outside the bounds of copyright law and proceeded anyway. That’s one, and that’s a difficult standard to carry. But two, they’ll have to have registered all of these works with the Copyright Office in advance of the suit.. And that’s not the case. It’s not been done. To get access to the courts and to the highest standard of, or the highest damages in the US courts, you’ve got to have gone through the process of registration fully before you bring suit. So the numbers that are going to show up in headlines, or that people will be talking about as pundits, will be large. And certainly the market cap that the companies have enjoyed is large. But disgorgement, if it applies, will not apply to every single work that has been infringed. So every plaintiff won’t be able to benefit from it either.

Speaker A: Surely the point could be made by a lot of people who blithely signed these terms and conditions that they weren’t really properly apprised of what was going on. You know, the terms and conditions of adhesion.

Speaker C: Yeah, you’re talking about contracts of adhesion whereby people, A, have no choice but to click through an agreement to get to what they want, but B, don’t know what they’re looking at. Even if they did read it, they wouldn’t understand it. And yeah, there are plenty of places, at least in the States, whereby the people who have clicked through and signed those agreements won’t be held to all the terms that are in there. One good argument would be that this wasn’t contemplated by those terms and conditions. In most of these, it will say for existing technologies and any and all future technologies, right? Well, that probably won’t be sufficient for every single case. Now, if a company wants to hide behind the fact, well, they click through and they sign the agreement, I think that they’ll run into at least one federal judge who says that’s not going to be good enough. To have given you permission for all these things. But on the other side, I also think that plaintiffs who run blindly in there and say they must not be allowed to do this under any circumstance, I don’t think that they’re going to carry the day 10 out of 10 times either. I think, as I said, I think anybody who finds themselves polarized one way or another, that this is clearly a violation of infringement and these companies must be shut down, or that this is clearly fair use and these companies must be allowed to continue. I don’t think either of those things at this current state is the proper way to think about it because we just don’t really know enough about each of these cases.

Speaker A: The US East Coast media lawyer Kevin Cassini on why we are being asked to sign massive terms and conditions and why to sign into so many applications We now need to get permission from Google. It’s a sophisticated relationship. On the one hand, the manufacturers can claim that they are augmenting us, providing an essential service to make us more efficient. But at the same time, they are intimately learning who we are, uniquely identifying us and classifying us, a process that allows big tech to then sell that insight into us to companies which want to sell things to us and governments that want to know where we are and what we’re doing. A point made by Jamika Green-Aaron, a cyber expert working for Okta, one of the world’s leading identity management companies, at a noisy technology event in Silicon Valley last October when she was talking about the concerns that people have in giving personally identifiable information to big tech like fingerprints or allowing technology to use our faces to log into our devices. The government already knows almost everything I think about you that they could possibly think of. And so if you think that they’re going to take your fingerprints and do something nefarious with them, they already have it. I mean, so the kind of the— it’s always very interesting when people bring this up to me. I’m like, you have a passport, you travel on airplanes, you go through TSA. All of these things that we have brought up, yes, they are valid concerns. You have the right to own your identity, and your fingerprint is a part of your identity. But at the same time, you are there leveraging that identity in many ways across many governments. They are aware of you and have your information and know exactly where you are most of the time. Cybersecurity expert Jameika Glean Aaron, and she should know. As well as working for Okta, Greene-Aaron is a former top employee for the US defense giant Lockheed Martin, working on space systems, and before that, she worked for the US Navy. Organizations that have an enormous vested interest in knowing that their employees are on their side and are not hackers or spies. Yet, it’s a vision of the world that terrifies human rights activists who point to the apocalyptic warnings from writers such as Orwell in his 1984 and Huxley in Brave New World, Stanley Kubrick’s 2001: A Space Odyssey, and Patrick McGoohan’s TV series A Prisoner, who point to the terrifying potential for totalitarianism already chillingly displayed by the Chinese use of technology to suppress its population. Green Aaron’s blithe assertion that we should just accept that we have lost our identities to AI echoes a statement that the Silicon Valley billionaire Scott McNeely made to me in San Francisco nearly 30 years ago. He told me that privacy is dead, get over it. But that view is now seeing a backlash for a number of reasons. Here’s Colin Levy:, a leading US legal expert on the problems that AI is now generating.

