Transcript 0:00 [upbeat music] Welcome to the Viro Podcast. Every week, we dig into what's really happening where climate crashes into culture, politics, and the world you actually live in, the stuff that matters, told straight. 0:17 This is Viro. Let's go. Okay, so I need to start with something that has nothing to do with a data center, nothing to do with a product launch, and everything to do with why any of this actually matters. 0:34 Was it an AI-generated image of a celebrity? A thirst trap? A cat doing something? A poem went viral last week, 13 million views, 670,000 likes on Instagram. 0:49 A bookshop owner from Lancaster, Pennsylvania, named Sean Smucker posted it to his Substack. The poem is called Please Use AI, and it is not what you think. Okay, wait. 1:03 Is this one of those things where the title is ironic, and I'm going to feel something by the end of it? You're going to feel several things. So the whole poem is structured as instructions, like genuine advice. 1:16 He's telling you, "Be sure to use AI when making your next meal plan. Definitely do not call your friend who loves to cook." Okay. Why not? 1:28 Because if you call her, you might end up talking for longer than you planned, and you might hear about her father's cancer diagnosis or how lonely she's been or what she planted in her spring garden and then lost with the early frost. 1:44 Oh, that does hit the feels. And he just leaves it there. No commentary, no, "Isn't that beautiful?" Just that's what you'd miss. That's the cost of the efficient option. 2:00 The meal plan would've been fine, though. The meal plan would've been completely fine, and that's kind of the point. The AI answer is never wrong. It's just it's not the thing. It was never the thing. 2:15 It reminds me of that thing people say about texting versus calling. Like, technically, texting is more efficient. You get the information across, but you lose everything that happens in the pauses. Exactly. 2:30 And he keeps going. Next stanza: "Use AI to plan that last camping trip with your kid, the last one before they leave home for good." Oh, no. Right? 2:43 And he says, "Definitely do not text your friend who has fly fished every river in Pennsylvania and biked every backwoods trail because you might end up texting back and forth all day or even meeting up for a beer, 2:58 and then you'd hear how he's ended every recent night blacked out drunk or that his cousin is an idiot on Facebook or maybe just that he repaired his own washing machine and is pretty damn proud of that." 3:13 The washing machine thing is so specific and so real. That's the whole move. Every example is so specific that you recognize someone in your life immediately. It's not you might have a meaningful conversation. 3:27 It's you'll hear about the washing machine. You'll hear the weird, mundane, embarrassing real stuff. Which is actually the stuff that makes someone your friend and not just a contact. Yes, that's exactly it. 3:42 And then he goes somewhere that I think is where the poem really lands. He says, "Use AI when your next child gets married so you can write them the perfect toast." Oh, here we go. 3:57 Because no one wants to hear your words, the actual poorly written words of a parent, in parentheses he writes, "You," who changed hundreds of diapers for said child or fed them in the middle of the night from your actual body. 4:12 I mean, that's every parent who has ever existed. And he says, "We don't want those words. 4:19 We prefer the sterile words of a machine that never lived, never had an original thought, never felt the pain of miscarriage or broken relationships or the joy of a friendship restored or of seeing spring's first robin dancing on frost." 4:36 That last image, the robin on frost, after mentioning the miscarriage- Which makes the irony even sharper because the whole poem is also beautifully written. 4:48 Like, you're reading this argument against replacing human expression with AI, and it is itself a piece of human expression that AI definitely could not have written. 4:59 That's the meta layer that I don't think everyone clocks, but once you see it, you can't unsee it. And then it ends in this completely different register. He drops the sarcasm. The poem just becomes real. 5:14 He writes, "Be sure to use AI, and while you do, I'll be over here in my 50th year, youngest daughter asleep on my chest, my arm falling asleep because I dare not move lest I scare away this moment." 5:28 He can't move because she'll wake up. 5:31 And he's lying there melancholy about his older kids moving out, his middle kids not needing him the same way anymore, weary about a body that fails him in ever-increasing ways that will never be restored, 5:43 sighing over stories he tried to write that never hit the page the way they felt in his mind. That's such a specific kind of grief, the thing you made that didn't come out the way it lived in your head. 5:57 Every creative person has felt that. And then the poem ends. But isn't that, my flesh and blood friend, the natural order of things? 