Let the Abyss Stare Back Adaptive Falsification for Autonomous Scientific Discovery

ArXi:2603.29045v1 Announce Type: new Autonomous scientific discovery is entering a dangerous regime: once the evaluator is frozen, a sufficiently strong search process can learn to win the exam without learning the mechanism the task was meant to reveal. This is the idea behind our title. To let the abyss stare back is to make evaluation actively push against the candidate through adaptive falsification, rather than passively certify it through static validation. We