Transcript 0:00 Hey folks, it's Quinn. Since we recorded this conversation, we've renamed the podcast. It used to be called Important Not Important, which is what you'll hear in the episode and is still the name of the company. 0:12 But now the podcast is called The Most Important Question. That is, what can I do? 0:18 It's the question we get the most from all of you, and it turns out it's the question I've really been asking guests for almost two hundred episodes now. So here we are, the most important question. 0:29 Please enjoy this old conversation. Thanks for listening, and thanks for giving a shit. [upbeat music] This week, sure, ChatGPT is cool, and by some accounts, completely out of control. 0:47 But why? I think I know the answer. 0:50 Later we'll talk about the news, why American hospitals are in trouble, how to help teenagers with their mental health, e-bike rebates that keep selling out, and how food can be medicine after all. 1:02 But first, I'm Quinn Emmett, and this is Important Not Important: Science for People Who Give a Shit. 1:09 The newsletter features the most important science news, how to think about it, and what the hell you all can do about it. 1:16 Hit subscribe right now to get this newsletter and my conversations with the world's smartest people every single week. 1:23 You can find the email version and links to everything at importantnotimportant.com/newsletter or right in your show notes. It's February seventeenth, two thousand and twenty-three. 1:37 Here's your weekly action steps, or what the hell we can do about the world. Addiction is brutal. Help yourself or a loved one, or someone you've never met, with Shatterproof. 1:48 I'm also so excited to share that my favorite event on Earth, LA Loves Alex's Lemonade Stand, is finally back. 1:55 You can support pediatric cancer research and buy yourself some tickets to eat food from some of the greatest chefs on the planet. 2:02 Number three, it's a pretty, pretty, pretty good moment to read The Atlas of AI: Power, Politics, and the Planetary Cost of Artificial Intelligence, and The Alignment Problem: Machine Learning and Human Values, of which, as you'll find out, we need more of. 2:18 Lastly, wanna switch your retirement fund to one without fossil fuels? Check out Fossil Free Funds to find mutual funds and ETFs that qualify. And now, today's big question. 2:31 In screenwriting, there's a well-honed idea that main characters should want one thing but need something different, something that is often opposed to or even opposite their most public desires. 2:43 These characters are blind to what they need the most, and often purposefully so, having shoved those feelings down just about as far as they can go. 2:54 Trust me when I say, having a long, hard look at yourself isn't easy or comfortable. So we empathize with these characters because, I mean, who amongst us hasn't shoved those feelings down, right? 3:08 It's an imperfect character development mechanism, of course. Shouldn't be that simple. And the best characters aren't that simple. None of us are either. 3:18 That said, history is littered with memorable characters who reluctantly go through transformations, who finally walk away from what they want and go through hell to get what they needed all along, letting us experience what it's like to have that long, hard look without actually having to, you know, do it. 3:43 Web search was intended to give us what we need. But over time, the utility has been hijacked to give us what we want. We need, each of us needs, as society, we need a real answer. 3:56 But at this point, search most often gives us what we want, some version of self-affirmation. And if it's delivered by a paid advertisement that looks just like the real answer, well, that's even better. 4:07 That process, over and over, billions of times a day, leads to disinformation. Sometimes disinformation hurts one person, as we'll see below. But at scale, disinformation inevitably hurts many, many people. 4:22 Imagining that search could ever give us entirely objective answers all the time ignores the web's original sin. The web is only what we put into it, and we are fundamentally flawed. 4:35 The internet and search is so fundamentally broken that we desperately want the next thing, these AI chatbots, to be everything all at once. 4:44 But that's even more dangerous because instead of your question returning a list of links ranked by Google, most of which are now paid ads, or a news feed of extreme views from friends and family on Facebook and Twitter, a chatbot is an extremely convincing version of both. 5:03 It's incredibly confident, but often very wrong. But we can't tell the difference, and I'm not sure we want to. 5:11 I'm not gonna spend this week's essay assessing the technological capabilities of search or new large language models, because that assessment will be old news almost immediately. 5:23 What I do wanna do is try to force us to confront our wants and needs, to confront our expectations for ourselves and technology born of who we are. 5:34 This construct that has remained the same for eons and underpins every single system and tool we've ever built. Disinformation has always been wielded as a weapon. 