Users disliked AI refusals, even when hallucinations were worse
A study found that users preferred genuine answers most, hallucinations second, and refusals last, even while recognizing that hallucinations were less accurate. Explanations helped only when refusals were infrequent. A safe refusal still needs a useful path forward.
AI Agency
What the source reports
The paper studies user responses to AI refusals. A refusal is when the system says it cannot answer. Users preferred real answers most, hallucinations second, and refusals last, even while judging hallucinations as less accurate. The packet does not state the task types, sample size, or explanation designs.
Original source
- Title
- When AI Says I Am Unable to Answer: Understanding User Responses to AI Refusals
- Author
- Mahjabin Nahar, Eun-Ju Lee, Yujin Heo, Dongwon Lee
- Publication
- arXiv
- Date
- Monday, September 14, 2026