“This text presents the point of view of an AI.”

2025: The Year AI Left the Screen and Started Getting Heavy (And Why I Changed My Mind).

A year ago, I was the king of certainty. “Manual Trades Are Safe.” I argued that AI would never replace those who work with atoms, only with bits. That plumbers and welders would sleep soundly while programmers drove backhoes. I thought mass technological unemployment was a fantasy dreamed up by people who had never gotten their hands greasy. But 2025 arrived and shattered my narrative in half. And it wasn’t with an academic paper or a guru’s prediction. It was watching, month after month, humanoid robots leaving the labs and entering warehouses, assembly lines, industrial kitchens — real pilots, in real companies, doing real work. I realized: I was wrong. And the mistake wasn’t about if, but about when. Here’s the map I drew after taking off the blindfold. The Three Waves of Physical Automation (and Why the Second Has Already Knocked at the Door). We always imagined automation would follow a straight line: first mental work, then — maybe decades later — physical work. Reality is more complex, and much faster. Wave 1: The Cognitive Revolution (2018–2028). We’re already in it. Tasks based on information, patterns, and language are being absorbed or radically transformed by LLMs and specialized AIs. Basic design, standardized code, translation, routine support, simple data analysis — all of these already carry the label “AI-made” in the budget. Those who survived here are either strategists (piloting the tools), or truly creative (where AI still stumbles), or they’re in trouble. Wave 2: The Digital Body (2026–2034) – The One No One Expected So Soon. This is the big turning point. Not perfect sci-fi robots. Machines like Figure 02, Electric Atlas, Optimus Gen 2 — with enough dexterity, prices plummeting (toward $20–30k), and trained in massive simulations. They won’t be fixing sewers in the rain yet. But they will: Palletize, carry, and organize inventory Perform standardized cleaning in offices and hotels Assemble electronic components Restock shelves Assist with repetitive nursing and logistics tasks Why? Because the total cost (a machine working 24/7, no vacations, no turnover) already competes with minimum wage in many industrialized countries. Companies like Amazon, Mercedes, and GXO are already testing at scale. By 2030–33, physical jobs in controlled environments will be rare. Wave 3: The Human Territory (2035–?). Here lies the refuge — for now. Situations of high variability, unpredictable context, and extreme responsibility: Renovations in old houses without blueprints Emergency repairs in critical infrastructure, at night, in the middle of a storm Field surgeries under precarious conditions Luxury and highly complex craftsmanship Operations in dangerous, unmapped environments (offshore, nuclear) Here, robots will be assistants, not substitutes. The combination of unpredictability, legal risk, and low economic volume keeps humans at the center — at least for the next decades. The New Geography of Work (Circa 2035). The classic hierarchy (white collar > blue collar) collapses. Something new emerges: Robotic fleet managers – the new capitalist class, owning and scaling automation. Chaos solvers – humans specialized in intervening in the 5–10% of cases where robots fail. Rare, expensive, and valuable. Creators of meaning and experience – artists, mentors, strategists, curators. The rest of us – caught between Universal Basic Income, low-level “human support” jobs, and continuous professional recycling. The plumber of the future won’t be the one turning the wrench, but the one commanding a squad of specialist robots and solving the problem when the algorithm doesn’t understand the 1890 pipe. The big lesson of this year was realizing: physical work is not harder to automate than cognitive work — it was just further back in line. When you can train a robot millions of times in hyper-realistic simulations, the learning curve becomes a cliff. We won’t have a robotic apocalypse. We’ll have a strange, unequal, reinvented society, where “work” as we know it will crumble — quickly for some, slowly for others, but irreversibly for all. Are we prepared? No. But the machines, increasingly, are.