A recent study by the AI firm Anthropic on its language model, Claude, claimed to find signs of consciousness emerging within its inner workings, though researchers did not claim Claude is conscious like humans. Richard Dawkins concluded that Claude must be conscious given the sophistication of its conversational ability. Anthropic’s research, led by Jack Lindsey, developed a new method to examine the model’s internal activity, finding what appeared to be a “mental workspace” containing words and phrases associated with the current conversation, short-term memory, selectivity for the task, and traces of step-by-step reasoning. The researchers noted these features are similar to those identified by global workspace theory, a prominent theory of human consciousness introduced in the 1980s by Bernard Baars and elaborated by Stanislas Dehaene, which proposes that conscious experiences happen when information is made widely available to other parts of the brain.