A new research paper from Anthropic explores the concept of ‘functional emotions’ within its AI assistant, Claude. The researchers argue that while Claude does not experience subjective feelings like humans, it exhibits internal states that serve similar functional roles. These states, such as simulated ‘curiosity’ or ‘frustration,’ are engineered to improve the AI’s performance on complex tasks by guiding its attention and resource allocation. The paper clarifies this is a technical framework for making AI systems more effective, not a claim of sentience. The findings contribute to the ongoing scientific discussion about the nature of intelligence and the internal workings of large language models. Read the full article at: https://www.wired.com/story/anthropic-claude-research-functional-emotions/