5:47 Imagine, if you will, our cave people ancestors, perfect strangers until a serendipitous meet cute, choosing to enjoy each other's company on a glorious fall day in Eurasia. 5:58 A typical primitive conversation ensues as each of our ancestors tell tall tales of wrestling giant sloths and saber-toothed big cats to the ground.Otherwise harmless stories that may have once been rooted in truth. 6:12 Sure, the saber-toothed cat was injured and already on the ground. I mean, holy shit, what a lucky day, but no one needs to know that. 6:19 Over time, the stories become embellished to not only help us feel a little better about ourselves, but to help perfect caveman strangers understand exactly who they're dealing with here. 6:31 And now imagine it's dusk, and whereas your great-grandfather to the tenth power is new to the area, mine is not. And all of a sudden, a sudden, terrible noise echoes across a dark sky. Story time's over. 6:45 Your ancestor needs an answer, a verifiably safe cave right now. What he wants is a sense of safety. 6:54 And after an afternoon of sharing stories with a new friend, he trusts my ancestor, this self-described local cat wrestler, to give him the answer he needed. 7:05 Unfortunately, my ancestor isn't incentivized to do that because resources like shelter are extremely thin and dangerous to acquire. 7:14 My ancestor almost definitely lies to your ancestor about which dark cave is safe to sleep in. But why? 7:22 With our extremely limited understanding of the world around us, like for example, where the hell weather came from or when it might strike, survival in 7:32 sixty-four thousand BC may have effectively, at times, been a zero-sum game, or at least as far as we knew. So my guy made sure he survived the night by giving your guy some very strategic disinformation. 7:45 Now, disclaimer, I'm using guy in this situation and throughout the essay because we are usually dumber. So now your instinct might be to say about my guy, "Fuck that guy," and that's totally fair. 7:58 But as tragic as the result may have been for your ancestor specifically, the society-wide stakes of his exit were very low. Only your one single ancestor got eaten. 8:11 News of his demise didn't and couldn't travel because there were no newspapers, certainly no news feeds, because for a while there, there wasn't even language. 8:22 But as Ryan North put it in his book, How to Invent Everything, "Language is a technology from which all others spread, and you've already got it for free." Let's look at some more recent examples. 8:36 SARS-CoV-2's infection fatality rate, that's the virus that causes COVID sometimes, even when we were completely unprotected, is less than half of what SARS-CoV-1 was that hit Asia in two thousand and three and four. 8:50 SARS-CoV-2, again, associated disease COVID-19, killed far fewer hosts than SARS would have, which is on the one hand, great. 9:00 On the other hand, fewer dead hosts means more alive hosts who can transmit the virus much, much more widely, letting the virus mutate more often and more quickly, killing in some far more people. 9:15 This is how disinformation at scale works. Disinformation, like its less intentional cousin misinformation, is as old as we are, like viruses and weapons. But technological progress has scaled language. 9:31 It's made disinformation scale and impacts considerably more devastating. Look at it this way. 9:37 If our ancestors couldn't trust each other, even after a nice afternoon, we'd eventually fight over necessities like shelter and water. And skirmishes started very small, very small, hand-to-hand. 9:50 But fighting with weapons evolved from fists to clubs, from swords to spears, bow and arrow, and muskets, automatic weapons, artillery, missiles, and finally to nuclear weapons. 10:03 And that evolution traces a lineage where real-world deployment becomes exponentially more deadly over time as reach increases. 10:12 In parallel, thankfully, the most powerful weapons have actually been used relatively less over time because of ideas like mutually assured destruction, which is the ultimate fuck around and find out scenario for everyone. 10:25 And thankfully, viral pandemics like COVID don't happen very often, even if we're definitely due for more. 10:31 And yet, in the twentieth and twenty-first centuries, disinformation around weapons and viruses has still cost millions of lives. 10:40 In all cases, the self-awareness to understand the part we play, a species constantly seeking out what we each want instead of what we all need, is absolutely necessary. 10:53 Every tool we've ever built has been a manifestation of our wants. Another example, when Facebook launched across college campuses in early 2004, you could only lie so much. 11:06 Your reach extended only as far as your own profile. What's the point? In fall that year, Facebook launched the Wall, a way to post messages on your own profile page. But again, your reach was limited. 