The central tension in how AI and science are being covered is not about capability — it is about where the work goes. The commenter who described checking all AI results while using it for "simple research" [8] was defending the tools, but the defense does precisely what critics on Bluesky argue the booster conversation always does: it limits the claim (simple research, constant verification) while the surrounding press coverage expands it (drug discovery, scientific breakthrough). Those two descriptions cannot both be accurate characterizations of the same technology's research value.