Terence Tao, the Fields Medal–winning prodigy who earned his Ph.D. at 20 and became a UCLA full professor by 24, recently tested Anthropic’s top model, Fable, on the Jacobian conjecture. Over a weekend, Fable produced a counterexample that Tao called “a massive miracle” unlikely to surface by manual search. He then fed that example into ChatGPT to check calculations and explore generalizations. Two days earlier, Dmitry Rybin—an accomplished competition mathematician turned startup founder—used ChatGPT alone to disprove another long-standing conjecture. His entire prompt chain was just four commands, yet it led ChatGPT to generate a complete counterexample with no expert steering beyond the initial instruction. These cases illustrate a new mode of problem-solving: automated hypothesis generation, rapid testing, and iterative refinement—in effect, an industrialized “brute intelligence.” Rather than random brute force, this approach resembles running millions of specialized labs in parallel. If a concise counterexample lies hidden behind tedious searches, an AI agent can unearth it in hours. Andrew Blumberg, a professor cited in the piece, says this outcome matches expectations: when a problem admits a short, verifiable counterexample, AI will find it if it can check its work. The article argues that we should reshape more fields into “math-like” puzzles so they fall into this AI-friendly loop. Code, drug discovery, finance models—any domain where solutions can be formalized, tested, and fed back—may soon rely on AI agents running continuous search cycles. Instead of handcrafting problems around human workflows, we’ll reframe them as closed-box experiments a machine can optimize. The key shift isn’t smarter AI but speed and scale: fast iteration over well-structured tasks. What remains uniquely human is work rooted in subjectivity, open-ended creativity, or hands-on experimentation. But every challenge reducible to a sequence of calculations or logical checks faces the risk of being ceded to “brute intelligence,” quietly churning away inside someone’s computer.