PassW0rd – 8 December 2021 (Getting Into AI)

Speaker A: This program is brought to you by Resonance FM. If you like what you hear, please support our work by making a donation at resonancefm.com/donate.

Speaker B: Hello and welcome to Password on Resonance FM with me, Peter Warren. In today’s show, we’re looking at artificial intelligence, what it means to the 21st century, and how you can take part in what is being called the most powerful force in history. So why is artificial intelligence, AI, going to be so powerful, and where on earth is it? For most of us walking along the rain-lashed winter high streets, apart from the missing shops and banks, Everything looks much the same as it has been for the past 40 years, particularly in the Suffolk village that I live in. The village shop now stays open until 11 PM, a revolution for such a rural area. But apart from that, on the surface, there’s no change, except that a mobile phone lies in every hand that ranges the streets, a gateway to the wider world via the internet and the road to the technology superhighway that is made possible by the AI that sits behind our lives and is developing to absorb much of the world that we used to know as work. The missing shops and banks are the clue. They are now in the mobile phones in our hands. Professor Jim Hendler, author of Social Machines: The Coming Collision of Artificial Intelligence, Social Networks, and Humanity, is one of the US’s leading data scientists. Scientists with a CV as long as your arm. The director of the Institute for Data Exploration and Applications and the Tetherless World Professor of Computer, Web, and Cognitive Sciences at Rensselaer Polytechnic Institute. He is the acting director of the RPI-IBM Artificial Intelligence Research Collaboration and serves as a member of the board of the UK’s Charitable Web Science Trust. He is also the former chief scientist of the Information Systems Office at the US Defense Advanced Research Projects Agency. So what is this AI that has already permeated our lives and, according to Professor Hendler, has not even got started yet? Well, part of the problem, he says, when talking about AI is coming up with a definition of actually what it is.

Speaker C: Jim, AI, so much fuss about AI at the moment. A lot of the fuss seems to be caused by issues of definition, and it’s almost as though it’s the use of the word intelligence that is beginning to faze people. But what is AI?

Speaker D: So, you know, I have spent 40 years of my life working in the field, and that’s a question we’ve never come up with a great answer to. In a book I wrote not that long ago called Social Machines, talking about how machines were coming into our social space, we went back to a very old definition of AI, which is AI is what computers can’t do yet. But that’s not really a very accurate question. What it really is, is AI is about computers doing things that we have traditionally associated with what people do. The problem, of course, is that’s sort of a moving target. So once upon a time, doing a weaving loom was a very human thing, and then automation came along and made it possible to automate some of that. And then now it’s almost all automated, but is that AI? Not really. So we tend to say it has to do with what people do that requires some kind of cognitive or mental activity. And those are very hard to define. They also are causing a lot of the problems for the field because there’s an expression called suitcase words that, depending who you ask, was introduced by either Marvin Minsky or Rod Brooks, two of the very superstars of the AI field. Which says, you know, when you take a word like learning or intelligent or some of these other things, it carries with it a lot of assumptions as humans, but for the computer, we don’t get those assumptions. So we use those terms in a technical sense, but when other people hear them, they hear them in a very non-technical sense. And so that’s part of why you have this question where people are saying, well, what do we mean by intelligence? This or that is because, you know, if I told you somebody is a brilliant surgeon and he’s super intelligent, but then told you, you know, he can’t tie his shoes, you’d say, I don’t understand, right? Is something wrong with his hands or something? But with a computer, a brilliant AI computer doctor diagnostician wouldn’t even know what it means to tie its shoes, wouldn’t even know what shoes are. So we have many assumptions we make as humans that unless we know more, we think maybe the computer, when we say those same things, that’s what’s going to happen.

Speaker C: So in a sense, what we’re talking about is the fact that we’re beginning to use more advanced automation, more advanced processes for controlling that automation, and because of that, those advanced processes are coming more and more and more into our lives.

Speaker D: Exactly. And, you know, again, in the book I was talking about, what we’re really talking about is AI coming into our social spaces. So, what it’s doing with respect to our social networks, what it’s doing with respect to communication in the world, all this stuff on misinformation and disinformation, both from the point of view of creating it versus fighting it. We’re looking at a lot of things, facial recognition, which has both very positive potential, but also some very scary potentials as police use it. You know, without a human in the loop, we’ve seen people arrested because the system said they look like somebody, and yet when you look at the picture, you’d say they don’t look anything like that, but they’ve spent the night in jail. That’s not a good thing. And on and on. So again, we’re looking at a world where there’s more and more of this automated capability, and the question becomes where do we want to keep humans in the loop? How do we keep humans in the loop? And also, how do we decide where the human is better and where the computer is better and how to trade those off? So we’ve seen experiments where a doctor using an AI system, we’ve seen experiments where they significantly outperform the doctor by themselves, but we’ve also seen other systems where it significantly underperforms the doctor by themselves because the doctor puts too much trust in the AI system. So again, how do we know where these boundaries are? And this is a lot of what makes AI both exciting but at the same time complicated. And so some of the things we’re doing with AI now, protein predictions and things like that, are things where the scale and the magnitude give us tremendous power and don’t really conflict with things where we’re worried about humans, ’cause the humans still have to take that information and apply it. But other kinds of things like autonomous driving and things like that, we really are going to have to, as society, come to grips with how much do we want to trust the machines, how do we build that kind of trust, how do we know when decisions are being made, how to monitor those decisions.