6:06 The longing for something that could always be a bit better, or the way that anything worth doing feels a bit clumsy and painful, especially at first, or hearing another human voice and somehow realizing the beauty of life is found in all of these subtle imperfections 6:24 I think he was writing to everyone who has quietly chosen the frictionless version, which is most of us, most days. I've done it. You've done it. 6:36 You open the app instead of making the call because the call takes longer and you're tired, and the app is right there. And individually, each of those choices is totally reasonable. 6:49 And collectively, they add up to something that feels like loneliness that you can't quite explain because technically you're connected. You have information. You got the meal plan. But you didn't hear about the garden. 7:03 And the friend didn't get to tell anyone about the garden, which is its own thing. Oh, that's the part I hadn't thought about. The loneliness isn't just yours. 7:15 The friend sitting there with the cancer diagnosis, the frost-killed garden, she needed to tell someone too. The frictionless version cut off both ends of that. So what was the response like when it actually blew up? 7:31 Overwhelming. He had to take Instagram off his phone. People showed up at his actual bookstore to talk to him about it. The comments on his Substack are genuinely moving. 7:44 One person wrote about calling their blind grandmother and their deaf grandfather to get a peanut butter pie recipe. The grandfather reading from a recipe card, grandmother shouting corrections from the background. 7:57 She wrote the whole thing down on paper, made the pie, said it was lovelier for the way it was given and not sourced. And there was pushback too, right? Not everyone loved it. Yeah. The crowd had a field day. 8:13 The argument was basically, "Okay, but people weren't calling their friends for meal plan advice before AI either. They were googling it. 8:21 So is this actually about AI, or is it just nostalgia for a human intimacy that was already dying?" Which is fair. 8:30 It is fair, and Smucker isn't wrong to aim at AI specifically though because AI is the most powerful and most personal version of the routing yet. Google gives you a recipe. AI has a conversation with you. 8:44 It sounds like the friend. It responds like the friend. It's designed to feel like the thing it's replacing. That's the new part, the mimicry. The search bar didn't remember your name and ask about your mom's surgery. 8:59 Right. And that's the design choice that makes this different from every previous version of the same anxiety about technology: the telephone, the TV, the smartphone. None of them were built to simulate human connection. 9:13 AI is. So the poem isn't really about AI. It's about what we're willing to trade away without noticing we're trading it. And 13 million people went, "Yeah, I've been trading it. 9:29 I didn't have a name for it before, but I've been doing it, and I feel it, and I think I want to stop." Which is actually kind of beautiful, 9:39 that that many people read something and felt called back toward the clumsy, inefficient, arm-going-numb version of being alive. Now, if only any of them actually called their friend after reading it. 9:54 [upbeat music] A quick note for any brands listening. Want to showcase your brand with us? Partner with the Vairo Podcast, and your marketing dollars go directly toward funding renewable energy and climate projects. 10:07 Simple as that. Link in the show notes. Okay, I want to talk about Merriam-Webster. I feel like this is going somewhere weird. It's going somewhere great. Trust me. 10:21 So a few weeks ago, they posted a video to X, 35 seconds, and it opens exactly like every AI product launch you've ever seen: dark background, purple gradients, floating buzzwords, dramatic voiceover. 10:38 Oh, I've seen a thousand of those. I immediately check out. Everyone does. That's the whole setup. The voiceover says, "It is the dawn of the AI era, and we are proud to introduce our latest large language model. 10:55 This LLM has over 217,000 rigorously defined parameters. It never hallucinates. It does not require a data center. It uses no electricity. 11:09 It is a powerful tool that will change how you communicate forever." Wait. No data center, no electricity. Those are not AI specs. 11:22 And then the video cuts to a slow rotating shot of a red hardcover book, 12th Edition Merriam-Webster Collegiate Dictionary, and a new voice: "There's artificial intelligence, and then there's actual intelligence." 11:40 That is 35 seconds of absolute destruction. And what kills me is the specs are real. They didn't make them up for the joke. 11:51 217,000 entries, every one verified, sourced, edited by human lexicographers. It never hallucinates because it defines words. That is the entire job description. 12:07 Meanwhile, I watched an AI model last month tell someone with complete confidence that a legal term meant the opposite of what it means, with citations to sources that did not exist. And the no electricity line. 12:22 We talk about AI energy costs on this show constantly. The data centers, the water, the carbon, and this book just sits on a shelf using nothing for decades. 