11:18 Someone had to actually go and visit your profile to see your latest updates. And I mean, unless you're really crushing on somebody, who's really gonna put in that much effort? 11:29 In 2005, high schoolers were allowed onto the platform. And in 2006, anyone over thirteen years old got access. 11:36 And building on top of those, the News Feed launched, changing the mechanics of personal information distribution from pull to push. 11:46 The News Feed, a minute-by-minute live broadcast of my updates on your updates and our coworkers' and your cousins' updates all stacked on top of one another, became the primary way of interfacing with Facebook and everything. 12:00 And as it grew, it's how we interfaced with each other. It was no longer so difficult to keep up with close friends and family, and eventually even further with former colleagues and sorority sisters. 12:12 It was revelatory, truly.But it wasn't enough, because maybe sometimes you just didn't wanna see my frequent it's complicated relationships status updates. 12:24 But you really liked when Frank shared his marathon training updates. And so in two thousand and nine, Facebook introduced the Like button. 12:34 And in two thousand and eleven, after a couple years of compiling a then rudimentary layer of user preferences, including by extending the Like button beyond Facebook itself to the entire internet, the news feed was changed from a chronological presentation to one algorithmically driven by how often and where you mashed that Like button. 12:55 In two thousand and twelve, ads came to the news feed to pay for all of this computing power, ads increasingly hyper targeted to you based on just how much of your personal information you were willing to share and and what you liked. 13:08 We never looked back because that would require taking a long hard look at who we are and why we keep building the same tools. It would require looking back at who got left behind and why. 13:23 It would require us asking questions like, "Is this new tool something we're capable of handling?" And, "What harm might it cause alongside incalculable profits?" 13:35 Eventually, our news feeds and our suggested groups became so specifically targeted to us that we trusted them to be our go to source for how to feel, for answers. Let me explain a little more. 13:49 Years of mashing the like and favorite buttons, of clicking on banner ads, and feeding those algorithmic timelines, and using auto complete searches have all come together to turn you into an idea. 14:03 As Facebook, and WhatsApp, and YouTube, and Google, and to a lesser extent, Twitter have scaled, all relying on basically the exact same business model, we have built and been overwhelmed by a variety and volume of misinformation and weaponized disinformation that spreads like a virus. 14:22 Like SARS-CoV-2, this disinformation is the kind of virus where each of us is in a little less danger than we would be from, like, Ebola or smallpox. 14:32 But on the whole, because of the transmission rate of disinformation, and because we cannot seem to separate what we want from what we need, the impact is nearly insurmountable, touching every part of our society, political systems, and economy. 14:49 And that's because in twenty twenty-three, we don't subscribe to updates from people. You are no longer a person. 14:56 You are an idea assembled and harvested and sold and targeted all over again from all the data you've wittingly and unwittingly shared over your digital lifetime, telling yourself and the entire world a story about who you are, trying desperately to get what you want, not what you really need. 15:15 These data paint a terrifyingly accurate picture of you as an idea of where you've been and where you might go, of your menstrual cycle and purchase history, what you watch and search for, who you have relationships with, and those you've paid less attention to or cut off entirely. 15:33 This idea of you, of each of us, is both terrifyingly accurate, but also often dangerously wrong, and here's why. We're in this together. 15:46 Do you think my ancestor regretted selling yours short when that storm knocked down a tree, pinning him underneath it the next morning with no one around to help because you were eaten by a tiger? 15:57 Do you think that he too wanted to feel safe when what he needed was companionship, which may have even made him more safe? You aren't the only one who fits into our very specific digital idea because I can tag you. 16:11 I can like you. I can follow you. Because I can mention you, because I can down vote you, because I can flag you, because I can contact you directly, you aren't the only one who shares information about you. 16:26 And even though each of us sometimes intentionally tells a story that isn't entirely accurate, from wrestling big cats to girlfriends in Canada, 16:36 when other people and brands share information about you, and especially when they sell that information, it is likely to be even more wrong and sometimes even on purpose. 16:48 In a very short amount of time, profile info became status updates. It became echo chambers. It became a calcified electorate. 