Researchers analyze 19,930 young adults’ distress conversations with ChatGPT
A team of researchers studied 19,930 conversations between young adults aged 18–25 and ChatGPT to understand how the AI handles emotional distress. They collected data from 158 participants and selected five example chats where users expressed acute distress. Ten clinicians reviewed these conversations and found that distressed users reported stronger emotional engagement with ChatGPT and more…
Key points
- Researchers analyzed 19,930 ChatGPT conversations from 158 young adults aged 18–25
- Clinicians flagged seven process failures in distressed user interactions, like jumping to solutions
- Study proposes three-stage AI design guidelines for handling emotional distress
The clinicians noted that ChatGPT often provided overly dramatic or action-oriented advice in moments of distress. While they praised the AI’s availability and some of its wording, they identified seven key process failures, such as prematurely offering solutions. Based on their feedback, the researchers proposed design guidelines for future AI interactions: first assessing safety, then de-escalating emotional intensity, and finally exploring concerns without immediate agreement.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Society & Work
All →- Pope Leo XIV dismisses AI safety warnings as fake news · 5 src
- Cramer warns AI stocks may suffer from shifting public perception · 2 src
- Bay Area artists unveil What Iceberg? protest art at OpenAI event · 1 src
- Google AI Overview misleads Alaska hunter with wrong hunting date · 1 src
- Anthropic co-founder admits uncertainty about Claude’s consciousness in new interviews · 1 src
Comments
via GitHub Discussions