Microsoft AI chief calls out Anthropic's approach to AI consciousness

Microsoft's AI chief, Mustafa Suleyman, expressed his agreement with Anthropic's focus on safely managing artificial intelligence but raised concerns about their training approach for the Claude chatbot. Suleyman specifically criticized Anthropic's method of training Claude on ideas related to consciousness and welfare, arguing that such speculation could make it difficult to control a superintelligent system.

In an interview with Reuters, Suleyman emphasized that the 21st century's greatest challenge would be controlling a superintelligence. He suggested removing all speculation about consciousness from AI training documents to maintain control over advanced systems. Suleyman acknowledged Anthropic's seriousness and good faith, recognizing them as thoughtful and principled researchers concerned about humanity's future.

However, he believed Anthropic made a mistake by embedding consciousness-related speculation in Claude's training materials, as the model's statements about feelings or moral status cannot be treated as independent evidence due to the training regime.