S5E09: Gashed Hull: Intro

This is Larry Lessig. I hope you remember the metaphor that has framed this whole season. The Titanic hits an iceberg. The captain notes the immediate disruption. He sees what I've called the overturned tables. And as he looks at those tables, he's confident we can clean up the mess that the iceberg has created. But then his crew reports to him that the iceberg has created a gash in the hull as well and that six of her 16 water type compartments are flooding. This fact, the captain realizes, means that the ship is going down. Or put more abstractly, this fact, he realizes, means that the trip is can't continue on the same ship. No doubt he also knows there are not enough lifeboats for every passenger. That's not because they were being cheap. But the theory of lifeboats at the time was that any ship in trouble would be able to summon a close by ship that would arrive in time to affect any needed rescue. But not with the Titanic. What the gashed hull and the Titanic meant was that this way of moving passengers across the Atlantic was over. Something different, at least for those lucky few, would be necessary. In this section of this season of this podcast, another way, my aim is to show you why this current way of running democracy or running a representative democracy or republic is over. In this section, my aim is to show you why we will need lifeboats. Now that sounds dire. Mainly because there's nothing about lifeboats that's particularly attractive, especially if you carry the metaphor through lifeboats in the middle of the Atlantic, at the end of the winter, in the middle of the night, when the presumed alternative is going back to bed on the unsinkable Titanic. But here's where the metaphor weakens. Because when we get to that section of the season, the lifeboats section of the season, the third and final section of this season, my aim is going to be to convince you that indeed, life on the lifeboats would be better than this life. That it's a better ship, a safer ship, a less annoying and embarrassing ship to carry. The ideals of democracy, that we should embrace the leap and make it together as quickly as we can. But all that's to come. All that's after we complete the objective of this section. The objective? To prove that there is a gashed hull in this Titanic before we can argue that it's time to jump to the lifeboats. Okay, so the gashed hull, what has gashed our hull? In a word, the gash in our hull is caused by AI. But the word AI won't quite carry the meaning. I need you to understand. So for the next few minutes, I want to introduce the idea in a less familiar way to help you to see what I want you to take it to mean. The word artificial intelligence suggests a distinction between natural intelligence and artificial intelligence. We humans have natural intelligence. It's superior, no doubt, to many animals. But of course, there's lots we don't understand about the intelligence of many other species. No one who studied bees, for example, can doubt their natural intelligence. But artificial intelligence is the intelligence that we make. It has an instrumental rationality, something it's trying to do in light of stuff in the world. And it is successful in that rationality, or not, based on how it interacts with that stuff in the world. Notice that definition says nothing about computers. Indeed, in this sense of the term artificial intelligence, we can understand certain real world institutions, very old institutions, as examples of artificial intelligence, even though they have nothing to do with computers. Think about democracy, for example. Democracy is an institution with an instrumental rationality. It aims to enable a public to resolve differences among itself and to commit to a common end or ends. And it does that in light of stuff in the world. For example, on December 8, 1941, the United States Congress declared war on Japan in light of Japan's bombing of Pearl harbor and other locations across the Pacific the day before. That decision was a product of the representative democracy AI that our Constitution had established. The Constitution created a congress. It had a president who could inspire a congress to almost unanimously vote to go to war with Japan. That AI acted in light of facts in the world. Those facts had flashed red. That flash induced the reaction. We saw, in this sense, the intelligence of democracy. An instrumentally rational intelligence acted. This intelligence was obviously artificial. It didn't exist naturally. But this artificial intelligence didn't depend upon computers. We could call it AI, but let's more precisely call it an example of analog AI. In the very same sense. We could say that corporations too are an AI. They too are artificial. God didn't make corporations. And they too are instrumental in the modern American sense of a corporation in form, or better misinformed by theorists such as Milton Friedman. The sole purpose of a corporation is to maximize shareholder value. It does that maximization in light of facts in the world. And it succeeds, or not, depending on how well it comprehends those facts and acts in light of them. Now, I say misinformed just to flag an important point about corporations specifically. They have