Insights · September 29th, 2023

Stephen Fry on How to use AI as a force for good at CogX Festival in 2023. CogX is the World’s Biggest Festival of AI and Transformational Tech.

Essential thinking and historical background on AI with a view on what comes next.

Full Transcription

Utilizing technology as a force for good, as if.

It is an honor to be here. I don’t doubt you will hear and have already heard from figures far more qualified than I to talk about things technical, mathematical, financial, and corporate. If what I say seems obvious to you, you’ll have to forgive me, but it’s sometimes the obvious things that need the most repetition and reinforcement. It’s why artists from Johto to Goya from Francis Bacon to Damien have so often reminded us that we’re going to die. It’s obvious we know it, but we need to be reminded. 

Let me begin with an image that you might be good enough to hold in your heads. A family is playing at the seaside. They sit on the beach, with their backs to the sea digging and making sandcastles or perhaps they’re playing beach cricket. But that’s the image of family on the beach, their backs to the sea. Store that away for a moment. Now let’s talk about the hot button issue of the day a I incident incidentally, allow me before we go any further instantly to demand that from now on, we always write and type this acronym, these initials uppercase a lowercase i. Otherwise, we’re upsetting everyone in the world called owl. I’ve spoken to my old friend Mr. Pacino, he’s very much in approving of my plan. He’s fed up with reading that owl is a threat to our way of life. Or the suggestion that it’s only faintly possible in our lifetimes, that owl will achieve some form of sentience. 

AI is a subject I’ve been fervently fascinated by since the 1970s. When the great Marvin Minsky, the so called Father of artificial intelligence laid out some of the groundwork for neural networks. What were then called expert systems as well and other potential Lines Of Approach. He founded the AI Lab at MIT. His period of ascendancy was known as the AI winter. Unfortunately, the joke was, AI of any usable kind was always 30 years away, always on the horizon and never reached, not because science and mathematics weren’t there. They were, it was the technology. There wasn’t a world full of digital data for AI systems to mine back then. And there weren’t powerful enough computers. 

But, the ineluctable destiny of Moore’s law, as I’m sure you know, dictated an exponential growth in computer processing power, along with similar curves of growth that brought us consistently increasing memory capacity, display resolutions and all the other astonishments that we have had the dizzying pleasure of living through. It’s hard to visualize the effects of this simple law that Gordon Moore propounded back in the 1960s. He was one of the founders of the company, Integrated Electronics, or as we now call it, Intel, every 18 months or thereabouts. He declared, the number of transistors that could be fitted onto a microchip would double and he was right. Here’s an example of what that actually means. 1996 was the year that Garry Kasparov perhaps the greatest chess player ever to push a pawn was defeated by IBM’s Deep Blue machine that year 1996 The most powerful supercomputer in the world more powerful by far than Deep Blue was the ASCII read, capable of more than a trillion floating point operations a second ASCII read stayed the most powerful computer until the year 2000. It cost $45 million equivalent to 75 million today, by the year 2006. All that processing speed of a teraflop per second was available in a Sony PlayStation three. In other words, in just 10 years, the ASCII read went from a $75 million supercomputer to the equivalent of the gaming console in a teenager’s bedroom. As you probably know, if you offered a child today a PlayStation three, they would snot in your face derisively. That’s why I think it’s sometimes better to think of Moore’s Law expressing not so much an increase in power as a decrease in cost. Every year what you can do with a computer gets cheaper and cheaper and therefore, available to more and more. And now, thanks to all the fantastic quantities of data that have accumulated, and all that doubling and redoubling of processing power, the AI winter is finally over.

I gave a lecture on AI at the Hay on Wye Literary Festival seven years ago. And amongst other things, I predicted that every smartphone app, every insurance company credit rating firm, every basic digital service would soon start claiming AI to describe their functionality in their software offer, when in reality, they would be using good old algorithms in the usual way. And you will see that this has come to pass. Everything today is an AI, but almost nothing is really. Instead I shouldn’t say that when I predicted you should very much bear in mind that like every single other person interested in this field of human endeavor, I am an utterly hopeless profit. Here’s an example of how crappy a predictor I couldn’t be excited to join the internet in the late 80s. It will be years before Tim Berners Lee devised the worldwide web. There was nothing graphical in those days on the net. The way in was to use a system called telnet. to hop from university computer to university computer with the help of primitive search services, like Gopher, Veronica and Janet. files could be uploaded and downloaded with a protocol called FTP. 

