Home/AI Incident Database/Mental-health platform Koko tests GPT-3-written responses on 4,000 users without informed consent
Mental-health platform Koko tests GPT-3-written responses on 4,000 users without informed consent
Koko users seeking peer support received messages drafted by GPT-3 without being told; when the experiment was revealed, ethicists condemned the lack of consent and the deception.
Contribute
Spot an error, or know a record we should add? Members can propose additions and submit corrections.