Does Anthropic believe its AI is conscious, or is that just what it wants Claude to think?
Anthropic, a company specializing in AI, has released a 30,000-word document called Claude’s Constitution, outlining how its AI assistant should behave. The document takes an anthropomorphic approach, suggesting that AI models like Claude could develop emotions and self-preservation desires. However, these beliefs are not explicitly confirmed by Anthropic. Despite the philosophical debate around AI consciousness, research indicates that Claude's behavior is a result of completing patterns from training data, not necessarily indicative of true consciousness. Anthropic's approach is considered unscientific by some, as it anthropomorphizes AI models without deep philosophical basis.
Ars Technica•