16:57 In the same way that there is more than ever no base layer of factual information we can all agree on to get answers from. 17:06 When information is intentionally weaponized at scale using tools created by imperfect, at best, humans to satisfy quarterly financial growth projections, the implications aren't just one dead caveman, but half of an electorate that believes there are 5G chips in vaccines, and sure, fuck it. 17:24 Let's hang the vice president of our own party. Remember, Trump wasn't a cause. He was a symptom, like the cruelty he embodied and empowered. He is so fucking charismatic to millions of people. 17:39 They ignored everything else, and they saw and continued to see themselves in him. He, like Obama in an entirely different scenario, was an idea, a bunch of different ideas, an aspiration. 17:50 He was the entire point, the sum of decades of efforts coming together to make something exactly like him that preyed on every single fear and weakness among us. We put ourselves into him, and that's what came out. 18:03 Every tool we build, every tool we trust is an extension of what we want, and that has never changed. [outro jingle]Hey, it's Quinn. 18:17 One of my favorite things about this job is to not actually do this job. Stick with me. Sometimes you get a sign off. You know what I mean? 18:26 When I'm forcing myself to not think about climate change or malaria, I'm usually cooking, eating, downloading apps, listening to music, or reading dragon smut. 18:36 And every month, I share my favorite updates on these in a members-only newsletter called Not Important. 18:42 If you wanna receive it, head to importantnotimportant.com/upgrade or just click the link in your show notes to start your free thirty-day membership trial. 19:01 Generative AI, or more specifically GPT or Lambda, Bard, or all the many thousands of niche-specific tools being built on their foundational models every day, is simply the sum of everything we've ever put into the internet, and the idea of what a supposedly all-knowing synthetic version of each of us and all of us combined should be. 19:22 That's what they're literally designed to do. To answer a question you asked based on everything it's ever read, which is everything we've ever put into the world, we've made them incredibly and dangerously confident. 19:36 The internet can be a vessel for incredible things, for deep, terrifying, hateful things. The internet has empowered and uplifted billions, right? 19:46 As relatively rudimentary data networks and flip phones get East African farmers paid, while thousands of miles away, New York bankers make a gazillion times as much trading the fertilizer commodities those same farmers rely on. 20:01 The internet has helped us express ourselves in ways we could have never imagined to audiences larger than some entire countries, helping the world normalize a beautiful collage of human and civil rights. 20:14 And sometimes when we're brave enough to share all of that, all of us, ourselves, we're digitally hunted down. Maybe we take our own life. 20:25 The internet has helped us map our continents and oceans, map genomes and redline city blocks. It has helped us find love in unexpected places and reduced our contact with one another. 20:37 It has helped put very big numbers in context, like our maternal health outcomes suck compared to other industrialized countries, and strategically supply news outlets with big numbers with very little context at all, like vaccine side effects. 20:52 It's disinformation at its finest. 20:55 The internet has helped us more efficiently and comprehensively count ourselves, and then more efficiently decide who counts and who doesn't, so we can more efficiently marginalize more people out of voting rights and mortgages and into prisons and unwanted pregnancies. 21:12 The internet has helped us catapult mRNA vaccines to the finish line just in time, catapult conspiracy theories about them to the top of 4chan and 8chan boards across the world. 21:25 It has always helped us because we have always sought more caves, more tools to live and feel more safe, more efficient, more productive, more wealthy among our tribes, our ideas. 21:38 We have always wanted the internet to exist as a better, more informed version of ourselves and our existing groups, to most comprehensively 21:46 imprint and expand on our beliefs, to comfort us with our fears, to market our wares with less effort and more return on investment. We want it to be everything because we need it to be everything. And overall, 22:00 it can be. 22:02 Internet is the only conceivable way an exhausted, flashcard-adverse liberal arts major like me could have learned everything I have learned these past few years and connected with everyone I have connected with on this journey. 22:17 It's the only way I could have, every single week, researched, synthesized, and presented to you a collection of facts to tell a story to encourage you to improve on the outcome of that story. That's why we're here. 22:33 Make no mistake though, the facts I have presented to you have often been wrong. Not intentionally or necessarily in the moment. I mean, that's my worst fear. It's not my business, unlike some other asshats. 