3 ms·
Can we bootstrap AI Safety despite being unable to even define it?
- cryptohell 11mo agoGiven several models, assuming only that some unknown subset is "safe", can we construct a single model as safe as that subset? This reduces obtaining a trustworthy model to a plausibly easier task.
- conartist6 11mo agoAI output is modeled on human behavior. Are humans safe?