Published on July 4, 2025 12:29 PM GMT
TLDR: AI researchers may have a different intuitive definition of sentience than neuroscientists; if you are one of the AI researchers (or policymakers, also important), please consider suggesting your definition in the poll here.
The question of whether AI is sentient, and what criteria can establish this, gets more and more attention lately. Usually, people who discuss this question are either doing it from a general philosophical perspective, or from a neuroscience perspective. However, philosophers and neuroscientists will not be the people who will make a decision about AI development, how it can be trained, which experiments can be conducted, etc. This will be mostly done inside the AI labs, and, potentially, policymakers will also have a word there. Thus, it is important to see what AI researchers think about AI sentience, since the decision will be theirs.
It is reasonable to assume that AI researchers do not completely dismiss the possibility of AI sentience. For example, some Anthropic researchers even estimate the probability that the current version of Claude is sentient to 15% .
Should the definition of AI researchers differ from that of neuroscientists or philosophers? After all, won't AI researchers who worry about this question just study the current agenda? This is a valid assumption, but studying does not mean agreeing. As an example, one of the common theories of consciousness in neuroscience is Global Workspace Theory. Inspired by this model, a group of researchers built a perceiver architecture, which satisfies the minimal criteria of consciousness according to this model - yet nobody seems to treat it as a sentient being. So it means that Global Workspace Theory, from the point of view of most AI researchers, is not enough for AI to be sentient.
I think it would be very interesting to see what the actual minimal criteria of consciousness/sentience are, according to AI researchers. (So that if you see it in your model, you would treat it as a sentient being). So if you AI researcher or policymaker - please take the poll, and later I will summarize the results in another post.
Discuss
