“Claude, what are you thinking about?” What sounds like the beginning of a conversation between a couple who have little left to say to each other is actually what is particularly driving AI developer Anthropic these days. Specifically, the AI makers at the OpenAI competitor wanted to find out whether the in-house AI chatbot Claude even has something like self-reflection, i.e. can understand what is actually going on in its “brain”. Anthropic’s answer: “Yes, he can”. Postscript: “At least sometimes.”
