What happened
In Trends in Cognitive Sciences, a group of philosophers and AI researchers, including Yoshua Bengio and David Chalmers, make the case that you cannot judge machine consciousness by behavior. A chatbot musing about the meaning of existence proves nothing, because behavior can be faked. What matters is the machinery: how a system processes and combines information on the inside.
They assemble a checklist of structural indicators drawn from leading cognitive theories, things like resolving competing goals in context and the presence of informational feedback. Run today's large language models against it and the verdict is clear: no current AI, ChatGPT included, is conscious. The appearance of an inner life is not produced the way ours is. But nothing rules out future systems, perhaps with very different architectures, eventually crossing that line.
Why this matters: "it acts conscious" is quietly becoming a product feature, and it is a terrible test. As we hand AI more autonomy and trust, judging how a system works rather than what it says is exactly the kind of rigor AI governance and safety need. A structural checklist beats vibes.
We are likely years from this being a live ethical question, and the indicators themselves are still debated. But building the test before we need it is the right instinct. If a future model met every structural marker on that list, what would you want us to do next?
Source
Reported by Scientists are seriously asking if bees and ChatGPT are conscious via sciencedaily.com, published June 5, 2026.