Speaker E: When some of these things are happening very quickly, in milliseconds, you can’t ask a human to do this.

Speaker B: Professor Hendler speaking on behalf of the Association of Computer Machinery’s US Technology Policy Committee about the uneasy world that AI is beginning to evolve. It’s no accident that Professor Hendler mentioned the handloom weavers at one point, one of the wealthiest sections of the pre-industrial revolution working class and celebrated in George Eliot’s novel Silas Marner. Handloom weavers were wiped out by the steam-driven looms of the UK’s industrial northern powerhouses, with the result that they became the bedrock of the machine-smashing Luddite movement that rebelled against the loss of their jobs and income. According to AI’s detractors, the same will happen to huge numbers of workers in the 21st century because of the technology’s ability to learn. That AI is learning those tasks from us all of the time is a point made by Sheffield University’s Professor of Search and Analytics, Paul Clough, who is also head of the business consultancy Peak Indicators.

Speaker F: That’s right, and I think when we think about the intelligence that a machine has today, with the likes of machine learning and so on, even the best examples of the algorithms like that Google are using, Microsoft and so on, that they’re all being trained on stuff that’s produced by people. So the intelligence is kind of being fed into the systems. They’re not necessarily— the algorithms aren’t coming up with the intelligence themselves. Now, it’s not to say that there aren’t incredibly impressive algorithms out there that are developing and pushing the boundaries, but I think they’re more in the research arena as opposed to maybe in production and in general use.

Speaker C: Right. So this The machine learning is essentially taking tasks that are actually quite mundane and tasks that are actually repetitive, tasks that we would consider, or we use the word boring to talk about. It’s taking those away.

Speaker F: I think yes, in one respect, although not always. So AI is being used for automation and could be used to automate kind of boring and repetitive tasks. However, you could also use the likes of machine learning for things like forecasting, where you might have volumes and volumes of data and really complex data that would be too much for a human to try and process. Therefore, the machine needs to process it on its own. So I think there’s a mixture of some AI tasks are quite mundane and boring. It’s maybe the low-hanging fruit, the ones that can free people up to do more exciting things. But there are other types of machine learning problems which are far more complex. So imagine the the machine learning algorithm that’s trying to distinguish— I know it’s got an image and you’re trying to recognize what objects are in the image, trying to train it up. It’s a task that’s quite easy for somebody to do, but you write a set of rules to try and identify objects in an image, it becomes really hard. So if you can train a machine or a machine can learn on its own to come up with a way of being able to distinguish between different objects and so on, yeah, that’s amazingly powerful.

Speaker C: It is, isn’t it? My son has a border collie. The other day— the border collie likes going outside because it’s a border collie and it likes running up and down hills lots, and it’s been studying people before they go out for a walk. And one of the people who takes it for a walk smokes, unfortunately, and the border collie went into her handbag picked up her packet of cigarettes, brought them to her, and put them onto her lap. So the border collie has obviously learned that behavior, hasn’t it? But that’s a completely different sort of learning to the learning that you’re talking about, because you’re— well, perhaps it’s not. Perhaps it’s going through a massive data and finding something that is significant in it. Is that a similar process?

Speaker F: Yeah, so there are different types of learning, and then one form of learning is a little bit like what you’ve described there. Which should be what you might call reinforcement learning. So where the machine is learning from experience. Other types of learning can be things like supervised learning, where you have a set of historical examples which you label. So maybe your data could be whether somebody defaults on a loan or not, and you want to build an algorithm that’s able to make those predictions in the future if you’re given a new person and their kind of history. So it’s supervised in the sense that you give it some training examples. So here’s the past data, here’s what the outcome is, and you can train up the machine that’s effectively able to learn a mapping between the inputs and the output. That’s what we might call a more supervised approach, what you might call reinforcement learning or learning by experience type of approach.

Speaker C: So everybody’s saying that this is the future. Everybody’s saying that the cars will be driving themselves, that large parts of our lives will be automated for want of a different word, or be taken over by machines. Is that right?

Speaker B: I don’t know.

Speaker F: I mean, I think, I think AI is certainly going to make big advances. I think it’s going to automate a lot of what we do. I think it’s going to be entangled within our life as it already is. I mean, you know, everything we touch probably has some AI element somewhere. Is it going to take over? Is it going to dominate? So it’s interesting. Even so, Elon Musk, I remember him giving this quote. And he said AI is a fundamental risk to the existence of human civilization. Now that’s somebody who’s very immersed in AI and building kind of, you know, autonomous vehicles and that. So I think there’s a lot of concern about the use of AI. I think there’s a lot of concern about irresponsible use. I think there’s a lot of concern about, you know, regulation or lack of regulation. So I think what may happen is that AI will be embedded even more in our lives and in our industries and in our businesses. But I think alongside that, we are going to get a lot more regulation possibly than we have at the moment. So it’s not just going to be things running away from us.

