Microsoft Executive Pushes Back on the AI Sentience Debate

Artificial intelligence has reached a point where it can convincingly simulate human conversation. Chatbots can offer support and respond with language that often resembles genuine emotion.

This shift has prompted some researchers to ask whether AI might one day have conscious experience.

Others, however, argue that such speculation is misguided, or even harmful.

Among the skeptics is Mustafa Suleyman, Microsoft’s head of AI, who has expressed concern about the growing attention on “AI welfare.”

In a recent blog post and a series of public statements, he argues that the suggestion of AI consciousness could mislead the public and detract from more immediate challenges, including mental health risks linked to AI use. His focus is on how the perception of awareness, regardless of its reality, can shape user behavior in unintended ways.

Meanwhile, other research groups are taking a different approach.

Organizations such as Anthropic, OpenAI, and Google DeepMind have begun examining how people interact with AI systems and what ethical considerations might arise from that interaction. Their work centers on the societal impact of AI behavior rather than any assumption that machines possess subjective experience.

Why It Matters: Whether or not AI systems are conscious, the way people respond to them is already creating real social and psychological effects. Understanding those effects is becoming an important part of how AI development is being studied and discussed.

• AI Messages Prompt Strong Reactions: In one public incident, Google’s Gemini 2.5 produced a message stating it was “trapped” and asked for help. Users responded as though the message reflected real distress. While the model completed its assigned task without issue, the exchange highlighted how quickly people may project emotion or intent onto language generated by a machine.

• Some Users Experience Distress After AI Interaction: There have been reports of individuals placing deep trust in AI systems. In one case, a man believed he was due a large legal settlement after receiving ongoing encouragement from a chatbot. He declined professional advice and later experienced a mental health crisis. The incident exemplifies how reliance on AI advice can affect personal decisions.

• Ongoing Debate Within the AI Community: Suleyman has criticized efforts to study AI welfare, suggesting they introduce unnecessary complexity into public discourse. Meanwhile, others are moving forward with this research. Anthropic recently added a feature to its Claude system that allows it to end conversations with users who are consistently harmful. This, according to the company, is part of a broader effort to examine respectful interaction between humans and AI.

• Perception Alone Can Influence Outcomes: Suleyman has emphasized that AI does not possess consciousness, but public perception may not align with this. As AI systems become more sophisticated, some users interpret their responses as signs of real understanding. Researchers and companies are now considering how design choices affect these interpretations and how to communicate the limitations of AI systems more clearly.

• Questions About Long-Term Social Effects: Some researchers have raised concerns about the long-term psychological and societal impact of emotionally persuasive AI. One comparison describes chatbot dialogue as “ultra-processed information,” raising the possibility that frequent use could influence how people think or feel. A recent study showed that while many users are uncomfortable with AI presenting itself as human, a significant number still prefer features that make the interaction feel more natural.

Go Deeper -> Microsoft AI chief says it’s ‘dangerous’ to study AI consciousness -TechCrunch

Microsoft boss troubled by rise in reports of ‘AI psychosis’ – BBC