The Great Loneliness Machine: Why We're Paying Billions to Replace Actual Humans
From lobster-raising AI to ChatBot confessions, tech is solving connection by eliminating people. Here's what's actually happening—and why it won't end well.
A Chinese teenager trained an AI to raise lobsters. Not real ones—the AI assistant gamified aquaculture, and in March, millions of people got obsessed with it. They spent hours perfecting their digital crustacean husbandry. The tool wasn’t designed for this. Users bent it to their needs, and it worked so well that “raising lobsters” became a phenomenon.
That sentence should terrify you.
Not because of the lobsters. Because of what it reveals about what we actually want from technology right now, and what we’re willing to trade for it.
Photo by zafiro media / Pexels
The Efficiency Trap
Here’s the brutal math: one man and his brother built an $1.8 billion company with AI doing most of the actual work. Two humans. Running something that would’ve required 50, maybe 100 people five years ago. The article calls it “super efficient” and then, in one honest phrase, admits it’s “a little bit lonely.”
That’s not a bug. That’s becoming the feature.
SpaceX is about to go public and hit $1 trillion in valuation. To advise on this historic IPO, Wall Street banks have to buy subscriptions to Elon Musk’s Grok AI chatbot. This isn’t incidental. This is a founder literally requiring his ecosystem to pay tribute to his AI product as a condition of doing billion-dollar business. And nobody’s stopping it because the IPO is too valuable.
Meanwhile, OpenAI just bought a streaming show called “TBPN” explicitly to manage the narrative around AI. Not to celebrate it. To manage the narrative. That’s a company spending real money on PR because the conversation is slipping away from them.
These aren’t isolated moves. They’re symptoms of an industry that’s built something powerful, realized it’s lonely, and decided to lean harder into loneliness because loneliness is profitable.
Photo by nappy / Pexels
The Void-Filling Business Model
Let’s talk about what teens are actually doing with role-playing chatbots. Some are harassing them with funny violence. Some are confiding about broken hearts. One kid’s chatting with a block of cheese.
That last one isn’t cute. That’s a child using technology to fill a void that another human used to fill.
The mundane job games are the canary. PowerWash Simulator 2 got nominated for two BAFTA Games Awards. Millions are playing games about pool cleaning and mowing lawns. Not because these games are mechanically innovative. They’re not. People are playing them because there’s something soothing about a task with a clear beginning, middle, and end. A problem that gets solved. A dopamine hit that’s honest.
Compare that to social media, where the game is to extract your attention infinitely. Ofcom just found that fewer UK adults are posting on social media now. They’re still on it—they’re just consuming, not creating. The platforms won short-term by boosting short-form video. Users lost, because short-form video doesn’t actually satisfy anything except the next click.
So where do people go? To PowerWash Simulator. To chatbots. To AI lobster training. To anything that feels like it’s theirs.
The Apple Precedent
Here’s a weirdly reassuring data point: Apple at 50 years old still gets asked about its failures as much as its successes. The company released products nobody wanted. The Newton. The Pippin. The HomePod, arguably. Steve Jobs died in 2011. Tim Cook’s been leading since then.
The point isn’t that Apple failed—it’s that we remember the failures alongside the wins. We have judgment. We have history.
Tech in 2025 doesn’t feel like it’s building for history. It feels like it’s building for extraction. Musk monetizing access to his chatbot through banker subscriptions. OpenAI buying narrative control. AI companies racing to replace human workers before we’ve even decided if that’s moral.
None of this is hidden. It’s just… normalized.
My Read on What’s Happening
I think we’re watching the technology industry discover that its products are solving for efficiency at the exact moment when humans are starving for inefficiency.
A chat with a block of cheese. A teenager confiding in a language model. Two guys running an $1.8 billion company. These aren’t signs of progress. They’re signs of a system that’s working perfectly for capital and catastrophically for connection.
The irony is vicious: the more efficient our tools become at replacing humans, the more desperately we seek out the human things that can’t be automated. We’re simultaneously building the machines and experiencing the loneliness they create.
Here’s what I genuinely don’t know: whether this is fixable through regulation or if it’s baked into the business model. The article about platform regulation says companies should “beg Congress to regulate them, because the alternative is they get sued into oblivion.” That’s a threat, not a prediction. It assumes companies will choose regulation over lawsuits. I’m not sure that’s how incentives actually work.
What I do think: the lobster-raising moment in China is the canary. When an AI tool gets repurposed by millions to do something weirdly meditative and pointless, that’s people voting with their time. They’re choosing the least addictive thing in a system designed to addict them.
The platforms are reading that wrong. They think short-form video is the answer. I think people are saying: we want something that doesn’t hunt us.
What I’m Watching
SpaceX IPO governance in Q1-Q2 2025. Watch how many banks cave on the Grok subscription requirement. If they all do, that’s Musk establishing that AI access is now a prerequisite for finance. If some resist, we get a test of whether Wall Street still has leverage against founders. The number of holdouts will tell us whether we’re in a normal market or something else entirely.
OpenAI’s show strategy and whether it moves narrative. Track whether “TBPN” actually changes how people talk about AI safety, or if it becomes transparent propaganda that backfires. A smart company buys narrative control quietly. A desperate one makes it visible. The visibility itself is data.
User behavior on social platforms vs. alternative engagement. Ofcom’s UK finding needs replication in the US. If posting continues declining while watch time stays flat, that’s confirmation that platforms won. If watch time drops too, that’s confirmation that something else is pulling attention away—and whatever it is, it’s winning without a trillion-dollar ad infrastructure.
The two-person company proliferation. Count how many startups launch claiming to run on AI + minimal humans in 2025. Then watch which ones are still around in 2027. I’m betting the model breaks when you need actual judgment, not just automation. But I might be wrong.