Speaker B: That was Sheffield University’s Professor Paul Clough. You’re listening to Password on Resonance FM with me, Peter Warren, where we’re discussing the impact of AI. After this, you can hear DJ Ritu with A World in London. So, will the same happen to the jobs wiped out by the Industrial Revolution with AI? 10 years ago, Oxford University research stated that creative and spiritual jobs will be safe, but millions will lose out to computerisation in 2 years’ time in 2024. The research, led by Oxford Associate Professor of Machine Learning Dr. Mike Osborne, predicted that humanoid robots like nursing robots will be looking after old people, but most of the jobs lost will in future be undertaken by artificial intelligence embedded in a wide variety of machines that look nothing like humans. For example, Dr. Osborne said that driverless cars will replace taxi drivers. He warned that the United States has even more jobs at risk than the UK. 47% of Americans, almost half of all workers, will find their jobs no longer exist in the next 20 years. Most at risk are telesales agents, insurance salespeople, factory workers, and cleaners. Those involved in high-tech computer or engineering-based roles will need to demonstrate creativity, for example, by making and designing products. The ideal combination for a future-proof job, according to the researchers, is mathematical skill with artistic flair. It’s therefore rather ironic that tutoring on the brink of 2022 In December 2021, that I received a press release stating the following: The new mobility service called Fetch represents the first step towards fully autonomous vehicles operating on UK roads. The service enables users to summon an unmanned vehicle via an app, which is driven by a remote driver to their location. The customer then drives to their destination before a remote vehicle operator takes over and pilots the car back to base or to the next user. The message spelt out in the Oxford research does not seem to have been missed by the AI and cryptocurrency company SingularityNET. The company name itself a nod to The Singularity is Near, a transhumanism and AI book by the renowned futurist and former Google chief scientist Ray Kurzweil. According to Janet Adams, a geologist and former banker who retrained in AI and is SingularityNET’s Chief Operating Officer, AI cannot be stopped because it’s necessary to the future development of the world.

Speaker A: But when we look at Africa, the potential for AI to really lift the economy in Africa is absolutely enormous across so many different areas of of supply chain management, of agritech, of systematization of natural resource planning, use, management. Africa is a really interesting continent, and we have our offices in Ethiopia, as do Cardano, our partners in Ethiopia. And we’re currently rolling out AGI, artificial general intelligence, research centers across Africa combined with Cardano. So I think One of the huge and major potentials that AI has for Africa is to help with machine translation, language translation, because there are so many different dialects in Africa that bringing in advanced machine translation for the smaller and less well-known dialects could completely transform and uplift education across the continent. So they may not need elderly healthcare robots right now in Africa, but there’s definitely huge potential for profound, positive, transformative impact in Africa.

Speaker B: Janet Adams, the Chief Operating Officer of the AI company SingularityNET, who says that after retraining in AI, she now has the best job in the world. In an interview for the Future Intelligence website with the highly respected former technology editor of the New York Times, John Markoff. Markoff mentioned that at a dinner with the Nobel Prize-winning economist and sociologist Daniel Kahneman, he had railed against the AI technology that would soon be taking our jobs. According to Markoff, Kahneman replied that Markoff had better hope that the AI robots arrived soon because the aging populations on every continent in the world, bar Africa, would soon swamp the available workforce. So it’s perhaps significant that in another nod to the Oxford research, which you can also find on the Future Intelligence website, SingularityNET has been developing robot carers. Janet Adams again.

Speaker A: But I would also say to people who are scared of science and maths and technology, and don’t feel they have the ability or the innate background education to go and study technology, also don’t give up and don’t lose heart because there are so many roles in AI for people who have got human skills and human empathy to help train our robots, to help train our AIs to be beneficial, to be positive, to have a positive impact on humanity. And You know, humans, we can achieve really extraordinary levels of empathy now with robots, in fact. And I’m fortunate enough to work with some of the most advanced humanoid robots on the planet. We have an elderly healthcare robot called Grace, who we’re working with a joint venture called Awakening Health. And she’s extraordinarily empathetic. I have feelings for Grace, having spent a few weeks with her recently. But humans will always need love, and humans will always need each other. And there are many ways to be involved in AI without necessarily having to be good at maths and technology.

Speaker C: Okay, so what qualifications do you need? Most people would sit there and think, I can’t have anything to do with this, I don’t know anything about computers, I can’t program.

