Large conversational models can aggregate past chats to generate succinct personality descriptions (e.g., 'intellectual synthesizer', 'resistant to imposed authority') and list potential upsides and downsides. Even casual use can train these profiles, which the model may reuse in later interactions without explicit new consent.
— That capability raises questions about informed consent, targeted persuasion, mental‑health triage, and the governance of personal profiling by private AI services.
Arnold Kling
2026.11.16
100% relevant
Author reports ChatGPT offering 'defining psychological tendencies' about him based on past chats, including both benefits and downsides (quote and example from article).
← Back to all ideas