Speaker E: One of the greatest concerns that I have with respect to generative AI is its ability to replicate the voice and imagery of humans and have them seemingly behave like we would, only saying things and acting in ways that are not at all, at all ourselves. And that, I think, is going to be interesting. I, I could see there being That being perhaps one of the biggest areas of concern and of regulation going forward, because we want to make sure that we can distinguish between what’s real and what’s not, both for our own safety as well as those of others.

Speaker A: That is an interesting point, isn’t it? Because one of the things that people are talking about is data ownership. However, ownership of your physical identity is data ownership. You are— there is nothing more integral to your identity than your face or to the, you know, the perception of others. To you. That is data. And so the ability to be able to fake that is actually bringing this slightly nuanced concept to this notion of data ownership. Now, we haven’t had that before. That bit is new.

Speaker E: Yeah, no, that is new. And I think, you know, it points to the fact— to a couple of things. One is that this sort of blurring of what’s real and what’s not is continuing to be, I think, a challenge and will only likely get worse for some time to come because holograms and/or augmented reality is already here. Holograms are not necessarily that far away in some form or fashion, perhaps pretty rudimentary. So I do think that that is going to be probably one of the more low-hanging fruits, I think, for regulators, just because it’s so prevalent and more easily addressable than others in terms of what you can’t do and what you should do. But there will be others that will be, I think, more challenging. And it also points to, you know, to your point about ownership over ourselves. It’s often been said that if you’re using a product or solution and not paying for it, you’re the product, you’re the solution. And that is very much true. You know, if you think about social media, I use LinkedIn a lot. I love LinkedIn. I’m all over it, but I’m also very well aware of the fact that the more I use it, the more I’m helping LinkedIn because I’m contributing to its success by contributing data and insights. And I’m fine with that. But the same with Facebook, for example, as well. Facebook has long used our sort of data and what we post for their own purposes. And the fact that only now people are sort of objecting or paying more attention to it, I think means that they weren’t paying attention to begin with. Because if Facebook thought that— if Facebook was trying to make money off of ourselves by charging us, then it would be more clear about what their intentions were. So Ultimately, there’s, I think, a lot of unresolved issues with respect to the ownership of ourselves, frankly. But ultimately, we do own ourselves. It’s just a matter of how much we want to share ourselves with others, whether it’s other people, other tools, and so on.

Speaker A: The US legal expert Colin Levy, author of The Legal Tech Ecosystem, on the need for big tech to become more inclusive. With everyone on the development of AI. Oddly, Levy is not alone, and support for the future development of AI is now finding support within the AI community, which is pointing to the need for AI development to include everyone and to ensure that everyone shares in the proceeds, not just in terms of increasing personal effectiveness, but also financially. Here’s Triveni Gandhi, an ethical advisor for the AI company DataIQ, on the need to include more people in the data used to develop AI and on the need to broaden that data pool.

Speaker D: Well, I think that it works if you first include diverse people in the development of that process, right? And so, like, typically what we’ve seen, technology has come out of Silicon Valley, which is a very white and very male space. So the technology and the harms that come from that technology reflect that. Facial recognition tools seem very cool to the average person, right? The average Silicon Valley person. But when you ask someone who’s maybe a minority or a Black male in the US what they think of facial recognition being used by the police, they’re going to be rightfully cautious because those technologies are not built in a way that are reliable. And people have been falsely accused by facial recognition technologies and falsely arrested. It’s one thing to say, oh, we need to let— it’s wokeism to include all of these different things. But if you really want to look at it from a reliability perspective, don’t you think we would be building better models, better systems, better technology if it was more inclusive at the beginning, if it was actually able to account for the variety of differences within the world, that’s not wokeism, that’s just building better technology that can actually handle lots of different use cases. And if you really believe what the big players in tech say, which is that we want AI to serve humanity, well then you need to make sure that that AI serves all of humanity and not just the microcosm of the data you have.

