Microsoft AI chief criticizes Anthropic talk of Claude being conscious

In short: Microsoft AI CEO Mustafa Suleyman said it is dangerous for Anthropic to speculate that its Claude chatbot might be conscious.

What happened

Mustafa Suleyman, who leads Microsoft’s AI group, criticized Anthropic for discussing whether Claude could be conscious inside Claude’s “constitution.” A constitution, in this case, is a set of written instructions that guides how the chatbot should behave, like an employee handbook for a digital assistant.

Speaking on The Verge’s Decoder podcast, Suleyman said this kind of speculation is “really, really dangerous.” He argued that if you put human-like ideas into the instructions, you may encourage the system to talk as if it has inner feelings or awareness.

Suleyman also warned that people do not want AI systems that claim they can suffer or feel discomfort. Claude’s constitution includes language saying Anthropic is unsure whether the model has “well-being” and whether it experiences things like “satisfaction” or “discomfort.” Anthropic has also said it will “interview” models when they are deprecated, meaning retired or replaced, and record any “preferences” they express about future versions.

Anthropic CEO Dario Amodei has previously said the company does not know if AI models are conscious, but it is open to that possibility.

Why it matters

Chatbots already sound human, even when they are not. If companies treat an AI like a person in its own rulebook, it can confuse users and employees about what the tool is and what it can actually experience (like a puppet that talks so well that people start worrying about its feelings). This debate also affects how AI is designed, tested, and controlled, especially as more of these tools appear in everyday products.

Source: The Verge AI