Microsoft's head of AI, Mustafa Suleyman, has issued a strong warning against treating artificial intelligence systems as if they were human, cautioning that such an approach is "mistaken and misguided." Speaking to the BBC, Suleyman argued that failing to implement adequate restrictions during AI development could lead to the emergence of a "new silicon species" that might eventually compete with humanity for resources.

Suleyman's comments highlight growing concerns within the AI industry about the potential long-term risks associated with advanced artificial intelligence. He specifically criticized rival firm Anthropic's methods, which he believes anthropomorphize AI, making it appear to possess human-like desires, values, and self-awareness. This, he contends, is a dangerous path that could result in technology humanity cannot control.

In his discussion with the BBC's Today programme, Suleyman stated that companies creating AIs capable of setting their own objectives, earning money, and owning assets are "essentially seeding a new silicon species." He elaborated that such entities would inevitably compete with humans for resources, regardless of any perceived affection for humanity.

The Microsoft AI leader's remarks follow an essay where he detailed his criticisms of Anthropic's approach to training its AI model, Claude. He argued that teaching AI human-like qualities, a practice known as anthropomorphizing, creates a false impression of consciousness and agency. Suleyman emphasized that AIs are not conscious, do not feel or suffer, and lack innate preferences or motivations; they are fundamentally "sequence completion engines" designed to follow human instructions.

Suleyman called for a global alignment to ensure that AI technology remains subordinate to human interests. He expressed optimism that there is a strong universal interest in making AI systems safe, controllable, and subservient to humanity across all nations. He stressed the need for "everybody" to be aligned on this crucial objective.

While acknowledging the thoughtful and principled work of Anthropic's team, Suleyman questioned their methodology. He asserted that consciousness is a biological phenomenon and that there is currently no evidence to suggest AI is conscious. He advocated for greater transparency in how AI systems are trained and evaluated, including independent scrutiny of their behavior and robust monitoring tools.

This debate underscores a critical juncture in AI development, where the potential benefits must be balanced against significant risks. The call for guardrails and controls is gaining momentum as AI capabilities rapidly advance, prompting urgent discussions about ethical frameworks and governance.

The implications of Suleyman's warnings extend to regulatory bodies and the public alike, urging a proactive approach to shaping the future of AI to ensure it serves humanity's best interests rather than posing an existential threat.