What should go in a model spec? (opens in new tab)

Suppose an AI company is considering whether to include some particular quality X – a rule, virtue, heuristic, default, attitude, goal, or style – in a model spec.Perhaps they are considering whether their LLM should have . Perhaps they’re wondering if the LLM should whistleblow to help prevent . Or perhaps they’re uneasy about whether the LLM should be so exactingly honest that it always tells the truth to children . And so on.What kind of reasons might be invoked over the course of such con...