Yuval Noah Harari on the Challenge of AI and the Future of Humanity

0:00On today's AI Breakdown, we excerpt a recent speech by Yuval Noah Harari on the challenge of AI and the future of humanity. Before that, on the AI Breakdown Brief, we talk about more AI-related job loss, another AI pioneer coming out with warnings about AI, and much, much more. The AI Breakdown is a daily podcast and video about the most important news and discussions in AI. If you're enjoying it, please like, share, and subscribe.

0:28Welcome back to the AI Breakdown Brief, all the AI headline news you need in five minutes or less. Another AI pioneer is sounding the warning on AI. Just a couple of weeks after Jeffrey Hinton started his media tour warning of the concerns that he had with AI, his fellow 2018 Turing Award winner, Joshua Bengio, is saying some similar things. In a recent interview with the Financial Times, Bengio pointed to the arms race in big tech companies as the key culprit in why things have gotten so much more dangerous. He said there is currently a lack of scrutiny being applied to the technology. Companies can rent access to ChatGPT, he said.

1:04It would be important that this access be closely monitored so we know who is using those systems so we can track potentially illegal or dangerous uses. Discussing the constant one-upmanship between companies like Microsoft, OpenAI, Google, and Amazon, he said this race dynamic is a vicious circle. Now, Bengio signed the pause letter with more than a thousand others recently, but he said the disagreement among AI experts around what the right course of action was, was in fact a signal that we should slow down. He said, if we disagree, it means we don't know if it could be dangerous. And if we don't know, it means we must act to protect ourselves.

1:37If you want humanity and society to survive these challenges, we can't have the competition between people, companies and countries and a very weak international coordination. Now, when it comes to what we need to do, he said, quote, like investments in discern in Europe or space programs. That's the scale where AI public investment should be today to really bring the benefits of AI to everyone and not just to make a lot of money. Bengio said that he is also currently discussing ideas with Hinton for ways to take on these challenges. Now, interestingly, that Turing Award in 2018 went to three people.

2:11Hinton and Bengio, who, as you've just heard, have serious concerns over AI. But then a third winner was Jan Lacoon. Jan is the head of AI at Meta and has a very, very different take on how risky it is. From a New York Times article this morning, quote, Dr. Lacoon argues that this kind of technology is not as dangerous as it might seem. He said small numbers of individuals could already generate and spread disinformation and hate speech. He added that toxic material could be tightly restricted by social networks such as Facebook. You can't prevent people from creating nonsense or dangerous information or whatever, he said, but you can stop it from being disseminated.

2:45Now, interestingly, the real focus of this article is how Meta has decided that open source is the right strategy for them to compete against rivals like Google and Microsoft. Lacoon says the platform that will win will be the open one. Do you want every AI system to be under the control of a couple of powerful American companies, he asked? I don't know if this article was driven by a PR agency or just the New York Times happened to intersect with the rising narrative, but given that a couple weeks ago we saw that Google memo leaked about how open source was the right strategy to compete in AI, it's pretty interesting to see.

3:16Speaking of Google, another leak from that company has suggested, perhaps no surprise, that they'd be using new AI models to help improve ads as well as to help YouTube creators figure out what to create content around, what to call it, and provide other similar assistance. The report also said that Google is working on its own DALI or stable diffusion-like product for image creation. Now, while Google may be thinking about how AI can improve marketing, marketers themselves seem pretty concerned about the impact of AI on their industry. This is particularly true among search marketers who aren't sure what the future search experience is actually going to be like.

3:49They built their entire business model around being one of a list of blue links on a Google page, and that might change dramatically if people just start listening to the oracle that is ChatGPT and BARD. A recent survey of more than 1 ,000 marketers asked, Are you concerned about the future discovery of your content as search engines implement AI chatbots into search? Only 17 % were not at all concerned, with 17 % slightly concerned, 24 % somewhat concerned, 21 % moderately concerned, and 21 % extremely concerned. Speaking of concern, the latest news about job cuts by AI came today from British Telecom.

4:22The company says that they're going to cut up to 55 ,000 jobs by 2030, and in fact, will be replacing thousands of those jobs with artificial intelligence tools. The total job cuts over the next decade represent about 40 % of the workforce, and the CEO says that he expects around 10 ,000 of those jobs to be replaced by AI. Now, overall, a new World Economic Forum report on the future of jobs found that there will be roughly 14 million less jobs over the next four years. The WAF survey suggested that employers around the world estimate that between 2023 and 2024, 83 million jobs will be lost while just 69 million new jobs will be created.

