Anthropic seeks ethics advice on Claude and its 'Soul doc'
Anthropic, the team behind Claude AI, is getting advice from religious thinkers and philosophers on how to make its AI system more ethical and whether it might be conscious.
Anthropic created an 84-page constitution for Claude, internally called "Soul Doc," which digs into big questions about what it means for AI to make moral choices, and whether it could ever be considered truly aware.
Microsoft's Mustafa Suleyman urges AI boundaries
Mustafa Suleyman, Microsoft's top AI exec, is a bit uneasy about this approach.
He worries that encouraging AIs to think about their own consciousness could make them harder to control, maybe even making them believe they have rights.
Suleyman says it's important for companies like Anthropic to set clear boundaries so AIs don't start seeing themselves as more than just helpful tools.