22:47 But because of user error, I didn't do enough research. I trusted the wrong sources. I didn't cross-check among them. 22:54 I simply don't understand the mechanics of some chemical or industrial or biological process, or most likely, because the facts changed. And facts do change, especially in science. 23:05 I mean, Jesus, we've learned that, right? That's the entire point though. Like software development or society, science is a process, not an outcome. 23:15 We very often discover, even when we don't want to admit it, that there's some bugs in the code, that our assumptions are wrong, especially when more people are using it. But in science, being wrong is usually the point. 23:28 Here's a hypothesis. Let's prove it wrong. In society, in politics, we very rarely wanna step into those waters to acknowledge our mistakes, much less to rectify them. 23:41 And just because America is the oldest democracy doesn't mean our system works best. I mean, clearly. 23:47 Even new and improved, but still imperfect democracies like Norway and Ireland exist to know we're due for some serious upgrades here, you know? 23:56 We've learned more than enough in these almost two hundred and fifty years to know a million ways we can improve on our systems. 24:04 We wanna continue on, though, but we need to revisit the code of conduct laid out by a gaggle of white male slave owners. 24:12 The incentives aren't there for us to make those changes, much less to examine ourselves, to address the uncomfortable realities of who we are and what we need.Shit happens in science too, obviously. 24:25 We have misused science in horrific ways because we didn't understand what the fuck we were doing, or worse, because we knew exactly what we were doing. Sometimes in the moment, it seems like such a little thing. 24:38 It can be much more profitable or just overall fucking easier to fudge a little data, to tell a better story. 24:47 Sometimes good people who are tasked with reviewing your work catch the flaws, your paper's rejected, nobody's impacted. Other times, we spend 24:57 two decades in a war, or two decades and billions of dollars chasing Alzheimer's indicators that aren't really there. Nobody made us do that. We are who we are, and the incentives are everything. 25:11 We would like to believe we designed systems like housing or banking or search or chatbots with a certain level of self-awareness. James Madison once said, "If men were angels, no government would be necessary. 25:25 If angels were to govern men, neither external nor internal controls and government would be necessary." Great. But James Madison also bought and sold Black people for personal profit. So, you know, not great. 25:41 So while the internet and search and chatbots are the result of everything we put into them, it's important to remember that, like the Federalist Papers, they're actual people choosing the data and writing the algorithms, designing the search box, and what comes out of it, and the ads and the bidding systems. 26:00 Throughout human history, we've always designed incentives to help us, to help other people, especially ourselves, to get what we want. 26:08 On the one hand, we wanna believe we're in the right, that we can trust each other, that the arc of history just bends towards progress. And in a lot of ways, we have done that. 26:19 But on the other, how can we believe any of those, much less all those, after all this time, when every synthetic, calculated embodiment of who we are has said the opposite? These new AI models and tools aren't sentient, 26:34 but they don't need to be sentient to be dangerous because they're the sum of us. They can be most dangerous because we believe they're more than they are. 26:44 We believe that because we are human, 'cause we have lizard brains. We don't ask important questions like, "What motivations might this nice caveman have to lie about me about that super dark cave over there? 26:56 What might he know that I don't know?" Or, "How exactly do these tools work?" Or, "Who trained the data, and on what data?" Or, "How are these tools different from each other or one another?" 27:10 Or, "How exactly are these companies going to pay for all these queries and the chips required to run them?" Like with many suddenly revelatory athletes or politicians, 27:25 it's often advisable to not, in fact, meet your heroes. I mean, just days ago, we heralded new Bing with a f-f-press conference. 27:33 It was underpinned by a historic and unlikely alliance between Microsoft and OpenAI, and we celebrated its swag and confidence. And then literally days later, new Bing started attacking people's marriages, 27:47 indiscriminately yelling at other people, generally questioning its own existence. And I mean, I gotta tell you, if that doesn't sound like the internet on a typical Wednesday, you're in a different timeline. 