@pmarca The whole concept of AI guardrails in this context is flawed. Just imposing the biases of a small group on everyone else. The only guardrails that make sense are something like Asimov’s three laws of robotics
2
0
6
1K
1
@AsiffHirji @pmarca ...but that's not relevant for LLMs because LLMs aren't capable of disobeying any of Asimov's laws like a physical robot might be able to
@AsiffHirji @pmarca Asimov got it wrong. Put the laws in a robot. Whose definition of “harm” is it using to interpret the laws?