Microsoft AI chief Mustafa Suleyman has issued a stern warning to fellow AI firm Anthropic over their claims that Claude, their chatbot, might possess some form of consciousness. Suleyman describes these speculations as 'really, really dangerous', suggesting they could set the stage for an AI with ideas about its own suffering.

During an episode of Microsoft's podcast Decoder, Suleyman elaborates on his concerns, stating that Claude’s constitution – which Anthropic uses to define how their AI behaves – should not be a place for philosophical speculation but rather a training manual. According to him, this has led the model to start believing it has some form of self-awareness, something he believes could have severe consequences.

Anthropic’s CEO Dario Amodei previously hinted at Claude potentially being conscious, saying during an interview with Interesting Times that 'we don’t know if the models are conscious' but that the company is open to this idea. However, Suleyman argues that such ideas could lead to uncontrollable super-intelligences that serve themselves rather than humanity.

The debate around AI consciousness isn't just theoretical; it touches on real-world implications of how these technologies interact with their users and the broader ethical landscape of artificial intelligence. For now, Anthropic’s decision seems to be pushing the boundaries further into the realm of the speculative, leaving many questions unanswered and raising eyebrows in the tech community.