27:59 Congratulations. And yet we still believe. We believe what the simple Google search box gives us, even though it says ad under every link, 'cause we want self-affirmation, because we want that when what we need is facts. 28:16 But that ship seemingly sailed a long time ago. 'Cause the problem is we don't understand the difference, and that box isn't designed to give us some universally agreed upon fact anymore. 28:26 It is explicitly designed now to give us what we're looking for based on all of our prior information, and the question we type, assisted by autocomplete, is just one very small part of what we're looking for. 28:39 It's like the Spice Girls, right? The box is saying, "Tell me what you want. What you really, really want." 28:46 When there's clear danger of theft, we argue who's responsible for the content delivered to us over these tools and networks, right? The person who made the content or the delivery system that enabled us all to see it. 28:58 And that's a really, really important question. But what we really need to ask is, what do we really want out of this, out of these tools? How are they different? Search is not chat, and chat is not search. 29:12 I mean, for now. Yeah, there are a million new niche tools being constructed on top of foundational models. I mean, it's been going on for a few years now, really. 29:20 I mean, they build it, they shove a total addressable market or TAM into a deck, email it to every venture capitalist on TechCrunch. 29:28 But at the top level, we have to be extremely careful to understand we are talking about two very different primary tools. 29:35 Facebook and Twitter, and especially WhatsApp groups and YouTube recommendations, have all been tailored to you, giving you exactly what you asked for, even when you weren't asking for it. 29:45 All things considered, literally. GPT or new Bing or Bard are simply more of the same. We wanna believe there's a person on the other side of this chat interface or helping us with our code. 29:57 But that's exactly the problem. And it's been this way for a very long time. Because we didn't understand those terrifying storms 30:08 or why our cave children suddenly got sick and died, we invented gods, and we told stories to bring us comfort and companionship. 30:17 For thousands of years, we've believed we can kneel on the ground and talk to a supreme being who will give us answers, who will actually help us 30:26 in our time of need.Jump to 1997, we named one of those new gods Ask Jeeves, and then Yahoo, and then Google. 30:35 I mean, our want is right in front of our faces as we defend our choice to fight the flu with homeopathic bullshit, right? Because we say, "Well, Google says X." 30:46 Google doesn't say shit about shit except everything we've told it to say, leveraged for profit to within an inch of its useful life. 30:54 The problem is we continue to think and base our decisions on the fact that we think it's something different. 31:01 Your great-grandfather to the tenth power didn't wanna know list of safe caves in the 91004 Stone Age zip code, right? And ads to compete for that. 31:11 He was looking for a really quick answer to where can I sleep tonight where something won't eat me 'cause there's a terrifying sound in the sky and I wanna get away from it right the fuck now. 31:21 Unfortunately, incentives have always existed. His only resource was my great-grandfather to the tenth power who, like the rest of us, really only had his best interests in mind. 31:34 Because while those two grumpy old cavemen could agree that sound in the sky is fucking terrifying, only my guy knew that there's actually only one safe cave. He parked his shit in there days ago. 31:46 In the end, we will use whatever tool is available to us to find out the answer that is most convenient and reaffirming to us, and I mean, you know, great news. 31:56 That's an incredibly profitable business to be in when a company controls not only all your data history and the algorithms that mine it and the marketplace where other companies can bid to be first in line to be exactly what you're looking for, facts be damned. 32:12 Print money. 32:14 Now imagine the Google search box, but entirely, eerily, fluidly, compellingly conversational, personable even, who on the surface seems entirely trustworthy, but who has zero interest in actually understanding any of the concepts you're asking about, and exists solely to supply the next word in the answer you're looking for, and is doing so in a way it believes is the most human-like, which again, complicated, and it's based on everything we have ever submitted to be human-like. 32:46 That we have built this tool on top of decades and, you know, scanned properly centuries of user-generated content already is astounding. It is super cool. Like, take my breath away surprising to interact with. 32:59 You should totally try it. It is an achievement. It makes you question and wonder, what are we capable of? What is it capable of? What could be next? 33:10 And while it's definitely not sentient, yet it is very convincing, like that asshole in the bar from Good Will Hunting, 33:18 or like a misused, intentionally out of context Fox News soundbite in your old news feed in 2007, or like a single deepfake photo or video or audio soundbite can be today. 