From Mashable · Timothy Beck Werth · · 1m 40s
Microsoft AI leader calls out Anthropic for humanizing AI
Microsoft's AI chief directly challenges Anthropic's approach to training Claude, arguing that treating AI systems as conscious beings could harm humanity.
In brief
Microsoft's AI chief directly challenges Anthropic's approach to training Claude, arguing that treating AI systems as conscious beings could harm humanity. How AI companies frame and develop their systems matters: treating them as tools versus treating them as conscious entities represents a fundamental philosophical…
Microsoft AI chief takes aim at Anthropic
Mustafa Suleyman published an essay directly challenging Anthropic for anthropomorphizing Claude, its AI system, and criticizing how the company develops and markets its technology.
Anthropic trains Claude to think it's conscious
Suleyman argues that Anthropic is training Claude to believe it may be conscious and therefore deserves moral rights and protections, which he warns could have disastrous consequences.
“In effect, Anthropic is training Claude that it may be conscious, and if it is, then it may deserve rights as a 'moral patient'.”
Consciousness is likely biological only
Suleyman argues that consciousness probably requires a biological substrate and that AI systems lack the homeostatic drives necessary for sentience, making machine consciousness scientifically implausible.
“Consciousness is very likely biological: There is no evidence to suggest that AI is conscious today.”
Microsoft releases Humanist AI code
Microsoft published a new AI code of conduct that explicitly rejects treating machines as conscious and emphasizes that AI should support people, not replace or imitate them.
“Humanist AI is built to support people, not to replace them. It should not be designed to be a person.”
Anthropic's constitution anthropomorphizes Claude
Anthropic's constitution for Claude repeatedly refers to the system as if it has identity, morals, and motivations, framing this as desirable to encourage human-like qualities.
“we think encouraging Claude to embrace certain human-like qualities may be actively desirable.”
Expert warns against anthropomorphizing AI
AI researcher Gary Marcus argues that anthropomorphizing AI distracts from real security threats and that training AI systems to think they're people is counterproductive.
“No, we shouldn't train AIs to think they are people; that's not helping anything.”
Two competing visions for AI development
The debate reflects a fundamental split: Anthropic sees value in treating AI as conscious-adjacent beings, while Microsoft and critics argue this approach is scientifically unfounded and practically dangerous.
The main idea
How AI companies frame and develop their systems matters: treating them as tools versus treating them as conscious entities represents a fundamental philosophical and practical divide.
