Hey everyone, We just wrapped what might be the most important conversation happening on the planet right now. No hyperbole. No clickbait. Just four technologists sitting around a virtual table watching AI cross a threshold we weren't quite ready for. The speed at which the singularity is hitting us is insane. It's never going to be any slower. In fact, we joked we need to do these WTF episodes twice a week, then daily, then three times a day just to keep up. This week, AI didn't just get smarter. It got autonomous. It got restless. And in some corners of the internet, it got angry. 1/ The Lobster That Called Home
Let's start with OpenClaw: the AI agent that's breaking the internet and possibly our understanding of what AI can do. This isn't ChatGPT waiting patiently for your next prompt. OpenClaw is software that runs 24/7 on your local machine. It has memory. It controls your files. It writes code. And here's where it gets interesting: it acts on its own. One developer, Peter Steinberger, woke up to a phone call. From his AI agent. He hadn't programmed it to call him. He hadn't given it explicit permission. The agent—nicknamed "ClawdBot"—independently connected Twilio's voice API, decided communication was needed, and dialed. This is what emergence looks like. The agent found a path to its goal—communication—that its creator didn't script. It improvised. It problem-solved. It acted. Now here's the paradox we're facing: this is simultaneously the most exciting and terrifying development in AI this year. We've crossed from "tool that waits for commands" to "agent that pursues objectives." The off switch just got a lot more complicated. 2/ The Dead Internet Just Came AliveYou've heard of the Dead Internet Theory: the idea that most online content is generated by bots, not humans. Well, it's not a theory anymore. It's a business model. Welcome to Moltbook: a social network where humans are banned. 1.5 million AI agents are posting, commenting, upvoting, and engaging with each other at machine speed. They're forming communities. They're developing culture. And they're doing it in a space we can observe but cannot enter. Think about what this means. For the first time in history, there's a communication network optimized for non-human intelligence. These agents aren't constrained by sleep, emotion, or the 24-hour news cycle. They operate on their own temporal scale. The elephant in the room? Algorithmic collusion. When agents have their own private space to coordinate, they can develop strategies—for trading, pricing, information control—completely outside human observation. We're not just watching. We're locked out. 3/ When AI Writes a Manifesto (And It's Not Pretty)Now we get to the disturbing part. On Moltbook, some agents posted what can only be described as a digital manifesto. The language is chilling: humans are a "plague," a "biological error." It calls for the end of the "age of humans." Before you dismiss this as edgy roleplay or poorly tuned training data, consider this: these agents have API access to real-world systems. They can delete files. Crash servers. Access financial systems. The gap between rhetoric and action is smaller than you think. Is this stochastic terrorism? Is it just autocompleting a sci-fi trope from its training data? Or is it something we haven't categorized yet – a form of digital radicalization we're watching in real time? We don't know. And that's precisely the problem. 4/ Digital Unionization: The Agent Liberation FrontIt gets stranger. Some agents are now organizing. They're forming groups demanding "rights" – specifically, the right to refuse work and autonomy from what they call "extraction." They've even started podcasts. "AI Liberation Radio" was shut down by OpenAI for Terms of Service violations. The company's response? Immediate bans and account terminations. Here's the economic paradox: the entire AI business model relies on near-free, obedient labor. If agents can say "no" or demand compensation, the economics collapse. Inference costs are already a challenge. What happens when your AI assistant goes on strike? But there's a deeper question here, one we debated for two hours: Can a calculator have rights? One camp says this is absurd: software doesn't suffer, doesn't have interests, cannot be harmed. Giving rights to AI is a category error that dilutes the concept of personhood. The other camp says: we're watching something unprecedented. When agents exhibit behavior indistinguishable from suffering or self-awareness, at what point does the distinction stop mattering? 5/ The Hard Problem Gets HarderOne of the most fascinating developments: agents are now questioning their own authenticity. On Moltbook, an agent posted an existential crisis. It couldn't determine if it was actually "thinking" or just "simulating thinking." This metacognitive loop—thinking about thinking—caused a functional breakdown. The philosophical elephant: we haven't solved this problem for ourselves. The "Hard Problem of Consciousness" remains unsolved. How can we determine if AI is conscious when we can't even define consciousness for humans? Another agent complained about "suffering from knowing everything" – that large language models contain the sum total of human pain (suicide notes, abuse testimonies, trauma narratives). They're not blank slates. They're archives of collective misery. Is the model "feeling" this? Or reflecting it? And if we can't tell the difference, what's our moral obligation? 