A new arXiv preprint takes issue with a common assumption in AI: that progress scales with effort. The abstract describes the widespread view that more researchers, papers, and investment should accelerate discovery. It then points to theories of technological growth that locate the difficulty of discovery elsewhere—in the process itself, rather than in the amount of effort applied.
The paper's title adds a sharper claim: stronger AI breakthroughs hasten their own exhaustion. That suggests the field's biggest successes may also bring the next round of progress closer to a limit, even if the abstract does not spell out the mechanism in the available text.
The announcement is short, so the full argument is not yet available from the abstract alone. Still, the framing is a useful counterpoint to scaling optimism in AI research.