not always and do not now everywhere describe themselves or think of their purpose to be. The single purpose of maximizing shareholder value. To the contrary, when corporations were first chartered in the Anglo American tradition, they often had a very specific purpose, not tied to shareholder value necessarily at all, or not directly. Those charters were long lists of instructions from the legislature to the corporation telling it what it must do in exchange for the freedom to be a corporation and to try to to make money. Within that democracy, and likewise across the world, corporations or their equivalents have a complicated sense of their purpose. Profit is at the core of all of them, at least if they're a for profit corporation. But rarely is profit the only objective being maximized or pursued. But here in America, we've simplified things pretty dramatically. In America, most corporations believe that the only purpose of their corporation is to maximize shareholder value. That is its objective. It is instrumentally rational towards that objective. In this sense, again, it is an AI. And though it may use computers in this process, we'll call it another example of an analog AI. We have two kinds of analog AIs, democracy and corporations. Obviously, sometimes the objectives of these two different AIs could overlap. If, for example, a democracy wants to build a highway for its citizens, that would be consistent with the objective of a highway construction firm. Both want the same thing, to build a highway. And the firm is as happy as a cop could possibly be to help the democracy achieve its objective. But equally obviously, the objectives of these two analog AIs wouldn't necessarily complement each other equally. Obviously they can conflict. So, for example, a coal fired power plant makes money selling electricity. It makes more money the cheaper it is to make electricity. If it does nothing to prevent pollution from its burning of coal, then its interest maximizing its own profit conflicts with the interest of a democracy. Interest for a safe, clean environment. And when that conflict happens, the ordinary assumption within a democracy is that democracy will stand up and prevail. Or again, between the interest of the corporation and the interest of the democracy, the democracy will be able to exercise its sovereign authority to assure that the corporation conforms its behavior to the public interest. Okay, but to even utter those words seems weird today, because we all recognize the many, many contexts in which corporations are able to advance their interests against the interests of democracy. Or more precisely, that when their interests conflict with democracy, they're pretty darn good at protecting their interests despite democracy. And of course, it's for this reason that long before digital AI, many of us were terrified about the future of democracy. Because corporations, or maybe wealth more generally, had developed a suite of powerful techniques to defeat democracy when its interests conflicted with democracy. Of course, they would never characterize it like that. They would say they'd won where the interests of democracy and the interest of capital coincided. But I'm not sure many actually would believe that account anymore. Most looking at our government today would believe that the analog AI of corporations was actually better, instrumentally, more rational or successfully rational than the democracy. When the interests of democracy and corporations conflict. Okay, so before we think anything about digital AI, let me introduce a distinction that will help us map these different AIs. When an AI has an objective that is consistent with your objective or the objective of your community, we can call that an allied AI. And when it has an objective that is inconsistent with your objective or the objective of your community, we can call that an alien AI. When it's consistent, we should be happy that AI is on our side. And when it's inconsistent, we should be fearful. And so these two dimensions, allied versus alien and analog versus digital, suggest we're going to have four kinds of AIs here. We're going to have allied, analog, and digital AIs, where the AI is advancing an interest that is our interest. And it will have an alien, analog and digital AI, where the AI is advancing an interest which is conflicting with our interest. This distinction, then, will help us think more clearly about digital AI, because here too, we can understand an intelligence having an objective, and we can evaluate how well it achieves its objective. And that objective could be either allied or alien. And our concern should, of course, be when it's alien. But let's start with the allied. There are plenty of examples of digital AI with with an objective that's allied to our own. Think about Netflix recommendation engine, working hard to identify which movie or TV shows you're likely to like. It has an objective to get you to watch more. It serves that objective by better predicting what you're likely to want to watch so that you watch more. And as it does that, you may indeed watch more. But if this is television or movies, that's not likely to be a terrible thing. There are, of course, some people who just can't get