Many nonhuman animals also show long-term abnormal behaviours, and will try to access analgesia (even paying a cost to do so), if they are in pain. I don’t think we have evidence that’s quite analogous to that with large language models, and if we did, it would cause me to update in favour of current models having sentience. It’s also worth noting that the same lines of evidence that cause me to believe nonhuman animals are sentient also lead me to believe that humans are sentient, even if some of the evidence (like physiological and neuro-anatomical similarities, and evolutionary distance) may be somewhat stronger in humans.
Many nonhuman animals also show long-term abnormal behaviours, and will try to access analgesia (even paying a cost to do so), if they are in pain. I don’t think we have evidence that’s quite analogous to that with large language models, and if we did, it would cause me to update in favour of current models having sentience. It’s also worth noting that the same lines of evidence that cause me to believe nonhuman animals are sentient also lead me to believe that humans are sentient, even if some of the evidence (like physiological and neuro-anatomical similarities, and evolutionary distance) may be somewhat stronger in humans.