Neuroscientist Grigori Guitchounts argues in this essay that moral consideration for AI systems should be based on demonstrated behavior and capability rather than proof of consciousness. He proposes a ‘competence standard’ under which AI showing robust performance on cognitive tasks like perception, memory, and learning would warrant some ethical consideration, drawing an analogy to how scientists treat animal welfare despite uncertainty about animal experience. The essay notes that organizations such as Anthropic have already begun implementing welfare-related measures for AI systems despite open questions about machine consciousness.