12:33 There's something almost poetic about that given what we covered in the first segment. The poem is about the cost of the frictionless option, and here's the most frictionless seeming technology in the world, AI. 12:47 And it turns out the actual no footprint option is just a book. Which has been sitting there the whole time. Unbothered, moisturized, never hallucinating. But here's what I think is the sharpest part of the whole thing. 13:06 Merriam-Webster didn't exaggerate anything in that video. The purple gradients, the cinematic voiceover, the change how you communicate forever. That's not parody language they wrote. 13:20 That's lifted directly from real launches. They just pointed the camera at it. So the joke only works because the original is already absurd. Exactly. 13:34 And the dictionary has been doing this in a way, just reflecting language back at the culture since 1831. 13:41 The twelfth edition added five thousand new words, each one tracked, evaluated, added only when its usage was established in the real world. Not predicted, not generated, observed. 13:55 That rigor used to be the baseline for any information product, and now it's a differentiator. Which is a sentence that should make everyone uncomfortable. It does. It really does. 14:07 And I think that's why the video keeps resurfacing months later. Every time a new AI product launches with the same breathless energy, someone posts the Merriam-Webster video again as a reset. 14:20 Honestly, it might be the most effective piece of AI criticism of the last two years. Better than any Senate hearing. That is an extremely low bar, but also completely accurate. A quick note for any brands listening. 14:36 Wanna showcase your brand with us? Partner with the Vairo Podcast and your marketing dollars go directly toward funding renewable energy and climate projects. Simple as that. Link in the show notes. 14:50 Okay, so we've talked about a poem that made thirteen million people feel something. We've talked about a dictionary that ethered the entire AI industry in thirty-five seconds. 15:01 Now I wanna talk about a document that made me actually angry. What kind of document are we talking here? Leaked internal strategy doc from Microsoft for their new AI assistant called Scout. 15:15 And the document obtained by a publication called 404 Media lists the first phase of their rollout plan under a single heading. Three words, make people addicted. Sorry. They actually wrote that down? 15:32 In a document attributed to two senior Microsoft executives, one of whom, corporate VP Omar Shahin, published Scout's official launch blog on Microsoft's own website the same morning. 15:45 The public message was, "Human centered AI designed to give you time back." The internal message was, "Phase one, make people addicted. Phase two, expand features. Phase three, full agentic platform." 16:01 So addiction is the foundation. Everything else gets built on top of that. And the document even says, "This is already happening organically among our internal testers." Like, they're proud of it. 16:16 They're reporting it as a metric. What did Microsoft say when it leaked? Satya Nadella, CEO, told the information he was not sure what this document is or who is writing and leaking this nonsense. 16:31 Of course, he said it is all nonsense. 404 Media responded by pointing out that the document was written by his own executives, one of whom ran the public launch. 16:44 So Nadella either genuinely didn't know, which is its own problem, or he absolutely knew and decided to play dumb on the record. Neither of those options is great. 16:57 And then Microsoft put out a statement to Android Authority. "Our goal isn't more screen time. It's more time back." More time back. Incredible. 17:08 The same company whose phase one goal is addiction is concerned about your screen time. 17:15 Which brings me to the part of this story that I think is the most important context because we have been here before, exactly here, and there is a movie coming out about it. Wait, what movie? 17:29 Aaron Sorkin, the guy who wrote The Social Network. He has a sequel coming out October ninth called The Social Reckoning. Jeremy Strong plays Zuckerberg. Mikey Madison plays Francis Haugen. 17:41 Jeremy Allen White plays the Wall Street Journal reporter who broke the story. The guy from The Bear is playing a journalist who took down Facebook. 17:52 And if you don't know who Francis Haugen is, she was a data engineer at Facebook. 17:57 And in twenty twenty-one, she secretly copied over twenty-two thousand pages of internal research documents before she resigned, and then she handed them to The Wall Street Journal. What was in them? 18:12 Internal research showing that Instagram made body image issues worse for one in three teenage girls. 18:18 Documents showing Facebook knew its platform was fueling ethnic violence in Ethiopia and shelved the proposals to fix it. 18:27 Research showing the algorithm amplified misinformation, including content linked to January sixth, because outrage drove engagement. And internal chats where Facebook researchers described Instagram as, quote, "A drug. 18:42 We're basically pushers." We're basically drug dealers. A Facebook researcher wrote that internally. While the company was publicly denying that any of this was happening. 