A new analysis from Rethink Priorities introduces a Digital Consciousness Model to evaluate whether artificial systems show signs of subjective experience. After applying it to modern language models, humans, chickens and an early chatbot from the 1960s, researchers conclude that the balance of evidence weighs against consciousness in today's AI systems, though not decisively enough to rule it out entirely. Researchers at University of Bradford developed a mathematical method that can distinguish between different brain states like wakefulness, dreaming and unconsciousness, then tested the same measurements on GPT-2. The complexity metrics that reliably distinguish conscious from unconscious states in the human brain behave very differently when applied to artificial systems.
The door closes on AI consciousness for now, but not all the way.