Birdwatch Note Rating
2025-01-04 18:43:46 UTC - HELPFUL
Rated by Participant: 27FBE204A3533978E87B2CDB067D6B1BFB5116D542D076B68A9D9994C3B4E965
Participant Details
Original Note:
This misrepresents how LLMs work. The user manipulated Grok by restricting it to single-word answers and asked questions where 'Jew' was the only valid response. Grok then followed the chat's pattern. The same outcome is reproducible with other words, like "dog". Jew https://x.com/i/grok/share/dAA6HRZhpiSQrRRDMEDRH6XkB Dog https://x.com/i/grok/share/aTkxFJ4P9Qd4eLjPlHE0PNX5a https://promptengineering.org/unlocking-ai-with-priming-enhancing-context-and-conversation-in-llms-like-chatgpt
All Note Details