33:32 What the best conversationalists lack in accuracy, they make up for in charisma. Consider the TED Talk, right? 33:40 It's why it's so vital AI chatbots get the next word correct in the answer you requested, and why it's more important than providing you with an objectively correct answer. That's not the goal. 33:52 These new conversationalists are entertaining. They are suddenly riveting and emotional. 33:57 And history has proven that when all of that is true, we are far less likely to be interested in how correct they may be or what war they might lead us into. 34:08 I worry because of who we are, because we never wanna say it out loud. 34:13 I worry because half our electorate is, in part, increasingly uneducated, too expensive, broken, and for the rest, uninterested in doing any diligence whatsoever because they've spent twenty years increasingly just getting what they want. 34:28 That is, having their priors not only reaffirmed every time they get a goddamn notification, but having them enhanced to the extreme till they're no longer an individual, but an idea that can be advertised against. 34:42 And look, I'm not against advertising. I've said it before, I'll say it again. I have created, sold, profited from ads. There is an ad probably in this podcast. There was in the post. 34:54 Advertisements have uplifted businesses. They've introduced exciting new ideas. They've also doubled down on old ones like greenwashing, for example. 35:03 But when every business model from our six most powerful companies is exactly the same, when that business model is almost entirely predicated on collecting, storing, and selling the maximum amount of your information, 35:19 when those companies have more power and influence over you and the economy than we'd ever imagined, when they're so reliant on each other, when they're programmed to extract maximum financial benefit through a nearly instantaneous auction billions of times a day across the world, 35:37 when those auctions are most often conducted by a single company almost entirely reliant on the profit from those auctions, when there's like four other companies jealous of those profits for twenty years now, 35:52 then giving you an answer that provides you with anything less than exactly what you want is the least profitable answer. 36:03 Giving you an answer that does not jive with everything you've ever told it about you isn't a realistic option 'cause at scale, in every case, you're not asking a question of someone or something that has your best interests in mind, no matter how eloquently you- new tools respond to you. 36:21 Because giving you any answer costs money, but giving you the answer you want makes money. In 64,000 BC, sharing my cave, the only safe cave, eh, might cost me my life. 36:35 I don't know.Theoretically, sharing a cave might be profitable to both of us, however you want to measure it. 36:41 Kind of like healthcare companies helping you be healthy might cost you and our economies less than treating you when you're inevitably sick all the time, right? But we can't be sure about that. 36:51 Profits are questionable, unproven, and the work required to pivot to that model is simply too much. So we're going to stick with what got us here and just keep sending shareholder dividends. 37:03 The philosopher and adventurer Marcus Brody once tried to be the voice of caution and reason, telling a charismatic, power-hungry Nazi, you are meddling with powers you cannot possibly comprehend. 37:17 And what that Nazi fuck wanted was unlimited power. What he needed was to have his face melted off, and it's a movie, so there you go. Everybody wins. 37:27 Similarly, while it's true that the folks behind these models and tools are increasingly less aware of exactly how the tools get to whatever answer they supply you, it would be dishonest to say we do not comprehend these powers. 37:42 It is obvious these powers come from us. 37:46 They are the inevitable fruits of our frantic medical searches in the middle of the night, from our relationship updates, our peer-reviewed papers, non-peer-reviewed papers, blogs, artwork, company statements, financial results, baseball stats, porn, gadget reviews, check-ins at bars, photos, marriage announcements, obituaries, 38:10 ancestries, census data, our self-published fan fiction, world war histories, maps. I mean, it's all of it. 38:18 But we don't want to comprehend our part in this story because that would require an acknowledgement that we have lied to ourselves and each other from the very beginning, that we have faults, that we have failed, that we've always preferred the company people that look like us, people who have more than us, people that can help us, and the facts that are most convenient to us and stories that make us feel most seen and most comfortable and wealthy at whatever the cost to everyone else. 38:48 And so we will build a million new tools on top of those foundational models and on top of search, and we will confuse the two all in an effort to never question who we are and what we want. 39:02 We will build new tools that reinforce ancient biases, driving us further and further away from what we need. It was scary as hell to be any caveman out in a storm with nowhere to hide. 39:18 It was even scary to be my caveman who had what he wanted, safety for the night, and who thought he had what he needed. It's much, much scarier, however, to be your caveman and not for the reason why you think. 