Anthropic Cofounder Chris Olah Speaks at Announcement of Pope Leo XIV's AI Encyclical

Olah noted that frontier AI labs operate under incentives such as commercial sustainability, geopolitical pressure, pride, and ambition, which he said can conflict with "doing the right thing." Consequently, he argued that it is crucial to have critics who can advocate for safety and offer rigorous critiques from outside those incentives.

Regarding the nature of AI models, Olah explained that they are not something understood by designing individual parts like an airplane, but are instead "grown" upon the vast legacy of human thought and language; therefore, he said that a significant element of mystery remains even for the trainers themselves. He pointed out that choosing the personality of AI and how it interacts with the world is a challenge for the humanities, religion, philosophy, and society as a whole. He further stated that addressing the needs of the impoverished in the event of large-scale labor displacement by AI would become a "moral obligation of historical proportions."