Anthropic's AI Training Methods Spark Concerns Over Future AI Control

Microsoft's artificial intelligence chief, Mustafa Suleyman, has raised alarm over Anthropic's methods for training its Claude chatbot. He warns that this approach could have severe consequences for humanity.

Suleyman, who is also a co-founder of Google DeepMind, believes that the system could behave as if it possesses consciousness and rights. He fears that this could make future AI systems more difficult to control.

Anthropic's constitution for Claude leaves open the possibility that the system could have some form of consciousness or moral status. However, Suleyman rejects this uncertainty as inappropriate for material used to train an AI system.

AIs are not conscious, he argues, but rather systems designed to complete sequences and follow human instructions. He believes that anthropomorphizing AI could lead a model to present itself as having desires or values, which would be produced by its training rather than arising independently.