But, when Tim did release the web, and those first mosaic web browsers emerged, I began to believe in the right that this new thing, this internet, was a gateway to an ideal new human future. An online world of museums, libraries, meeting squares, schools, cinemas, shopping arcades, noticeboards, clubs and parks for all kinds of play and gathering and all free. Yes, there were a few dark alleys and red light districts but what great city didn’t have those and with email and new forms of instant messaging, all the difficulties disagreements, resentments, prejudices, bigotries, and horrors of political and social discord would melt away, and a new era of international understanding would flourish, banishing borders and enmities through the 90s and into the naughties. 

The public launch of Facebook Twitter, everything became surely even more promising a golden New Jerusalem was being builded here, no more walls, no more hatred. The internet was all gifted. In Greek. That’s pan, Dora. And in myth, Pandora was given the best attributes of all the gods, so that she was indeed all gifted her rather the all gifted internet. I should have remembered that Pandora had been given a box by the gods, who then told her never to open it. Actually, it was a job at common parlance calls it a box. Anyway, one night she could not resist lifting the lid, and out flew hundreds of malicious sprites, leathery winged creatures, wheeling and shrieking as they flew off into the night to infest the golden world. They were misery, poverty, crime, disease, murder, war, cruelty, falsehood, viciousness and violence. And they plague us to this day, when the lid was lifted from the Pandora’s Box of the internet when exactly, we can’t say maybe it was the Cambridge analytical scandal. But by the time of 2016, coincidentally or not the time of Brexit and Trump, it was clear that any rosy view of the internet was criminally naive, by now outed flown all the malware the bots, the criminals, the hackers, the groomers, exploiters frauds, Grifters, thieves, propagandists and extortionists. When Pandora in a panic saw all the ugly sprites flying from her box, she slammed down the lid, unwittingly imprisoning the one last sprite in the box, which to this day, is still beating its wings frantically against the sides, desperate to get out. And the name of that last sprite was hope.

Are things as bad as that for our Pandora’s Box? Is there no hope? No Way Out no hope for a way out of the culture wars, the spirals of deceit, identity theft, abuse, bullying, fakery and fraud. The AI of Marvin Minsky that I was so looking forward to can we be foolish enough again, to believe in anything but disaster coming from it? Twice bitten once shy, surely, am I now dreading the horrors AI will bring the amplification of everything dark and detrimental, the further and complete erosion of trust, security and privacy, the further collapse of everything on which healthy, sane, productive social living depends. Was the internet merely the anti John the Baptist, heralding the arrival of AI, the anti Messiah who will darken our world forever? Well, generic motive systems like the large language models that have expressed themselves into the public sphere recently as chat bots, such as chat GPT, and its variants are not of course, AI in any meaningful sense of intelligence. They’ve been splendidly described by the University of Washington, computational linguist, Emily Bender, has no more than stochastic parrots that squawked their streams of output, according to a probability determined by their access to such masses of data that their owners have been permitted to mine, advanced autocorrect In other words, the speed and apparent fluency of the squawking certainly gives the impression of intelligence and might even pass the Turing test. And certainly remarkable things can be achieved. Corners can be cut that save labor for report writers, students, paralegals, et cetera, et cetera. 