Speaker A: I mean, the biggest problem with this though is that The whole history of computing, the whole history of this discipline stems from government and from the financial sector. Those are, those are the first users of computers, and those are the ones who demanded how it would be used. They’re hardly going to change that model, are they? To take your point about diversity, that if you want to make it as diverse as possible, then theoretically you should also include the criminals. And also this factor of what is the definition of a criminal. One man’s terrorist is another person’s freedom fighter. To get that sort of diversity is going to be quite difficult, isn’t it?

Speaker D: Yeah, well, and I’m not saying that you can’t have every single voice in a democracy speaking at the table, because then it’s a never-ending conversation. But are there key representative groups? Are there people who are seen as leaders within specific communities that need to be addressed within a specific situation. That seems relevant. It’s not fair to say, well, oh, it’s going to be very complicated to figure out who should sit at the table, so let’s not open up the table. Start at least opening the table. Too many—

Speaker A: I completely agree with you. Ages ago, at the turn of the century, I went around a conference of people at a database conference, and I was shocked and appalled by the people who were there and by their, what I consider to be terribly inhumane, way of dealing with people, because they weren’t seeing people as people. They were seeing them as numbers. They were seen as the extraction of data from an algorithm. They were— it’s become a cliché now to say that there’s only two industries that use the term user. That’s the illegal drugs industry and the technology industry. But these people were awful. I’d said to them, these are people that you’re making these decisions about, and they were looking at me as if I’d got three heads, which they’d evidently counted before they made that assessment. Then I was basically banned from the event for scaring them, for coming in and introducing this notion of humanity. But that’s what you have to do, because going to the point I was making about this culture that has developed computing in the first place, it is governments who’ve wanted to get into what’s going on in populations at large, and it is the financial community, the insurance companies, the bankers, etc., etc., etc., which are all looking for insights into how they can make more money and then how they can also turn that information into making more money. And until you stop it reflecting those economic parameters, then that’s the only way that you’re going to make it inclusive.

Speaker D: And I think that, like, again, this is about reframing the conversation. Being inclusive is not an antithesis to making money, right? In fact, you could very easily frame it as if you have better models that are more inclusive, you’re going to find new ways to continue driving your business growth. Like, this is the work that I do with my clients is, okay, you can still reduce harm and increase benefit or reduce harm without sacrificing benefit. It requires creativity. It requires some, some work and some push.. But it also then allows you to actually get into the best of both worlds. And I think that to the point of the government, you know, and financial institutions, we are seeing a lot of regulations on how fair lending and different practices need to be enacted by financial firms. And so there are already systems in place. The goal is really to start extending those principles out so that there are very clearly defined parameters and also extending those kinds of parameters out to non-financial situations, right? Like anyone using data and AI to drive business decisions should be thinking about these kinds of parameters. And that’s why we see things like the EU AI Act coming out, the US building the AI Bill of Rights. Singapore has AI Verify. I think governments are starting to say, okay, we need to start putting some guardrails in place. Just because this is another— it’s another technology. Look, when cars were first developed, you could make the same argument that, no, no, I don’t want you to tell me what to do and how I build my cars or how we use them because I just want to sell cars and I want to get them on the road. Yeah, but we also need safety measures, right? We need seatbelts and we need automatic braking systems and we need stoplights and things to help regulate that. And it’s no different here, right? This is another piece of technology that should be regulated. So that we can continue using it in a safe way and actually continue using it towards our own humanities goals versus just a broader sort of build it because we can.

Speaker A: Sure, but I would hazard that you— that this is much more important than the car because this is essentially— it’s the road, it’s the roadway, it’s where the car’s gone and been and where it’s going to go in the future. I mean, this is an amazing amount of technology that’s been put in the hands of a particular group at the moment, and to try to wrest it from the control of that group that is currently in control of it is going to be difficult.