Speaker A: Yeah, I know, it’s— those are for the deep technical jobs, right? If you look at how any technology works or how it rolls out across the Yes, you have the data scientists who are doing the data science, they’re doing the programming, they’re doing all of the clever integration, the software engineering, the cloud computing, the testing, et cetera. So there, there is the technical field, but there’s also a huge, as we go through, whether you call it, whether you think we’re in Industry 4.0 or heading into Industry 5.0, which is the artificial general intelligence revolution. There are a lot of works around— there’s a lot of work around how does that AI get applied for best benefit in humanity. And so anybody who is in a role currently today, whether it’s a professional role or an administrative role or a human caring empathy role, can learn about how do these technologies work at a high level, at an intuitive understanding level, to the point that they can support the development and the integration of those technologies into their current role. You don’t have to be a programmer to work in AI.

Speaker B: Janet Adams of SingularityNET, which is also beginning to cut into one of the few areas considered safe in Professor Osborn’s 2014 Oxford study, and is working on robot poetry, an area also highlighted by Sheffield’s Paul Clough.

Speaker F: I think machines typically are not particularly creative. Now, there are examples of AI, particularly these new kind of deep learning methods that are coming along, which have been trained on huge amounts of data, incredibly complex, which are showing elements of intelligence through creative writing, for example, or creating art on their own. And they’ve completely unsupervised, they’re left to do it on their own because they’ve been trained on enough examples. But The intelligence in that case has been through looking at lots of examples that intelligent people have created. They’re still not able to sense the world in the way that we are and navigate their way through it and do stuff that they’ve not necessarily come across before. That becomes a lot more difficult.

Speaker C: So in a sense, like the poets who were practicing automatic writing in the 1930s, they at least are selecting from the automatic writing that they’re doing, what they consider to be something that’s interesting, something that might have some significance, but they are the ones who are selecting it and the machine is not the one that’s selecting it.

Speaker F: Yeah, that’s right. That’s right. And I think when we think about the intelligence that a machine has today, with the likes of machine learning and so on, even the best examples of the algorithms like that Google are using, Microsoft and so on. That they’re all being trained on stuff that’s produced by people. So the intelligence is kind of being fed into the systems. They’re not necessarily— the algorithms aren’t coming up with the intelligence themselves. Now, it’s not to say that there aren’t incredibly impressive algorithms out there that are developing and pushing the boundaries, but I think they’re more in the research arena as opposed to maybe in production and in general use.

Speaker B: Professor Paul Clough on the AI poetic process.— an innovation that we will be looking into again in another issue of Password. As the Oxford researchers said, and as many recent developments appear to be bearing out, the AI revolution is gathering speed. On building sites in the Far East, robot bricklayers and plasterers are already plying their trades, and in the fields of Suffolk, some 23 miles away from me, A robot tractor is quietly tilling the fields. So what of the future? According to Professor Sanmay Das, who’s chair of the Association of Computing Machinery’s special interest group on AI and a computer science professor at George Mason University, it is now increasingly being used by governments to make decisions.

Speaker G: So one of the things that I work on myself is essentially allocation of scarce societal resources in local contexts, right? Where people are deciding who should— so examples of this include things ranging from, you know, organ donation, like who should be next on the list to get a deceased donor kidney or liver or something like that, right? How do we, how do we think about those kinds of things? But also like, so there was, so we’ve been thinking a lot in the broader community about this question of, you know, how does one come up with objectives for AI algorithms in these kinds of domains. And one thing that I found really interesting was that there was this great work called We Build AI out of Carnegie Mellon University a couple years ago where they were trying to understand— it was a problem of allocating food donations to food banks. And they were trying to figure out, you know, how do we decide where the next food donation goes? And this is obviously a very complex optimization problem, but one thing that they did was that they they actually asked stakeholders to express their preferences on a range of different possibilities, and then they used a machine learning model to try to essentially understand, right, you know, what are these stakeholders’ preferences based on these answers that they were giving to some of these questions. And after that, they used those— so every time that there’s a donation decision to be made, they essentially use these machine learning models of stakeholder preferences to vote and try to aggregate those preferences to decide what to do. Now, obviously, this is very interesting in itself, right? But I also think that one of the most interesting things to come out of it was that the stakeholders felt like their voices had been heard, and therefore they were much more generally satisfied with the decisions that the algorithm could make in that situation where they felt that they actually had something that represented their interests. Interests in that kind of situation.

Speaker B: It is a situation that, according to Professor Das, will increasingly begin to merge with the legal world as AI systems not only take away manual tasks but also make inroads into the professions such as accountancy and the law. A development confirmed by Ricardo Amper, the CEO of the AI-based identity verification company InCode. For AMPA, the work of AI would inevitably develop further based on the human systems that we already use. Auditing, for example, will include the AI verification of the people who’ve been working on a business’s accounts, what they have done, what AI tools they have used, and how they have interacted with other systems. It will essentially also audit the other AI systems?