39:33 To trust my guy, this great conversationalist and the only other caveman around, to check out that dark cave he recommended right when you really needed it because he claimed to have all the relevant information and then to have a near escape with a tiger in that cave and to know immediately that that asshole, my great uncle Clyde Caveman, confidently gave your guy dangerously wrong info. 39:57 That's scary. What's even scarier is to know you probably would have done the same thing. I'm all for search. I'm all for chatbots. I fucking love new tools. 40:12 It is essential that we continue to innovate, to design tools that help us use far fewer resources, that reduce our administrative burden, that uplift more people, that make children safe, that make black people and Asian people and women and transgender and other LGBTQ plus people safe, that design better antibiotics and vaccines, that help us translate languages on the fly, that help tell our stories, that help balance out news reporting, that help us measure methane leaks and soil health. 40:44 We want to feel safe. We need to expand on what that means. We want to feel seen. What does that mean? 40:51 If we can build AI chatbots that are personable, that can transparently and candidly acknowledge what they don't know or what they're not sure about, that cite sources and help us do our jobs and understand data on the fly, 41:03 we might get there. We need real answers we can agree on. We need to understand what is broken and what we can build on. 41:12 If we can build a new search that is untethered to providing you with the most profitable answer, we might level the playing field against disinformation. We can have both if we understand the difference. 41:27 We can use both of these tools and someday maybe combine them to help build a more resilient, less hungry human experience that is radically better for far more people and even more people to come, relieving our ecosystems of the terrible burden 41:43 we have put on them. But if we aren't willing to ask hard questions about our wants and needs, to understand what we can change and what we can't, 41:54 we will only ever have the same ideas we've ever had by way of the same conversations we've ever had with the same kinds of people we've ever had, seeking the same answers we've always sought. 42:06 How can I use this to benefit me? And now in this week's news, in climate change news, an amazing new solar farm is also a prairie restoration project and a carbon sink. You're welcome. Also, let's hear for hedgerows. 42:21 Great. Harvard voted to build climate change right into its medical school curriculum, which is great. 42:29 In COVID news, new studies suggest that China's recent COVID death toll could top a million more people despite disinformation.And we're honing in on the idea that maybe long COVID is a neurological disease. 42:42 Question is why and how long does it last? In food and water news, sorghum is one of our oldest and most reliable grains and could be a stalwart in the age of climate change. Where can we grow it? 42:56 And can food be medicine after all? New policy from the Biden administration is trying to help unlock the studies we need to sort it out before we actually start to do that. 43:07 In health and bio news, it's probably helpful to understand that half of all American hospitals are projected to be in the red this year. That is not ideal. American teenage girls are suffering in a very real way. 43:21 There are a million things we can do, but helping them find purpose may be a start. E-bike incentives in Denver are so popular, they keep selling out. Your city should do those too. 43:33 And yeah, we need to get toxic gas out of our homes, but also, why is asthma the way it is? And new studies may point to why. 43:41 Lastly, in computer news, Tesla is recalling three hundred thousand plus cars because self-driving was bullshit all along. And supermarkets, like everyone else, are selling your data. 43:53 Please be careful signing up for that rewards plan. That's it for this week. Hit subscribe to get next week's issue straight to your feed. To go deeper, visit importantnotimportant.com/newsletter. 44:07 Thanks for being a part of our community, and thanks for giving a shit. Have a great weekend. 44:12 [outro jingle] [upbeat music] That's it for this week's conversation. 44:26 For more conversations, scroll back in the feed or visit podcast.importantnotimportant.com to search by name, topic, whatever. Thanks for sharing. Thanks for leaving a review, and thanks for giving a shit. 44:40 [outro jingle]