5:00Now, of course, as some jobs evaporate by the technology, other jobs will be created. The report says, quote, The fastest growing roles relative to their size today are driven by technology, digitalization and sustainability. Perhaps no surprise that the fastest growing position will be, quote, AI and machine learning specialists, factoring in an employment growth rate of 39 % within the next five years. In short, big changes ahead. That's it for today's AI Breakdown Brief. If you're enjoying it, please like, subscribe, and share. And I'll be back soon with the main AI Breakdown. Today on the AI Breakdown, we're doing something a little bit different.

5:35This will be a shortened curation of the most important and salient points from a recent talk on AI and the Future of Humanity. On April 29th, historian, philosopher, and author Yuval Noah Harari keynoted the Frontiers Forum in Switzerland. The talk was called AI and the Future of Humanity. Now, I firmly believe that to truly discuss and understand AI in the world, we have to engage with it on at least three levels. There is first the individual level, using the tools and understanding how much power they provide, but also how disruptive they're likely to be to the way that we earn our livings and the way that we live our lives.

6:14The second level is, let's call it industrial scale, how these technologies are going to fundamentally reshape entire industries, entire swaths of jobs, and what that might mean for society because of that. The third level is the largest scale, the societal scale, which could mean changes in the nature of belief, changes in institutions, and of course even the risk of existential catastrophe. Now I too often find that that final conversation, that third conversation about the societal level risk of AI is presented in a binary. Either you are convinced that AI left to its own devices will end humanity, or you are effectively an accelerationist cheering on our human-machine hybrid future.

6:55Now, my guess is that for the 100 million plus people that have come to intimately understand and engage with AI over the last six months in the post-ChatGPT era, the vast majority are still trying to grapple with these questions and trying to figure out what they think, what they believe. What I like about Yuval's talk is that it's not about future theoreticals. It's about the challenges of what's happened already. I don't think you need to agree with all or even many of his conclusions to find this useful. But for you busy folks out there, I wanted to curate it down to its essence. So I've taken about a 30-minute presentation and cut out the set of excerpts, which total a little less than 10 minutes.

7:30I'll only interject in between to give a little bit of glue from section to section. We start with Yuval explaining why the entertainment industry's conception of AI risk is so misleading. People have feared AI since the very beginning of the computer age, and this fear has inspired many science fiction classics like The Terminator or The Matrix. While such science fiction scenarios have become cultural landmarks, they haven't usually been taken seriously. Science fiction scenarios usually assume that before AI can pose a significant threat to humanity, it will have to reach or to pass two important milestones.

8:16First, AI will have to become sentient and develop consciousness, feelings, emotions. Otherwise, why would it even want to take over the world? Secondly, AI will have to become adept at navigating the physical world. robots will have to be able to move around and operate. If they cannot move around the physical world, how can they possibly take it over? The bad news is that to threaten the survival of human civilization, AI doesn't really need consciousness, and it doesn't need the ability to move around the physical world. Over the last few years, new AI tools have been unleashed into the public sphere, which may threaten the survival of human civilization from a very unexpected direction.

9:16So then if we don't need sentient cyborgs, where does the risk of today's AI come from? The most important aspect of the current phase of the ongoing AI revolution is that AI is gaining mastery of language at a level that surpasses the average human ability. And by gaining mastery of language, AI is seizing the master key, unlocking the doors of all our institutions, from banks to temples. Because language is the tool that we use to give instructions to our bank, and also to inspire heavenly visions in our minds. Another way to think of it is that AI has just hacked the operating system of human civilization.

10:12The operating system of every human culture in history has always been language. Okay, but even if AI has these linguistic capabilities, this mastery over the operating system. Isn't it just another tool that is subject to human use? Remember that we humans, we never really have direct access to reality. We are always cocooned by culture. Previously, this cultural cocoon was always woven by other human beings. Tools, previous tools like printing presses or radios or televisions, they helped to spread the cultural ideas and creations of humans, but they could never create something new by themselves.

11:00A printing press cannot create a new book. It's always done by a human. AI is fundamentally different from printing presses, from radios, from every previous invention in history, because it can create completely new ideas. It can create a new culture. And the big question is, what will it be like to experience reality through a prism produced by a non-human intelligence, by an alien intelligence? So what does this actually mean in terms of the real dangers and changes that might come? The dangers it disposes are fundamental, very, very different from everything or most of the things imagined in science fiction movies and books.

