AI Could Soon Outnumber Human Researchers 20 to 1

Imagine if we could clone the world's best scientists and engineers millions of times over. That's essentially what one researcher predicts could happen with artificial general intelligence (AGI) — AI that can do any intellectual task a human can. According to the analysis, the computer chips sold by mid-2026 could power AI workers equivalent to 240 million full-time employees. That's 20 times the number of all human researchers today. If all those AI workers focused on research, one year's output would equal more than all academic research done since 1900.

To put that in perspective, today there are about 6 million academic researchers worldwide, and they've produced 150 million researcher-years of work since 1900. With AGI, we could have hundreds of millions of AI researchers running simultaneously. That's like having thousands of Apollo moon missions or tens of thousands of Manhattan Projects happening at once. Such a surge could lead to breakthroughs in medicine, energy, and technology at a pace we've never seen.

But there's a catch: research needs more than just brainpower. It requires equipment, labs, factories, and physical materials. For most big projects, researchers are only a small part of the cost — the rest goes to building and running things. So even with millions of AI researchers, progress could be bottlenecked by our ability to build labs and manufacture goods. Physical production can grow, but not as fast as AI numbers.

Still, some fields like mathematics or software don't need much physical infrastructure. In those areas, AI could race ahead. The bottom line: AGI could massively accelerate research, but the real world might not keep up. This means we could see incredible advances in some areas while others lag. It also raises questions about jobs, economic disruption, and how we manage such rapid change.