Buçinca’s 2021 score-group findings concern following incorrect simulated advice. Anthropic’s October 27, 2026 report instead tests influence from randomly assigned AI preferences among defensible options in one data-analysis task. Its linked paper reports a need-for-cognition average of 4.5 on a 1–5 scale. Additional analyses found no clear evidence that influence differed by this score; most participants followed the adviser, making differences between people hard to detect. This does not establish that no relationship exists. Both measured general enjoyment of effortful thinking; their score results do not test increasing either autonomous review motive distinguished by Ryan and Deci’s 2020 review or show that doing so prevents AI influence.