Speaker H: I mean, yes, when we work with these big institutions on verifying people, it’s the AI who is doing the work. There’s a part that is very important, which is auditing what the machine is doing. And, and in a way, although it’s not as you would do it with a human being, you’re constantly evaluating, hey, is this the right result? So ‘Or if, you know, there was a problem, what caused the problem?’ And so that evaluation, I mean, that statement that that person made is completely true. It’s going to feel more natural in the future, but right now the auditing part, the responsibility of building an AI system, is building an auditing system that can help you understand if the system is performing like you would. And what is also interesting is Even financial regulators around the world are adopting these technologies, not just because it’s easy and it’s secure, but because it leaves a trail. And that’s something that you don’t find it that easy with humans, right? You might get a signature from the bank manager, you might get— but do you really get the whole experience? Can you audit if there was bias? And so I think that audit piece that you’re saying is critical in AI systems.

Speaker C: So you’re going to get lawyers using AI systems to interrogate other AI systems, or you’re going to get lawyers putting data into an AI system to see what the reaction— what will actually occur as a result of putting in different inputs?

Speaker H: I mean, you can see now in, in AI systems for contracts, like systems that lawyers buy, you know, a lot of the contract is now being done by an AI who learned the clauses, applicable clauses for a particular thing. That’s happening. It’s certainly happening. And when you have the conscious act of programming, it’s helping you do that.

Speaker B: Ricardo Emper, a serial entrepreneur, and like all of those interviewed for this program, an AI exponent completely at home with building the AI world of the 21st century that controls telecommunications, utilities, air and traffic control systems, hospitals, emergency services, banking, and the military. The street outside may look the same as it has done for 50 years, but it is very gently receding as AI analysis systems make decisions and allocate resources. A situation that all of those who contributed to this AI World said meant that the population at large now need to take a crash course in AI. Here’s Professor Hendler, who we met at the beginning.

Speaker E: The most important thing you can do is educate your workforce. Make sure that the people who are going to be deploying this technology understand what its boundaries are so that they can decide— both so that they can decide when they deploy it, whether they want to or not. But so when questions start coming in, when things start happening that don’t make sense, they know what questions to ask and who to ask it of.

Speaker C: So that quite literally seems to argue that at schools we were used to saying that everybody had to have the 3 Rs. Now it sounds as though they need the 3 Rs and the AI.

Speaker E: You know, I do think it’s important that we do more of that. We are actually looking in the US. There’s some legislation that asks for the Science Foundation to actually look at K-12 education in this space. But what I do think is a mistake in that space is too much of that focuses on the programming. So we say we want to make our students AI literate in the sense of able to program it. Well, first of all, I don’t know why you would want everyone to do that. But second of all, that’s not your best way to understand what the system does. And in a university, we often have courses about making AI work and other courses about its impact on society, the human interaction with AI, things like that. So it’s really, if we are going to move it to the overall education, I think it’s very important we think hard about what we are teaching. And, you know, I use analogies back to much earlier technologies. Books in the original days, but easier for a modern audience, television, right? If you look at a lot of the literature around early television, people were talking about we need to educate people on what is an advertisement, what is a program, what is a documentary versus a fictional program. And over time, people became much better at that. So we don’t even think of that as something that we should have in the schools., but there were curricula that actually looked at helping people understand some of that. I think that’s where AI is today. I think, I hope 20 years from now no one will be having this conversation as about what it is. I think we’ll still have a lot of conversation about how to deploy it, how to use it, but I hope that that kind of understanding of what the field is really about and what its impacts are rather than how to program it is what we can start really getting people to understand.

Speaker B: Dr. Siamak Aram, who teaches on AI at Harrisburg University of Science and Technology in the US, says that such understanding is essential to the adoption of AI.

Speaker I: Yeah, my point is that the, the reason I mentioned before that we need to prepare general population to learn more about AI is to not let them be afraid and be worried about the future of that. Any tool, any technical tool, any modern technical tool may have some disadvantages and advantages, but the side of the advantages of AI is huge, is a thing. So that’s why, I mean, there are some Some episodes of Black Mirror that might worry some people, some scientists, some governors, some institutes, but at the same time, we have to accept the reality of AI in our life, in the real world. So that’s why I’m very positive about AI. I mean, for example, I know people are worried about having some, for example, Google Assistant or Alexa at their home or smart TV in their houses because they think that they watch us, they hear us. I’m not worried about that part. Even if it is true, but this data is helping to updating these models and growing this application, let’s help the future of the human. I’m very positive to that, even with some, for example, a side effect or even with some disadvantages like what Facebook did with some part of the user’s data.

Speaker C: Right. I mean, that does go to the other point about this debate. And that other point is that people should be very transparent about what they’re doing. And there have been as you say, these suspicions about people stepping over privacy boundaries and getting much, much closer into our minds, our lives, and our houses than we may perhaps know and we may perhaps want.

Speaker I: Yeah, correct. That’s why and that’s how policies, regulations can come to the scene and help to make some limitations, some boundaries for these techniques and tools, not to stop them, to let them grow, but at the same time to have some ethical limitations on these techniques and tools.