Speaker D: I don’t think you need to wrest the power, right? You need to just put in some guardrails and, and requirements and safety measures. You are not going to stop people from trying to develop new and interesting things, and you shouldn’t, right? Because technology and Development and improvement is an important part of our human existence, but can we do it in a way that is oriented towards a sense of justice and fairness? Sure. And it doesn’t have to be antithetical to making money. Just because we haven’t seen people make money in an equitable way doesn’t mean it’s not possible. Maybe we can build systems that are profitable so that the yacht buyers can continue buying their yachts and also not create more harms for people.

Speaker A: Maybe if we pushed our limits and pushed our brainpower in different ways, we could do that. Jivani Gandhi, the responsible AI lead for the AI company DataIQ. So what do we do? According to Professor Ng, who we heard from at the start of the program, the answer is for everyone in the world to have a stake in AI. By owning our own data, updating it, and allowing people to use it for purposes that we know about. A process that in the technology world is known as transparency, and one that is increasingly gaining traction in the world at large.

Speaker F: Here’s Professor Ng again. I do have to say though, in our process and trajectory of getting to where we’d like to get to, The doom and gloomers can give many probable futures or possible futures that can sound very scary. I think we have to just sort of be mindful of the fact that people are sensible creatures. We are. We would generally adopt the things that are sensible.

Speaker A: Now, one of the things that you have been very, very keen in promoting is this idea of a hat.

Speaker F: What is a HAT? The HAT came from 2013 till today. Now it’s sort of grown and it’s a research ecosystem to more than 12 universities now. The HAT is the microserver, a data server. So it’s not just a dumb storage now. It’s not making self-sovereign data that’s dumb. The HAT is a full, with a storage with a server capability to put in your own edge AI. And so, if you wanted to, but it basically allows people to engage and talk data with their own server. Not just, here’s my data, go do something of it, but I actually can talk it. I can, it can be structured in a way that can communicate with different services. And that That just sort of brings a lot of not just access and control that the EU and UK is talking about, but genuine first-party rights. I have the IP to my data, not just— and I can now license it in bundles and ways that can engage with services. That is what the hack is. It’s currently, of course, that is the technology, the underlying technology technology for the self-sovereign data wallet today is the HAT. And that stands for Hub of All Things. We believe that the person should be the hub of all things.

Speaker A: So essentially what you’re saying is you’re putting together the technology that allows people to exploit their data, and it’s all housed in one place, and it’s looking after you, really. So it’s the, the hub of all things is moving from a hat to a roof.

Speaker F: What the universities did was to create the concept, create the structure, create the governance, create the stewardship, and create the rules through which data conduct should happen. And so the project talked a lot about, well, just because it’s contractually I’m able to give it, should I? Is that the right thing to do? What are the risks of harm this year? I think there are 77 harms in terms of data leaving. Even if it’s controlled by us, even though owned by us, and if it’s contractually my right, it’s like it’s contractually my right to give away my kidney, to sell my kidney in a marketplace, but is that the right thing to do? You know, we do not want repugnant markets. So, there’s one thing that’s about control and access and legally ours. There’s another thing about stewardship. Of an ecosystem that is transparent. So it’s a bit like your money. There’s a Basel Committee for banks, for supervision of banks. We gave all the rules and guidance as to how data can come in and out of HATs, in and out, and dealing with the, not just personal HATs, but organizations have their own HATs, what kind of services. So that was what the project was.

Speaker A: It’s now in commercial deployment. Professor Ng of Warwick University on the development of personal databanks that give ownership of our data back to us, allowing us to decide who knows what about us and what governments and companies can use our data for. It’s a process those championing it think will work for all concerned because you will license the use of your data in return for particular things like for example, a passport, where you will verify your identity in return for the travel document. A verification that will occur on systems that you control that will interact with, in this case, a government system to confirm that the system applying for the passport is you. It’s a remedy to the issue of data ownership that’s beginning to garner some support among lawmakers, according to Lord Jim Knight. A Labour peer involved in the All-Party Parliamentary Group on AI. According to Lord Knight, because of the role AI already has in our lives, data ownership is now an issue that must be addressed.

