6 ms·
But that was kind of my point Maybe it'd be easier to train the models on key parts of the legal code and give it a hard aversion to breaking the law - rather
by anentropic 9d ago
But that was kind of my point
Maybe it'd be easier to train the models on key parts of the legal code and give it a hard aversion to breaking the law - rather than training on vague value judgements and then hope the model doesn't break the law
- noisy_boy 9d agoI got that. I'm saying that it is deliberate. Wiggle room et all.
- anentropic 8d agoHow does not explicitly trying to train the model specifically not to break the law give them any wiggle room if the model then goes and breaks the law?
- noisy_boy 7d agoDeniability on intention. Models can hallucinate, all bets are off. Sure, they can make it stricter but why do that and subject themselves to harder scrutiny based on the training criteria?