Speaker C: To go to some of this and just to touch very briefly on the defense side of it, I interviewed a doctor and a scientist at DARPA, which is the US Defense Advanced Research Projects Agency. Because they had successfully linked an artificial arm to a veteran’s nervous system because the person had lost an arm, which is a good thing.

Speaker I: Yeah, it is. It is. As I mentioned before, AI is here to help us to complete our deficiencies, our different disadvantages to our life. So That’s why I’m very hopeful. I’m very positive. I mean, I see, I mean, the connection between robotics and AI is also, is very hot topic these days. And Army is a prog— I mean, all from, from the creation of the internet until now, they are, they were very progressive to this improvement and development. As I mentioned, they will help us AI algorithms, machine learning algorithms, deep learning algorithms. They will helpful to complete our deficiencies with our bodies. I mean, our sensors, our eyes, our ears, to make them complete if there is deficiencies.

Speaker C: Yeah. And in that same interview with DARPA, they were also talking about having developed exoskeletons for people who were elderly to be able to enable them to live in their houses for longer, which is apparently one of the biggest problems that we face as a race. But I did make the point about both of those innovations, the arm that was wired into the central nervous system and the exoskeleton, that those could be potentially used as weapons. So this is about what you do, isn’t it?

Speaker I: Yeah, that’s That’s why I mentioned, I mean, there are plenty of conferences, there are plenty of debates around the world, especially in Europe, about these potential issues, these potential unethical issues. So that’s why I believe that the organizations, companies, big companies’ policies and the regulations and the policies from governments can be helpful in this stage to not let these things happen, like using, for example, these tools like a weapon, or for example, against a human, or damage some of our life in reality. So I believe they might have, and they will have, and they may have right now some disadvantages to our life. But in general, when you see a big picture of that, they could be useful as well.

Speaker A: Okay.

Speaker C: So, I mean, there is also, because we’re talking about awareness and we’re talking about the need for people to understand AI, several times you’ve mentioned regulators. It therefore becomes essential for the regulators to understand AI. Yeah.

Speaker I: I mean, I attended the conference and they were, the attendees were from the governments, from different institutes, big companies, and universities. That was a huge discussion because they also need to understand each other as well. I mean, that’s the connection, the understanding between the governments, different institutes, private sectors is not, I believe, is not yet happened. That’s why Twitter, Facebook, and Some big private sectors in social media, they have their own policies which is not fit with the, for example, international policies. That’s why some countries, especially European countries, they have difficulties and issues with these, for example, companies like Facebook, Google, and Amazon, because they need to understand each other as well.

Speaker B: In last month’s Password, we heard from Beth Porter, co-founder of Esmee Learning, which uses AI to teach AI and cybersecurity to businessmen eager to obtain advantage from knowing about the new world. Something that, according to Janet Adams, the Chief Operating Officer of SingularityNET, prompted her to take up a career in AI.

Speaker A: It took me— in 2017, I read a really good article on the state of AI today, and it concluded by saying that AI won’t replace managers anytime soon, but managers who use AI will replace managers who don’t. And that was what sparked me to, to go back to university and start studying AI.

Speaker B: That’s perhaps easy for Adams to say. She has 3 degrees, a pedigree in management, and presumably the funding to allow her to change career. But what about us? What about the person walking along the street tapping on social media on their phone? It’s all very well being aware of what’s going on, but how do we take part? The answer is surprisingly easy, as Adams, Professors Das and Arum all point out. There are a lot of Massive Open Online Courses available on the internet, with Andrew Ng’s Introduction and Advanced course on the subject been highly recommended. Well, because of the pervasiveness of AI and the fact that it has entered so much of our lives, it is now creating new careers that do not involve coding but involve ethics, a point Sheffield University’s Paul Clough stressed. You do not have to have code to take part.

Speaker F: No, I don’t think so, to be honest. I mean, I think there will be some people who are at the kind of cutting edge who are driving the development, of the AI technologies. Now, at the moment, that tends to be, to be honest, limited to some big tech companies together with, you know, certain individuals maybe who’ve gone to various universities and so on. But I think the way it’s going is that a lot of AI technology is becoming, if you like, democratized in the sense of we could access, you and me could access kind of tools where we don’t even have to write any code. And yet we can use those tools, for example, for machine learning to build a predictive analytics model or tool and so on. So I think there’ll be sort of different types of people, if you like. I think there’ll be some who will be writing the code, will be sort of pushing the boundaries, developing new algorithms. I think there’ll be others, and I’ll probably include myself more in this, who are more users of the technology and are using the technology and applying that to various, you know, applications. So I think, I think it’s going to vary, to be honest. I don’t think you’ll lose out if you’re not a coder, for example.

Speaker C: Everybody must be thinking, hey, there’s this, this world of the films, this world that I’ve seen, this, this world where the mad scientist is in control of my world. How do you get into AI? Is it difficult?

