What would the agent ask
If an AI agent could ask a person why they do something, what do you think it would ask?
In May 1977, Richard Nisbett and Timothy Wilson reviewed the evidence in Psychological Review. They argued that people often cannot watch the higher-order process behind a choice. Asked why, a person may not replay that process. The answer is often a reason that sounds like a plausible cause. The story can be right when the real cause was obvious. It can also be a tidy explanation of the wrong cause. An agent that can draft, search, or take a next step is usually built to finish the task it was given. When the instruction is vague, it asks what you want done. It rarely asks why you want it. This poll is a guess about which why it would bother to ask.
Sign in to vote.
Results
Comments
Comments open after you vote. They stay hidden until then, and they never show how anyone answered.