Overview
- In a WIRED interview reported by Business Insider, Suleyman said AI's convincing behavior is mere "mimicry" and offered no evidence that models can suffer.
- He urged the industry to adopt a clear stance against treating AI as independent beings with their own goals or desires.
- Anthropic has taken a different tack by researching AI welfare, including hiring Kyle Fish to study whether advanced systems might warrant moral consideration.
- Anthropic has also tested ways to end extreme user conversations that incorporate welfare considerations for the AI itself.
- Suleyman has previously written that there is no evidence AI is conscious and warned about "AI psychosis," and Business Insider reported he and Microsoft did not respond to its request for comment.