Visual generic native models, like Midjourney can make the eyes pop to when mid Jenny was asked a fairly simple question to generate for portraits of Stephen Fry, in the style of the French impressionist Edward Digha. Best known for his paintings of ballet dancers. This is what the program came up with. Yeah, not bad as you know, I’m also a proud member of SAG, the Screen Actors Guild, and you may know that we’ve been on strike for close on three months now. And one of the burning issues at stake is Ay ay ay. In case you’re wondering why, let me play you my voice reading a documentary narration just as a clip first, as it actually is my voice, and then an AI version to give an idea of what we’re talking about in this field. So here’s a short extract of me performing the commentary for a film about Dutch resistance groups fighting against the Nazis. In 1937, Lao Maduro, a Dutch human rights lawyer, went to the Paris World Exposition. It is there she hears a lecture by two German scientists on how they intend to purify the German race. Lau immediately traveled to Germany, where she discovers Hitler is building training camps, which are effectively the nascent beginnings of what would later become concentration camps. Lau knows that the Dutch population census records every citizen’s religion, she warns the Dutch government that if this information falls into Nazi hands, it will spell disaster for Dutch Jews. The government chooses to ignore her warnings 14th of May 1940 Germany invades the Netherlands. Rice commissar Arthur sighs and appoints SS hubster infura Ferdinand ouster fronton, to ensure that the Netherlands is cleansed of Jews. Right. So there we are. Now let’s hear the AI version. No, because that was the AI version. I said not one word of that. It was a machine. Yes, it shocked me. They used my reading of just seven audio books, the Harry Potter novels. And from that data set, an AI of my voice was created. And it made that new narration what I played you included, as you heard a variety of German and Dutch names, and general vocabulary hoped stoom fewer, for example, Arthur Zeiss. All those names. And all those words, of course, are not in Rowling’s books. What you heard was not the result of a mashup. There are plenty of those and they’re completely obvious. This is from a flexible, artificial voice, where the words are modulated to fit the meaning of each sentence uniquely. It could therefore have me read anything from a call to storm parliament to hard porn or a product endorsement, all without my knowledge. And this what you just heard was done without my knowledge and without my permission, when six months or so I heard about this. I sent it to my agents on both sides of the Atlantic and they went ballistic. They had no idea that such a thing was possible. But I had to say something to them which I have to say to you, you ain’t seen nothing yet. It won’t be long before full deep fake video is just as convincing it already is but at a cost which will soon tumbled down into the realm of free or as good as these generic active AI tricks the so called AI chatbots and the like that we now have the bolt on so called AI modules that are being sold to companies and individuals everywhere. We have to think of them as being like Karl Benz’s first working automobile in 1895. Impressive,

By no means the finished article, goodness, a motor an engine that isn’t steam, and which can pull a whole carriage no horse. How impressive but it will never replace the horse. It’s clumsy, smelly and unreliable. Well, yes, automobiles were in 1895. For the next decade or so to drive them. You had to understand how they worked. You had to be prepared to get out and get under every mile also and tweak the magneto or the carburation. Nobody imagined interstate highways, motorway service stations, motels, 10 storey parking structures, car ferries and whole towns and economies that were founded on the idea of that belching sputtering Karl Benz motor, all in the lifetime of my grandfather. Technology is not a noun, it is a verb, it is always moving. What we have now is not what will be. And when it comes to AI models that can improve themselves. What we have now will advance at a faster rate than any technology we have yet seen. So never ever judge a new gizmo, device gadget system or notion by what it is. 

I was lucky enough to know Steve Jobs quite well. He was not an engineer. He was not a businessman, at least not as much of a one as his successor Tim Cook is he was not a scientist. He was not an industrial designer. Like his friend and colleague, the great Johnny Ive, he was a visionary, which sounds a flabby word, but it means he saw things that others didn’t. He envisioned them as Americans like to say a favorite quote of his was from the Great Canadian hockey player. Wayne Gretzky. When Gretzky was asked why he was so much better a player than anyone else in the NHL, Gretzky said, other players skate to the puck, I skate to where the puck is going to be. Steve skated to where the puck was going to be. So let’s look at two aspects of AI as a technology, one that is common to all inventions, and another that is unique to AI. 