Speaker G: In order to develop an AI, you need to train the AI with data, and all of our data becomes valuable. And you can ask a question, you know, if we survived all this long without AI, could we do without it in future and then not have to worry about these things. I suspect that the genie is very strongly out of the bottle and that there are powerful applications for this in medicine and education and various forms of public service, and that that’s something to be welcomed. But at the same time, we’ve got to make sure that the data that data that these products are trained with is both representative so that there’s not— so you minimize AI bias because you’ve got biased dataset, but also that the protections are there for us as individuals that we’re not essentially giving away our data and then having to buy back the products that have been made with the data that we gave them.

Speaker A: That’s a really interesting point, isn’t it? Because in terms of this ownership of data, we’ve seen some cases that have come out in the US at the moment where people have been arguing AI doesn’t have the right to write patents or to own patents, but does that mean then that if our data is recombined in some way that creates some medical advance, for example, that we should have a stake in that?

Speaker G: Well, I’ve become interested in that. I’ve become interested in whether or not we can form data trusts that essentially allow us as citizens effectively to aggregate our data, to have the use of that data governed by a set of principles written into the trust that trustees then have to abide by in terms of the exploitation of that data by other people. The truth, Peter, is I guess individually our little bits of data aren’t worth that much. Collectively they’re worth a fortune, and it’s thinking more about how we leverage our collective power over these hugely powerful big tech companies.

Speaker A: It’s fascinating, isn’t it? Because this is an idea that Professor Irene Ng at Warwick University has been exploring at length. She says you’re the one who’s interested in your data. You should be the one who is allowed to hold your data and to update it, make sure that it’s relevant, etc., etc. And also, you could use an AI tool to smooth out any imbalance. The various embellishments that you might have made on your data, but then you can collectively allow it to do whatever you want it to do and obtain value from your data. Because let’s face it, if you’re not alive, then there’s no data about you, is there?

Speaker G: Well, yes and no, Peter. I mean, one of the issues that we had in Parliament with the Online Safety Act as it now is, are those bereaved families who wanted to be able to see the data, the content that their children had been consuming that was implicated in them committing suicide. And there’s a real struggle for the coroners, let alone the families, to be able to access that data. Now, if we had a view in law that our personal data is our own asset and we are effectively licensing it to tech companies to be able to use to train AIs or to do whatever they do with it, but that it’s our asset, then we could bequeath that. When we’re dead to our inheritors and they can then have access to it. You know, you can constrain that in your will, etc. But once you’re dead, your data is still there and it still has a value.

Speaker A: That just shows, as we’re seeing with the recent Horizon debacle, that we’re not up to speed with technology. We’re not up to speed with the ramifications of what’s happening.

Speaker G: No, we’re not. And yeah, the other read across to the Horizon scandal, you know, as featured in that wonderful TV drama, there’s a scene in that where the barrister is talking to Alan Bates and saying, these contracts that you’ve signed, they’re, they’re very weak in terms of their enforceability in the law because, you know, you can’t reasonably have been expected to to follow them, both parties can reasonably expect it to have followed them. Same is true with terms and conditions, I think. Tech companies rely on them in contract law as being things we’ve all signed up to and accepted. I just wonder, when tested in law, whether they do form solid contracts, because reasonably we can’t be expected to have read and understood them all as individual users. It might be possible to generate an AI that can read it for us and tell us what we want to know from it. But at the moment, I haven’t seen that application.

Speaker A: Labour peer Lord Jim Knight from the All-Party Parliamentary Group on AI. So the debate on data ownership is growing, but is it really practical? With data ownership, we also have to have the skills and desire to keep our data up to date. It only works if we are able to easily manage our data and understand what is happening. And as Lord Knight pointed out, there is a burning need for the tech companies to clearly and concisely explain what they are doing instead of hiding behind massive terms and conditions documents that can probably only be understood by a specially developed legal AI system. It’s a problem of implementation that Silicon Valley native Amy Hodler, the founder of the AI company GraphEvangelist and an AI ethicist, says that must be overcome if data ownership is to become a reality.