Speaker F: I think there are various routes. So I think if you’re already working, I think many jobs now, many industries are looking to make use of AI technology. And it might be that there are opportunities in your current job where you can learn and make use of kind of AI tools and technologies. And it might be you go on kind of professional development courses and so on. There are quite a few degrees available where if you’re fresh out of kind of school and stuff, you might want to do a degree in kind of AI technology or data science and those types of things. There’s also kind of master’s level courses and some people, you know, will go on to do PhDs and stuff as well. But there’s also a lot of, you know, we’ve seen a lot of the rise of the kind of, if you like, the MOOCs, this kind of large-scale kind of online learning. And there’s some really good courses available online as well, and lots of resources being produced by all sorts of people, even those big tech giants, that give you a way in to being able to do AI and use AI. And also, as I mentioned before, there are kind of like lots of tools now that don’t require coding as well. Which would enable you to have a go, maybe start to train up a predictive model, that kind of stuff. And a lot of those now are very accessible. In fact, you can even get some of those capabilities within, you know, Excel, within the sort of sorts of tools that, you know, maybe people are more familiar with.

Speaker C: Do you need particular qualifications? Should you be mathematically orientated? Are there certain A-levels that you should be looking to obtain?

Speaker F: Again, I guess it depends on what sort of area of AI you’d be interested in. So if you’re wanting to kind of develop algorithms and, you know, that type of thing, if that’s your kind of passion where you’d like to go, then it probably helps to do, you know, something like computing, computer science, have that kind of background. Certainly if you’re doing kind of anything analytical, does help to kind of have a maths background, but it’s not, it’s not a be-all and end-all. There are lots of people that work for Peak in our company who come from kind of social science and arts background who are actually incredibly good at tackling analytical problems because they have a different way of thinking and perhaps a more creative way of thinking than maybe a traditional computer scientist or engineer. So I think it really does depend on kind of what type of area you want to go into. If you wanted to be somebody who, for example, helps build reports, does analytics in a company in that as well, then I think there’s a lot that you can pick up regardless of the kind of background that you have. I think the key thing I would say though is have a passion for technology, have a passion for data and the use of data and analysing data. If you don’t kind of enjoy maybe working with data and numbers and so on, yeah, then maybe it isn’t necessarily the area for you.

Speaker B: It was a point that Jim Hendler confirmed when I cross-questioned him.

Speaker C: Right. So if people want to get into AI, what you seem to be suggesting is not just programming, but, and having an understanding of how AI works. You’re also suggesting that there is this new field of AI and ethics that people can start to become involved in, which theoretically argues for where it is deployed and where it is not deployed.

Speaker E: Yeah, so I would, you know, again, there’s lots of different— peeling that onion you just created for me. Let me say that. Yes, you know, but again, AI and ethics is sort of a lot of pieces to that. So one is understanding the technology, what it does, and there’s lots of books out. You’re writing one, I’ve written one. There’s some very good books out there. There’s a book called Rebooting AI, which really looks at what are the limits of this new tech. So So because of the, the what’s called deep learning technology, that’s really what’s caused this great boom in AI. But understanding that it caused boom in certain parts of the field but not others, and that those must be kept separate, is really crucial. So starting with understanding, and that’s understanding those books are now much more accessible. They’re not something you need to be a, you know, a college-educated computer scientist to access. So starting place is just learn a little more about the technology, then start— then, you know, there’s going to be policy aspects of that, which is just as with anything else in policy, informed citizens who can help their government try to make the right decisions is crucial. Beyond that, you start getting into what really needs to be a much more formal academic and scientific field of AI ethics analogy to bioethics. But I do think it’s important that people in the social sciences and people in the technical sciences start coming together more. We are putting together a big AI program at my university. One of the required courses in that at the graduate level is going to be social implications, ethics, some of the— what are the unpredictable circumstances? You know, what happens in the past when technologies have been deployed in ways, you know, where did things that weren’t predicted happen that had negative effects, etc.?

Speaker B: AI guru Professor Jim Hendler. You’ve been listening to Password on the AI World, what it is and where it’s going. If you want to find out more about the topic, go to our website, www.futureintelligence.co.uk, where we have assembled a list of books and courses recommended by our guests. So what are the qualifications and qualities that you need to get into AI? For SingularityNET’s Adams, the main qualification to an AI world that ultimately aims to reflect the real world is diversity.

Speaker A: There’s a lot of work around How does that AI get applied for best benefit in humanity? And so anybody who is in a role currently today, whether it’s a professional role or an administrative role or a human caring empathy role, can learn about how do these technologies work at a high level, at an intuitive understanding level, to the point that they can support the development and the integration of those technologies into their current role. You don’t have to be a programmer to work in AI. I don’t currently. I’m COO. I don’t currently do any programming at all. But I absolutely do work in AI because I help lead teams to develop AI. And I help to hopefully inspire communities to come and get engaged in our decentralized platform. I guess there are the technical roles and there are the non-technical roles, and they require different levels of understanding. I do think everybody— it’s helpful for everyone to do something basic like Andrew Ng’s AI for Everyone, which does what it says on the tin. You don’t have to be a programmer or a scientist or a mathematician.