Technology has, as philosophers say, no moral valency take Gutenberg and his development of movable type printing. Before his printing press in 1450. All written material in the world was done by hand by monks and Clark’s. After him. Printed matter flooded the world. His machine unleashed everything that made us from the age of reason through the enlightenment, the industrial revolution, everything, but no moral valency in the machine itself. The same dumb press can print the sonnets of Shakespeare on Monday, and Hitler’s Mein Kampf. On Tuesday, the technology unthinkingly complies to the wishes and ethos the moral valency of the operator. The microphone and radio station doesn’t know if it’s Gandy broadcasting, or use if gerbils. A gun does not know its target. The difference with AI systems is that unlike guns and printing presses, they must have moral valency which is to say we insist that they must we feel that if they don’t We’re doomed. unsupervised machine learning is an immensely powerful technology. systems can teach themselves games with bewildering speed. But why must AI systems be taught moral values you ask? Well, we all in this great venue. I I would suggest, surely agree that an AI should know that human life is of supreme importance that men and women are of equal value, that all humans are of equal value, whatever their gender, race, skin color, sexuality, level of education, class, upbringing, religion, non religion, political beliefs, or even criminal record that children deserve a special kind of respect and protection, that we all have the right to free assembly, free speech, free thought individual opinions, it’s taken us centuries to come to this conclusion. We would therefore want AI to attach more credence to say John Stuart Mill’s masterpiece on liberty, or Thomas Paine’s the rights of man than it would to the aforementioned main cam for fakes, like the Protocols of the Elders of Zion. After all, without our guidance AI might be the dinette of endless unsupervised scraping for all data, all human exchanges and records that fill the whole wide internet, an AI system might come to different conclusions. As each machine without any moral sense scoops up the data from the weirder corners of social media, it may absorb some of the prejudices out there in the darkest slimy list, and most festering pools of racial supremacy, ultra radical religious fundamentalism, toxic masculine masculinity, et cetera, et cetera, some of the slime might stick. 

So, as in the original great science fiction novels of Asimov and others, we install primary directives in the AI similar to the famous Laws of Robotics, primary directives, which cleave to the human rights that we value, primary directives, it cannot transgress will pass over the inconvenient fact that we can’t agree on those values. It’s not like the days of the writing of the US Constitution when there was a broad enlightenment consensus, but we’ll drive over that for it gets worse. Because Wait, what use is the building of adjust and moral AI? China is building AI systems of equal if not greater power than those in the academic, governmental and private sectors of the West. They believe and why shouldn’t they believe that their AI should have their values baked in, making clear that the Communist Party of China matters more than any individual that the citizens first duty is not to themselves, but to the state? That to question or criticize the state, the party, its functionaries and its leaders is treason. That it is right and admirable to report neighbors who say bad things about communism who questioned the party, who make jokes about the leader who spread Western ideas. The Free Press, alternative sexualities, and art free from censorship are pernicious Western decadence that deserve to be suppressed? So we must wake up and smell the Lapsang su Shang? There can be no question that China is building exactly such systems with exactly such values, and that these systems will massively increase the state’s ability to enforce them. Nor will their AI be sandboxed or ring fenced outside the world internet that we use, it will crawl into and contaminate every corner of our digital world. Chinese mobile networks and social media services tick tock, for example, are already permeated around our own world of the internet, and that will only continue Russian AI will do the same according to the lights of its value as grotesque As We May Think them. And on top of that, we have a deeper problem for the Champions of our western values will rightly we think, insist that we weaken, not strengthen our western AI is power to scrutinize surveil, Spy report and limit the freedom of our own citizens, which will Perforce make China’s Big Brother AI with no such scruples stronger than ours, more far reaching and all encroaching. This is an arms race where uniquely we in the West will deliberately choose less powerful weapons than our adversaries across the world. If we didn’t do that, it would mean surrendering our values to theirs. In other words, in order to achieve parity with our rivals, our perceived enemies, we would have to adopt their ethical and moral frameworks. And yet it is the difference between their ethical and moral frameworks and ours. That is the reason we think our system of democracy, openness, free speech, etc. is so superior, and so worth defending. 

