Researchers from the University of Bradford and Rochester Institute of Technology applied scientific methods used to assess consciousness in humans to artificial intelligence systems, including large language models similar to ChatGPT, in new studies published in February 2026. The tests revealed that AI can produce conscious-like signals even when degraded, highlighting why complexity doesn't equal awareness. Structurally impaired GPT-2 variants scored higher on some consciousness-style metrics than intact models, undermining the idea that passing consciousness tests proves anything. The findings could prove valuable for AI safety, reliability and future regulation by helping engineers understand when an AI system is functioning well or beginning to break down.
Congratulations: you've found a way to flunk the consciousness test by being too good at faking it.