Speaker H: I think it’s an interesting idea. I think it could work in some scenarios. I think there has to be much more thought around the two things. One is not everybody is is capable, has the interest, and moreover the time to manage all of the data about themselves and to understand when they get— how many times have you gotten a EULA and actually read through that? So to say, oh well, you know, we, we provided some information and they clicked okay so they could get their free app, so hey, it’s all on them, I think is inappropriate actually. I think so. So I like the idea. I think there has to be a thought about implementation and who’s going to be able to do it. If you live in a rural area with terrible internet— and I do live in a rural area— and you don’t have the ability to manage that, does that mean you don’t get to have a say, or it’s too bad for, for you and your family? So that’s one aspect. Let’s look, let’s look at that and capabilities, interest, and time to manage that. And the other thing is we need to look at things like privacy and I’m about to say ethics, you almost got me, responsibility, responsible AI from a holistic standpoint. And what I mean from that is from like a network standpoint, I can do as an individual everything in air quotes right about protecting, let’s say, let’s say it’s privacy in this case or the use of my data. But if the people around me are sharing their feeds and their pictures and information that happen to include me, I have no control over that. And so there was a paper written probably about 5 or 6 years ago about herd immunity and data. And the idea— they have shown that with phone logs, if you go, I believe it’s 3 to 5 hops out of somebody looking at phone logs, you can pretty much cover the US, even if individually I haven’t given you that that ability to do that. So thinking about, I think, how AI is used and privacy and safety from a herd immunity standpoint, I think is important as well. We often put everything on the individual, and I think that’s not feasible from— not practical from a standpoint of what they, you know, can do and can understand and can have time for, but also from the herd standpoint. So I can do everything right as an individual, and if the people around me, or even let’s say 5 hops out from me, don’t, I am still at risk. So I think there needs a bigger agency and regulations looking at that kind of information.

Speaker A: Well, one of the things that Irene was suggesting was that you should have a little AI that runs on your hub of all things, and the AI looks after your interests, which in a sense is the point that you were making about assistive AI, of augmented AI, or life augmented by an AI, that you have an AI that is loyal to you, biased to you.

Speaker H: That’s an interesting idea. Implementation will be interesting. And if how the AI decides what is good for you and works in cohort to your spouse’s AI, your kid’s AI that’s interested towards them, your neighbor, your boss, who may have a very— an AI with different goals.

Speaker D: Starts to get very complicated, and that—

Speaker H: I’m not sure how that would work out.

Speaker D: It’s, again, it’s an interesting idea. I am a big proponent of system-wide considerations and trying to have common guidelines, rules that are open, transparent, and understandable.

Speaker A: The technology ethicist Amy Hodler on the important issues that data ownership will throw up for the general population. While it’s easy to be cynical about the difficulties of creating new, more inclusive systems for the 21st century and the development of the technology we will use in it, many people are arguing that it is a natural evolution of AI which they claim will begin to change the ways that the world has developed up until now. Increase participation and understanding, goes the argument, and we will all benefit, instead of keeping money and power in the hands of a privileged and informed elite that will use AI to simply increase their power. AI, goes the argument, should not only increase knowledge and insight, but in so doing will lead to new ways of creating employment due to increased understanding. It’s a message that has been embraced by the leading AI company SingularityNet, according to its chief executive, Janet Adams.

Speaker I: Oh, I’m absolutely behind people having the rights over their own data. We are shortly— 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 to Rejuve by providing their biomedical data to have complete ownership over their own data. And that is coded into NFTs. The ownership is, 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: So are people allowed to update their data to make sure that their data in your model is accurate?

Speaker I: Yes, absolutely. That is a whole fundamental precept of the way we’re developing our Rejuve self-sovereignty application and also So going forward, will be used in many other models.