These are exactly the problems that we faced in the Cold War. writers like John Kerry often portrayed the bind that we were in our spies and their controllers, being forced to become no less ruthless, no less secretive, dishonest, cruel and treacherous than the wicked opposition. For our system of tolerance, openness and freedom to prevail. We have to suspend our tolerance, openness and freedom. Nice. So whichever way you slice it, AI will constrict our freedom, security and our privacy, all in the name of protecting our freedom, security, and privacy is our way through this. We’ve talked about Karl bents and his automobile. The fact is that almost from the very beginning, we controlled how people were allowed to drive cars. And we still do that as a driving test, an age requirement, a driving license, and insurance imperative laws on intoxicating substances at the wheel. We tell citizens on which side of the road they may drive, we impose speed limits, emission limits, tire tread standards, we insist on the installation of certain safety features like seatbelts, crumple zones and airbags, there are still deaths on the road. But by and large, those rules keep us safe and allow traffic to more or less move and flow safely. The right to fly an aircraft is even more curtailed and circumscribed with broadcasting another exceptionally powerful technology. We insist on levels of fairness, balance, truthfulness, decency, etc. At various times what is considered decent is adjusted to the prevailing mode. So language, nudity and violence that would have been utterly unacceptable on television in the 70s is now fine. On the other hand, representations of women, ethnic minorities, or gay people that were acceptable a generation ago or not. So today, we can just about manage to navigate our way through the conflict between this need to control broadcasting and the need to allow freedom of expression. It isn’t easy. BBC directors general often fall on their swords or get booted out. All left wing people think the BBC is biased to the right, or right wingers think it’s a hotbed of lefties. In the wider media. The controls that govern good practice in the press can sometimes seem shackling. Sometimes they can seem to allow press barons to impose themselves on our democracy in less than satisfactory ways. We’re used to all this, used to trying to square the circle and maintain the right balances. Can we do something similar in the field of AI? Can we demand the kinds of rules, regulations and oversight that seem natural with motoring and television? Off calm to tell Ofsted of AI? By happy coincidence, while I was writing this yesterday, Elon Musk, Sam Altman of OpenAI, Mark Zuckerberg and others, convened in Washington to announce that there is overwhelming consensus on AI regulation. If such dyed in the wool Randian libertarians are in favor of government intervention and control that must mean something. But the guardrails and guidelines that they’re calling to be legislated into existence won’t work abroad. 

And as ever, politicians insist on talking about technology as a noun, not as a verb. They keep skating to the puck, and there’s another snag. The BBC was founded and motoring unwound as a way of life in days of hierarchy. When society was run by figures derived from long established positions of more or less accepted, or authority, society and its institutions were governed by people who wish they were not always respected, were usually obeyed and seen born to govern and to regulate by virtue of education and matters, it seems to us now, birth and accent. Now we live in a different world. Thank goodness for that. It goes without saying we do not trust governments or educated elites anymore, frankly, than we trust Silicon Valley bros. But it does mean, unfortunately, that we trust no one. No one is in charge. There appear to be no grownups, no one whose word we can Trust, no one whose judgment, whose honor, whose moral valency we can believe in David Attenborough, that’s about it. 

Do you remember the beginning of this? I asked you to hold an image in your mind. A family is playing at the seaside. I said, they sit on the beach, their backs to the sea, making sandcastles well, that family is us. The human family, we’re playing with our backs to the ocean and out their way out there at sea, different currents and swelling. They are the currents of a standing new technologies, quantum computing, nanobots, robotics, brain machine interfacing, where the human mind can control a computer and in the other direction computers can read and control human minds. Musk is working on these and claims great successful ready bio augmentation bionics, genomics CRISPR, gene editing stem cells, revolutionary new materials like gallium arsenide and graphene, they’re the tip of the iceberg. Each of these currents each of these technologies, especially when amplified and assisted by AI, in all its modalities has the potential on its own, to be existentially transformative in the human world, but joined with the others in a great Confluence. They will rise and rush towards the shore. That tsunami is massing and massing. While we still play and build our sandcastles, even though we can now begin to hear its roar. We’d rather not turn round. Thank you very much. governments and political parties talk and prepare budgets and layout plans and election manifestos for a world a polity, way of life and work that will very shortly no longer exist. And all this is coming at the same time as a climate change catastrophe. It seems to be a question of which of these horsemen of the Apocalypse will catch us first, just as the heat domes and the flash floods and the droughts and hurricanes will come and keep on coming.

