3 ms·
Architecture wise the "admit" part is impossible.
by BoredPositron 11mo ago
Architecture wise the "admit" part is impossible.
- rbranson 11mo agoBricken isn’t just making this up. He’s one of the leading researchers in model interpretability. See: https://arxiv.org/abs/2411.14257 https://arxiv.org/abs/2411.14257
- Tepix 11mo agoWhy do you think it's impossible? I just quoted him saying 'by default, it will actually say, "I don't know the answer to this question"' We already see that - given the right prompting - we can get LLMs to say more often that they don't know things.