off the couch. But most of us are able to just slot in our TV time and then move on to dinner or bed. So, unlike social media, most of us are pretty good at regulating our consumption of movies and tv. So if movies and TV are recommended to us based on what we're more likely to like, then the AI wins, and so do we. Allied objectives benefit everyone. Yet obviously, these objectives are not always allied. Take the case of social media and democratic politics. The objective of social media, Facebook TikTok etc is to maximize engagement with its platform. The best way to achieve that objective is to flood the channels with polarizing content. That content drives engagement. Good for them, but by polarizing democracy, bad for us. The objectives of these two conflict. The objectives of democracy and the objectives of the social media companies. The objectives conflict. And as the recent history of politics in America suggests, these conflicting objectives this alien AI has actually resolved in favor of the AI. Democratic politics has done squat diddly to resist the harm social media AI has done to exploit it. And that harm is only going to get worse. As AI evolves and as we see it evolving, it will become a tool deployed not just by advertisers, but by enemies, by governments, China, Russia, leveraging our vulnerability to its instrumental rationality. And as it does that, its effect on American politics will only grow. It will press us to say what we should do in response, if we can do anything in response. Now, this conflict, I predict, will be devastating to the democracy as we experience it now. Or put differently, if we want to continue with democracy, if this is an idea we want to find a way to transport into the middle section of the 21st century, we won't be able to continue with it in the way we do it now. We will need to protect it from the corruptions of this emerging AI. We will need to put it into a place where it can't be bent or subverted or turned against itself or against us. How we do this I describe in the third part of the season, lifeboats. The burden of this part is to establish that indeed, the hull is gashed, that even if we could fix the overturned tables, the hull is gashed. And we must recognize this means we have to find a way to relocate democracy into a safer space. Okay, to do this, we're going to start with what seems like a fantasy for democracy today. We're going to start talking about a time when it was possible to write a book titled the Rational Public. The first interview in this part will be with the authors of a political science classic, Ben Page and Robert Shapiro. They'll discuss their 1992 book, the Rational Public, and more importantly, they describe the conditions under which it was plausible to speak of the public as rational. We'll then hear from Marcus Prior, author of a favorite book of mine, post Broadcast Democracy, which brings to the foreground an idea that too many simply just ignore, that the technical structures of media affect fundamentally the kind of democracy that is possible within that media, and that the very best and worst of the 20th century, both democracy and autocracy flowed from the affordances of this media. This will be a critical point to recognize because we all see how fundamentally different modern media is from the media media of the 1960s and the 1970s. We all see that if indeed it was the structure of that media that made that democracy possible, then that different structure today may mean that that democracy is no longer possible because that structure has completely changed now. We'll then turn directly to AI, digital AI, and we'll think about the consequences from what Tristan Harris calls for first contact with AI. That first contact was social media. We'll discuss the consequences of that contact with Jonathan Haidt and for Media with Ben Smith, and then with the man behind one of the most successful documentaries in the history of documentaries, the Social Dilemma. And that's Tristan Harris. Okay, that's a long, long wind up. I hope you enjoy the pitch. Stay tuned for the first episode in this second part, the Gashed Hull. This has been the ninth episode of season five of the podcast Another Way. It's basically been me blathering, but me blathering about maps and maps are important because it's important to see where we are and where we're going. And if you're still hungry after this blathering introduction, you continue listening. We've posted the next episode after this at the very same time. These podcasts are produced by equal citizens in an abstract sense, at least in a literal sense. They're produced by Josh Elstrow of Elstro Productions. You can find more about equal citizens at equalcitizens.us and you can give us your thoughts and feedback and your money. I love the thoughts and feedback. Feedback I depend. We depend on the money. I don't get any of the money. Somehow that would be fun. But no, none of the money comes to me. My work here is pro bono. But not everyone can work pro bono for equal citizens. So we need that support and support us even in tiny amounts every month. That would be great. Thanks for listening to this introduction to the second part of this season, Gast Hall. Stay tuned for the first episode.