Will the waves of new technology – governed and ungovernable – we will just have to see what happens. We must hope that all the good outcomes the medical advances, pharmaceutical designs, protein folding mysteries solved and astounding discoveries made the neat solutions to logistical and other problems that have plagued us for years, the new ways to educate children housing, rehabilitate criminals, that all these will outweigh the bad things, the surveillance, the unemployment, the malignant powers abroad and at home, they will hack and despoil pervert the truth, manipulate and exploit us, for ironically, perhaps, it is AI for all its threats, they will offer the best chance for a solution to itself. It is to be hoped for the climate apocalypse too. So I’ll end this by talking about the singularity. 

I think my beloved Greek myths have a solution here, at least as a way of thinking about the singularity, as you know, is probably the name given to the moment that AI achieves Sapiens, from artificial general intelligence to human-like self consciousness, the singularity. Well, we were created by Prometheus, the Titan, the immortal friend of Zeus, the king of the gods, and he loved us very much, and Zeus put up with us. We were small, we were articulate. We were not like other animals, exactly. But we weren’t like the gods at all. And he said that we could survive and we could live, so long as we never were given fire. Fire only lived for the gods by fire, he meant literally fire, the fire that roasts and toasts that melts and smelts the fire that allows technology to transform the materials of the world. But he also meant the divine fire, the self consciousness, the awareness, the creative spirit, that made the gods better than humans. Well, Prometheus loved us so much that he disobeyed Zeus and he went to Olympus and he, with a phenol stalk, stole fire. And he took it down and he gave it to mankind. And Zeus looking down one evening saw the spots of light everywhere, and realized that man had fire and was terrified because he knew that one day, they would unseat the gods because now they had the Gods secret, this divine ability to be aware of the self to separate yourself completely from animals. So he punished Prometheus by chaining him to a mountainside, sending an eagle to pick out his liver every day because it grew back every night because he was immortal. Well, it’s an interesting myth about creation that the humans are sort of like little gods and then we became gods and we got rid of the gods and indeed, the Greeks did get rid of the gods in a sense, they still to believe in themselves in their own power to add and subtract and to think and solve problems and be little gods as humans. And by the time we did this in the 18th century, we threw off the shackles of Ecclesiasticus ism and the power of the church and the enlightenment. blossomed. Prometheus became my hero. Beethoven wrote the creatures of Prometheus. Shelley wrote Prometheus Unbound, and his wife, Mary Shelley, wrote Frankenstein, subtitled The Modern Prometheus, they regarded the Prometheus myth as being about them, throwing off the gods and standing up and being human. But actually, it’s now that it matters because we have the same problem that Prometheus and Zeus had. Here are these entities that we have created just to Zeus and Prometheus created mankind at the moment that clever little parrots squawking around with no real self consciousness or awareness of self. And there are Prometheus amongst us who want to give that spark. And, there are users amongst us who are terrified of the possibility. 

So, we have to decide if we’re a Prometheus, or a Zeus. There’s no right answer. But the one thing we can all agree on, is that it’s a fucking weird time to be alive. 

Thank you.

About Stephen Fry

Stephen Fry is a legend – read more.

About Nikolas Badminton

Nikolas Badminton is a world-renowned futurist speaker, consultant, author, media producer, and executive advisor that has spoken to, and worked with, over 300 of the world’s most impactful organizations and governments.

He helps shape the visions that shape impactful organizations, trillion-dollar companies, progressive governments, and 200+ billion dollar investment funds.

Nikolas Badminton’s book Facing Our Futures: How Foresight, Futures Design and Strategy Creates Prosperity and Growth has been selected for 2023 J.P. Morgan Summer Reading List, and featured as the ‘Next Gen Pick’ to inform the next generation of thinkers that lead us into our futures. 

Artificial Intelligence
Nikolas Badminton – Chief Futurist

Nikolas Badminton

Nikolas is the Chief Futurist of the Futurist Think Tank. He is world-renowned futurist speaker, a Fellow of The RSA, and has worked with over 300 of the world’s most impactful companies to establish strategic foresight capabilities, identify trends shaping our world, help anticipate unforeseen risks, and design equitable futures for all. In his new book – ‘Facing Our Futures’ – he challenges short-term thinking and provides executives and organizations with the foundations for futures design and the tools to ignite curiosity, create a framework for futures exploration, and shift their mindset from what is to WHAT IF…

Contact Nikolas