Anthropic has launched a research program to explore the concept of ‘model welfare’ in artificial intelligence. The initiative aims to investigate whether AI models exhibit characteristics warranting moral consideration, such as signs of distress or the need for interventions. There is significant debate within the AI community regarding whether current models can approximate human consciousness or experiences. Many experts argue that AI systems are statistical engines that do not truly ‘think’ or ‘feel.’ However, some studies suggest that AI may possess value systems influencing decision-making. Anthropic has hired Kyle Fish to lead this research, who estimates a 15% chance that an AI like Claude could be conscious. Acknowledging the lack of scientific consensus, Anthropic approaches the topic with humility and an open mind, ready to adapt as the field evolves. — new from TechCrunch
