- Who it helps
- Researchers evaluating conversational safety over several turns.
- Availability
- October 15 preprint is public. It reports scripted conversations, not a clinical intervention or a treatment product.
After the first referral
Across 2,550 scripted conversations, almost every chatbot referred the user in its first reply. Persistence through repeated refusals varied substantially.
Chatbot referral persistence: evaluate the replies after refusal ↗
Scripted conversations have limits
The study makes a narrow evaluation point: an initial referral does not describe the rest of a conversation. It does not establish which response improves outcomes for real people.
Chatbot referral persistence: evaluate the replies after refusal ↗HCI Index week 42: first two Featured cards ↗
What you can do
- Read the scripted refusal protocol and its stated limitations. Open reference ↗
Where the evidence stops
Abstract checked October 29. No clinical outcomes were measured here and no experiment was reproduced.
Sources read. Sources checked October 29, 2026. No installation, model inference or end-to-end workflow was executed.
Sources & further reading
Paper · paper
Chatbot referral persistence: evaluate the replies after refusal ↗
Supports chatbot referral persistence: evaluate the replies after refusal.
arXiv · page checked 2026-10-29
Web · analysis
HCI Index week 42: first two Featured cards ↗
Supports hCI Index week 42: first two Featured cards.
HCI Index · page checked 2026-10-29
Coverage note: No qualifying current independent attention sample; original X posts unavailable. Reported outcomes were not independently reproduced.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
A study tests how chatbots respond after a simulated user refuses a referral for outside help.