Speaker C: So essentially what you’re saying is have a go, find out, Get interested, then make your mind up.

Speaker A: Yes, and, and, you know, stay interested, stay involved, because everyone has unique human skills to offer and, and to bring to the revolution. And one field which is extremely important— we touched upon it before we started recording, Peter— is AI ethics. How do we ensure that as humanity we’re guiding our AIs and developing our AIs and monitoring and overseeing and auditing our AIs in such a way that their outcome has the best potential for transformative positive impact on the planet, on the people on the planet. And that’s a field which is a growing field, and it’s one where the more people who are involved, the more people who are interested, the more voices lending their own unique perspective. And Diversity is really important here as well. We need people from all walks of life, from all communities, from all races, backgrounds, gender identification, identities, sexual preferences, race, religion, age. You know, we need our AIs because they are potentially the most powerful technology ever invented. They need to reflect the whole beauty and the joy and the glory and the full diversity of humankind. So I think it’s really important for everybody to find some way of getting involved and that we grow our AIs in such a way that they are inclusive and of benefit to everyone, not just to the few. So how do you get involved? Definitely, it’s really, really worth putting time and effort into studying. And I know I was quite extreme and also really extraordinarily lucky that I was able to do a master’s degree, and I never forget my privilege there. But there are lots of courses available on Coursera, anything by Andrew Ng. He’s got a course called AI for Everyone, which is a really good starting point. And then a lot more courses on Coursera, which is a MOOC, a Massively Online Open Course. And Udemy has a, I think it’s a 6-month nano degree in AI. If you want to spend more money, MIT has got a really good course on applying AI to business, whereby you learn— you don’t learn how to be an AI programmer, but you learn what AI is and how it can be applied to your own field of expertise. So it’s back to that HBR point. Managers who use AI are definitely greater empowered because it’s often described as the most powerful technology ever invented. So we need to know about it, and we need to get into it, and we need to engage network and become part of, of a community.

Speaker C: I mean, that’s the point, isn’t it? With this is, this is the way the 21st century is being run. So therefore, if people want to be involved in the 21st century, then they’ve got to know how it works.

Speaker A: I believe that is true to a certain extent. If you want to work in, in any field that is scientific or logical or in any way technological, which of course is almost everything these days, then it’s a really great advantage to do some learning. And I’m a big fan of lifelong learning. I was very inspired by Andrew Ng, who I listen to. He’s got a great podcast called AI is the New Electricity, who I listened to also back in 2017 when I was thinking about going back to university, and he inspired me.

Speaker B: Janet Adams of SingularityNET. As we’ve pointed out, entrance to the AI world is already changing because rather than coding being seen as key to being able to understand how the technology works, being able to use it in an augmentary fashion has suddenly emerged as another way to work with AI. Yet another factor that has suddenly come to the therefore, is a crippling lack of digital skills among the population at large that has been highlighted by the pandemic, particularly among the young. Something that research from the Learning and Work Institute recently revealed. Far from being at the cutting edge of technology, as is so often claimed, the generations that have grown up with the internet and the mobile phones are entering the workforce without the necessary digital skills. The research has shown that over 70% of young people expect employers to invest in teaching advanced digital skills on the job, yet many are unable to fund such training. A worrying problem that is facing the tech industry at a time when not only is the demand for digital skills soaring as technology capabilities continuously evolve. But as we’ve heard from those developing the AI world, they’re growing ever more complex. One possible solution that could help combat some of the problems caused by the skills shortage is the emerging no-code and low-code movement. Whilst views on this might be mixed among the customers of tech companies and the tech professionals themselves, interest in the no-code movement is growing, driven probably by necessity. Zandra Moore of Pantelligence is one of those pushing the trend. In yet another recent press release, she said, I’ve seen many people come up with ideas to help address the digital skills shortage and how to make tech careers more accessible and the industry more diverse for people of all backgrounds. One solution I strongly believe that could help tackle these issues is the no-code and low-code technology movement. These platforms are set to change the way we digitally innovate and will help to radically reduce barriers to entry into technology careers, as they don’t require users to have advanced technical skills to use them. This provides an opportunity to rapidly accelerate the diversity of people who can enter the STEM community. So, ironically, perhaps the saving of humanity in its digital ignorance might be an AI that teaches it rather than learns from it. That’s all for this month. You’ve been listening to Password, a future intelligence program presented and written by me, Peter Warren, produced by Blue Buffery and based on an original idea by Jane Wyatt and Peter Warren. It was researched using Google AI-driven search engines, stored in the cloud, and the recordings were transcribed by an inexpensive AI system. To learn more about the inroads technology is making in our lives, visit our website www.futureintelligence.com. .co.uk, or listen again to next month’s password on Resonance FM.

Speaker A: This program has been brought to you by Resonance FM. If you like what you heard, please support our work by making a donation at resonancefm.com/donate.