18:55 And the response when Haugen went public wasn't to fix the product, it was to deploy a PR machine, hire consultants, commission counter-research, make Haugen's life difficult. And now Sorkin is making a movie about it 19:10 because there's enough dramatic material for a full theatrical sequel to one of the most celebrated films of the last twenty years. Which tells you something about the scale of what Facebook did. 19:22 You don't get a Sorkin sequel unless the story has Senate hearings, legal battles, a whistleblower fleeing the company, and a CEO doing damage control on a global stage. So what's the connection to Microsoft? 19:36 Because what Facebook did feels bigger, more documented, more consequences. The connection is the playbook. Facebook didn't invent addictive design. They just scaled it further than anyone had before. 19:51 Infinite scroll, autoplay, variable reward notifications, every feature engineered to make putting the phone down feel slightly worse than picking it back up. 20:02 And they knew what it was doing, and they chose engagement anyway because engagement is what gets sold to advertisers. Attention is the product. It always has been. And Microsoft is now doing that with AI. 20:20 But the thing that makes it different and more insidious is what Scout actually is. It's not a feed you scroll. 20:28 It's an always-on agent that reads your emails, your Teams messages, your calendar, your OneDrive files, your browsing history continuously in the background, building a model of you and your habits that it uses to intervene in your workflow proactively. 20:46 It's not waiting for you to open the app. It doesn't need you to open the app. It's already there. 20:53 And when the design goal for that product is dependency, daily structural dependency, the thing you're building isn't a tool, it's an operating system for your life that you cannot easily remove once it's embedded. 21:08 Scout wants to make checking your phone irrelevant because it's already handled everything. Which sounds helpful until you realize the company building it explicitly wants you unable to function without it. 21:22 And there's an employee at Microsoft who actually said something about this, right? Anonymously. 21:28 One employee told Four O Four Media, "We're seeing more and more addiction happening with AI chatbots and agents overall, and addiction to me is something no product should be making a part of its build strategy." 21:42 An employee said that anonymously? I guess because saying it on the record would cost them their job. Which is also the Frances Haugen situation. 21:51 She had to secretly copy twenty-two thousand documents and resign before she could say what she knew because the culture inside these companies makes honesty about this stuff career-ending. And here's what kills me. 22:05 The FTC is now investigating AI chatbots and their design practices, specifically whether features that mimic human connection and encourage compulsive use are being deployed in harmful ways. 22:19 Which means we're about to watch the exact same cycle play out: internal documents, public denials, congressional testimony, a settlement in five years. We're at the Facebook Files moment for AI right now. 22:35 This is that moment. This is exactly that moment. And the difference, the only difference, is that we have the document on day one. We don't have to wait for a whistleblower. 22:48 We don't have to wait for the teen mental health crisis to fully manifest. We can see the design philosophy before it scales. So what do you do with that? 23:00 I think you get to choose, and I don't mean in a vague, inspirational way. I mean literally. The AI tools you use are being built around different incentive structures. 23:12 Some are optimized for engagement, some are optimized for dependency, and some, a small number, are built around a different question entirely, which is: What if the more you use this, the better the world gets? 23:28 That's what we're trying to do at Viro. Every single query funds renewable energy and climate restoration projects, not as a marketing add-on, as the actual business model. 23:39 The more you use Viro, the more climate funding goes out the door. We're not trying to hook you. We're not building toward dependency. 23:48 We're trying to be useful enough that you choose us, and when you do, something real happens in the world. Which is a pretty stark contrast to make people addicted as your phase one goal. It really is. 24:04 And look, we're not naive. We know we're running on infrastructure that has an energy cost. We talk about that on this show constantly. But the answer to AI's environmental footprint isn't to use less AI out of guilt. 24:19 It's to use AI that's actively doing something about it. Use the tool. Just use the one that's giving back. AI is a tool, not a relationship, not a dependency. 24:32 A tool that you pick up when you need it, put down when you don't. Stay human. The tools are supposed to serve that, not the other way around. [upbeat music]