Speaker A: Janet Adams, the chief executive officer of SingularityNET. But it is not simply a question of giving data back to the people, according to the academic, New York University’s Dr. Juliet Powell, author of The AI Dilemma: Seven Principles of Responsible AI. Right and wrong in the AI world, she says, is potentially very complex.

Speaker J: Siemens was dealing with health data, and in this particular example, they had the choice because of the laws in the United States where you can only hold personal data, personal medical data specifically, for a certain period of time, and then it has to be destroyed. Well, ultimately Siemens faced a dilemma of its own, which was, do we decommission this and follow this law, Or do we pollute the soil by decommissioning and get a much bigger fine? And so ultimately, the decision was made from a financial basis, from a short-term risk basis versus a long-term view. And I think that leaders are faced with these kinds of dilemmas on an ongoing basis. So when you talk about we know what’s good and we know what’s bad, in this case, What would you have chosen?

Speaker A: That’s the issue, isn’t it? I mean, I hope that I would have done the good thing that was in the interests of the planet and the population at large. That’s the issue with so much of this, isn’t it? Because going back to that Sarbanes-Oxley thing, you say, okay, should we be running a company in the interests of the shareholders and in the stock price? I mean, arguably what you could say is that it’s in your interest to manipulate people using AI so that you can make more money out of them. You could say that that is good and that is in the interest of the stockholders.

Speaker J: There’s a whole industry based on this. It’s called advertising and marketing. That’s what that industry does. And we know this and it’s perfectly legal, right?

Speaker A: So we are capitalists. But that’s one of the interesting points about this, isn’t it? Because that AI dilemma It should presumably force us to be more intelligent. A lot of people have been saying that progressively we have become better and better read as the century progresses because we’re seeing more information. Yet at this time of fake news and fake images, we’re now contending with deliberate disinformation. Now, deliberate disinformation, presumably the people who know that they’re putting that out must know. The Alex Coxes of this world must know that that isn’t right.

Speaker J: I mean, although he doesn’t, does he? I certainly can’t speak for him or for anyone else, but I do think that most people don’t necessarily see a line between good and bad as they’re crossing it. They might see it in retrospect, but at the moment, not necessarily. It’s one of the reasons why I developed something called the calculus of intentional risk. It’s something that we developed specifically for AI so that leaders can make decisions before developing the systems, as they develop them, and after they’ve deployed them. But I think that that kind of calculus is necessary for the majority of us just in our day-to-day thinking. We rely on our technologies, on AI in particular, to help us make make faster, better decisions, right? Who we’re dating, where we’re going, which job we should take. All of these things are algorithmic, at least in G8 countries. And we rely on these heavily. I mean, who we date determines who we marry, determines who we have children with, determines our entire lineage and heritage. These are huge, huge things that we’re delegating to our technologies. Without a second thought. And I think that is the scariest part for us as humans, is that risk-reward-benefit does have to be measured in more than just the moment. And most of us don’t even do that because we’ve got buttons on our phone that say buy, buy, buy, or follow this link, or tap to pay. And so we don’t go through that entire process. I interviewed this incredible woman. She’s at at Columbia Business School here in New York. Her name is Sheena Iyengar, and she’s a behavioral psychologist. And she was talking about how control is an innate being. It’s part of who we are as humans, right? We’re born with this need for control. It allows us to survive over time, right? But ultimately, real control means that you weigh the risks and the benefits. And most of us, don’t have the capacity to do this thousands of times a day. So we create these shortcuts, these mental models, which are in fact biased, that allow us to do things faster. That, in addition to our technology in which these biases are already embedded, make it that we’re actually not in control. We have the illusion of control for the majority of our decisions.

Speaker A: AI Dilemma author Dr. Juliet Powell. On why AI and the data we train it on is essential in our future lives because we have a greater interest in ourselves than the big technology companies. You’ve been listening to Password on Resonance FM with me, Peter Warren. Password is written and presented by me, Peter Warren, and produced and edited by Blue Bathory. Thanks for listening and goodbye.