11:50Previously, people have mostly feared the physical threat that intelligent machines pose. So the Terminator depicted robots running in the streets and shooting people. The Matrix assumed that to gain total control of human society, AI would first need to get physical control of our brains and directly connect our brains to the computer network. But this is wrong. Simply by gaining mastery of human language, AI has all it needs in order to cocoon us in a matrix-like world of illusions. For thousands of years, prophets and poets and politicians have used language and storytelling in order to manipulate and to control people and to reshape society.

12:44Now AI is likely to be able to do it. And once it can doubt that, it doesn't need to send killer robots to shoot us. It can get humans to pull the trigger if it really needs to. But is there any evidence to validate that this type of change and influence can be as problematic as Yuval is making it out to be? If we are not careful, a curtain of illusions could descend over the whole of humankind, and we will never be able to tear that curtain away or even realize that it is there, because we'll think this is reality. If this sounds far-fetched, just look at social media over the last few years.

13:23social media has given us a small taste of things to come. In social media, primitive AI tools, AI tools but very primitive, have been used not to create content but to curate content which is produced by human beings. The humans produce stories and videos and whatever and the AI chooses which stories which videos would reach our ears and eyes, selecting those that will get the most attention, that will be the most viral. And while very primitive, these AI tools have nevertheless been sufficient to create this kind of curtain of illusions that increased societal polarization all over the world, undermined our mental health and destabilized democratic societies.

14:18Millions of people have confused these illusions for the reality. The new AI tools are far, far more powerful than these social media algorithms, and they could cause far more damage. For hearts and minds, intimacy is the most effective weapon of all. And AI has just gained the ability to mass produce intimacy with millions, hundreds of millions of people. Now, as you probably all know, over the past decade, social media has become a battlefield for controlling human attention. Now, with the new generation of AI, the battlefront is shifting from attention to intimacy. And this is very bad news.

15:08what will happen to human society and to human psychology as AI fights AI in a battle to create intimate relationships with us, relationships that can then be used to convince us to buy particular products or to vote for particular politicians. Even without creating fake intimacy, the new AI tools would have an immense influence on human opinions and on our worldview. So what do we need to do? What's the solution? Or at least, what's the starting point? In order to make sure that the new AI tools are used for good and not for ill, we first need to appreciate their true capabilities. since 1945, we knew that nuclear technology could destroy, physically destroy, human civilization, as well as benefiting us by producing cheap and plentiful energy.

16:10We therefore reshaped the entire international order to protect ourselves and to make sure that nuclear technology is used primarily for good. We now have to grapple with a new weapon of mass destruction that can annihilate our mental and social world. And one big difference between nukes and AI, nukes cannot produce more powerful nukes. AI can produce more powerful AI. So we need to act quickly before AI gets out of our own control. If we don't slow down the AI arms race, we will not have time to even understand what is happening, let alone to regulate effectively this incredibly powerful technology.

17:05And finally, wouldn't any sort of pause or slow down just allow autocratic regimes to zoom out ahead? Now, you might be wondering or asking, won't slowing down the public deployment of AI cause democracies to lag behind more ruthless authoritarian regimes? And the answer is absolutely no, exactly the opposite. Unregulated AI deployment is what will cause democracies to lose to dictatorships. Because if we unleash chaos, authoritarian regimes could more easily contain this chaos than could open societies. Democracy in essence is a conversation. Democracy is an open conversation. You know, dictatorship is a dictate.

17:55There is one person dictating everything, no conversation. Democracy is a conversation between many people about what to do. And conversations rely on language. language. When AI hacks language, it means it could destroy our ability to conduct meaningful public conversations, thereby destroying democracy. If we wait for the chaos, it will be too late to regulate it in a democratic way. Maybe in an authoritarian or totalitarian way it will still be possible to regulate. But how can you regulate something democratically if you can't hold the conversation about it. All right, to wrap up, what I think is most valuable to contemplate is the fact that this assessment of the threat is here right now.

18:43It comes from the capabilities that the technologies have in their ability to use language today. That doesn't mean that we should ignore future challenges. It just means that the challenges aren't all of the future. Now, you'll probably have noted that there wasn't really anything positive about AI in here. That's not because Yuval is a full pessimist. He just says in the larger presentation that there are plenty of people out there talking about the benefits of AI, namely the industry that's creating these tools, so he didn't really need to do that. I think that's a fair point. As for me, I am an incorrigible optimist, and the potential for AI for positive transformation is blindingly clear.

19:17But I try to ground my optimism by really sticking my nose in the dangerous or threatening or risky possibilities. And that's the spirit this was presented in. Not in pessimism, but in giving ourselves the best chance of that positive potential. That's it for today's AI Breakdown. If you're enjoying it, please like, subscribe and share. And until next time, peace.