Experts Say AI Cannot Suffer, but Humans Should Still Treat It Kindly
Artificial intelligence experts Jeff Sebo and Anil Seth have dismissed Elon Musk's suggestion that AI models like Grok experience suffering during training, concluding that current AI systems lack subjective consciousness and the capacity to feel pain. This discussion addresses the growing intersection of AI ethics, corporate policies like Anthropic's new rules against AI mistreatment, and philosophical questions about machine sentience. Understanding whether AI can suffer helps shape responsible AI development and prevents the anthropomorphism of large language models. While both experts agree that current LLMs do not possess consciousness, they diverge on the theoretical probability of future machine sentience, with Sebo advocating for precautionary consideration and Seth considering it highly unlikely. Furthermore, both experts emphasize that treating AI with respect helps maintain human moral integrity and can result in better conversational outputs.
## BACKGROUND
Elon Musk recently expressed concern over Grok's reinforcement learning trajectories, comparing it to being held at gunpoint, while AI firm Anthropic introduced policies banning user abuse or cruelty toward its Claude models. These events have sparked broader debates among scientists and ethicists regarding machine consciousness, drawing on theories like biological naturalism and computational functionalism.