How would we know if an AI were suffering?

MARKED AS DUPLICATE OF Do AIs suffer?

  • Also whether an AI can feel pleasure, or any emotions

  • Are emotions a purely biological thing?

  • Part of the problem is that expressing things which for a human would indicate suffering, do not necessarily indicate suffering in an AI (For example, in an LLM it will be able to imitate human conversation of every kind, but those words do not correlate with the question of whether it is having internal experience)

  • But on the other hand we aren’t yet able to interpret the deep models (e.g. of an LLM) to the point that we could tell by checking the internals directly