“We’re at a critical moment. Language has long been seen as a defining line between humans and other animals, yet even in humans, verbal self-report is not always a reliable window into true understanding. That problem is even greater with LLMs. Just because a model gives a human-like answer does not mean it truly understands. To know what is happening under the hood, we need more careful, systematic ways of testing LLMs, drawing not only from human psychology, but also from animal cognition and comparative behavior research.” Prof Erica Cartmill on The Mindscape podcast when asked about understanding how LLMs communicate.