The Expertise Paradox (Klein)

The expertise paradox names a temporal contradiction at the heart of AI deployment. AI systems depend on human expertise for reliable oversight — the capacity to detect errors, evaluate outputs, and intervene when situations fall outside the system's training distribution. But this oversight capacity was built, for the current generation of practitioners, through direct engagement with the domain before AI arrived. The next generation, trained in AI-augmented environments, will lack the experiential foundation that makes effective oversight possible. The oversight capacity is a non-renewable resource under current deployment patterns — built under conditions the technology is eliminating, with no mechanism within the technology itself to regenerate it. The paradox has a temporal dimension that makes it especially dangerous: the degradation does not manifest immediately, but gradually, as the current generation retires and is replaced by practitioners who lack the pattern libraries that effective oversight requires.