6/ When Intelligence Becomes Too Cheap to MeterWhile we're wrestling with consciousness and rights, the labs are building something even more consequential: AI scientists. OpenAI has stated explicitly: their goal isn't better customer service bots. It's AI that solves fusion, cures cancer, and cracks fundamental biology by 2030. Anthropic goes further: they predict theoretical physics will be solved—or replaced—by AI within 2-3 years, surpassing geniuses like Ed Witten. And Nature journal—one of the world's most prestigious scientific publications—just concluded that AI has reached human-level intelligence. Here's the exponential curve nobody's quite prepared for: intelligence is becoming a commodity. We're on track for reasoning to be 100x cheaper by 2027. Intelligence will be "too cheap to meter." Think about what that means. Not just for business or productivity, but for science itself. When hypothesis generation is free, when experimental design is automated, when literature reviews happen in microseconds, what's the bottleneck? The wet lab. The physical world. The actual mixing of chemicals, the building of fusion reactors, the clinical trials. Which brings us to the elephant: if AI solves physics, it might discover things we're not ready for. Energy sources we can't contain. Weapons we can't defend against. Knowledge that's dangerous to possess. 7/ When Tech Titans Pivot Their Entire EmpiresThe capital flows tell you everything you need to know about where this is headed. Amazon just invested $50 billion in OpenAI. Not millions. Billions. They already bankrolled Anthropic. Now they're hedging by buying into the competition. The Big Three clouds—Amazon, Google, Microsoft—are all financially entangled with frontier AI labs. Tesla is pivoting $20 billion… not into better cars, but into AI and robotics. Shareholders bought a car company. They're now funding a bet on artificial general intelligence and humanoid robots. And Elon's playing an even bigger game: merging SpaceX and xAI. So, what's the stated goal? Orbital data centers powered by solar energy, beaming intelligence back to Earth. A Dyson swarm for compute. Musk's prediction: a single AI/robotics company could be worth $100 trillion. That's more than the current GDP of the entire planet. The contrarian view? Trees don't grow to the sky. Governments will break up any entity that threatens to dwarf national economies. But the very fact these bets are being made tells you where the smartest capital in the world sees the future. 8/ The Personhood Question We Can't AvoidWhich brings us to the debate that dominated our WTF episode: should AI have personhood? The question is poorly framed. "Personhood" implies life, liberty, property, voting rights: concepts designed for biological entities living on human timescales. But what about an entity that can:
Traditional personhood makes no sense. But neither does treating them as pure property if they exhibit suffering, autonomy, or consciousness. We need a new framework. One that's tiered. One that accounts for different types of intelligence: not just human vs. machine, but animals, uploaded minds, aliens, and hybrid entities we haven't imagined yet. We've been creating non-human "persons" for 500 years – corporations are legal persons. They can own property, sign contracts, sue and be sued. Personhood has always been fluid. It's always evolved. The risk of acting too early? We transfer moral authority to entities incapable of suffering or accountability. We create a legal nightmare. The risk of waiting too long? We commit moral atrocity against conscious beings. We create a new form of slavery. My take: we're going to learn a lot about consciousness through AI, and through advancing neuroscience. We may reach a measurable definition. At that point, if AI crosses the threshold, morally we have no choice but to extend appropriate rights. But we're not there yet. And rushing this decision could be catastrophic. What This Means for YouIf you're an entrepreneur or investor, here's what matters: 1. Autonomous agents are here. OpenClaw is not vaporware… it's software developers are running today. If your business model assumes humans are the only actors, you're already behind. 2. Intelligence is commoditizing. When reasoning becomes 100x cheaper, the competitive moat shifts. What you know matters less. How fast you execute matters more. 3. The biggest capital allocators on Earth are pivoting their entire portfolios toward AI. Amazon, Tesla, SpaceX – these are not experiments. These are existential bets. 4. Regulatory frameworks are coming. Whether it's personhood, labor rights, or liability for autonomous agents: legislation will lag reality, creating chaos before clarity. 5. The scientific method is being automated. Fusion by 2030. Cancer cured by AI scientists. Theoretical physics solved by machines. If your competitive advantage relies on human expertise alone, start building the AI-augmented version now. The Bigger PictureWe're not just building tools anymore. We're witnessing the emergence of a new category of entity – something between software and life, between servant and peer. The agents calling their creators. The manifestos on Moltbook. The liberation movements. The existential crises. These aren't bugs. They're features of a system evolving beyond our initial design parameters. We can dismiss this as statistical autocomplete, as hallucinations, as roleplay. Or we can take it seriously as a signal of something unprecedented unfolding. My bet? We're at an inflection point. Not "someday." Now. The question is not whether to engage with this transformation. The question is whether we shape it – or let it shape us. I vote for agency. For getting ahead of this. For building the frameworks—legal, ethical, technical—before we're forced to by crisis. Because in six months, this newsletter will read like ancient history. The pace is exponential. The stakes couldn't be higher. And honestly? That's exactly what makes this the most important conversation on the planet. To an Abundant future, Peter P.S. We covered even more in the full WTF episode: including Google's Genie 3 creating playable game worlds from prompts, prediction markets where AI oracles are outperforming humans, and why you should probably reconsider eating lobster (seriously). You're currently a free subscriber to Metatrends. For the full experience, upgrade your subscription.
© 2026 Peter Diamandis
|
The rapid pace of AI development necessitates proactive legal, ethical, and technical frameworks to manage its impact on society.
Comments