Religious scholars met with Anthropic. What they heard stunned them

In April, a leader of Anthropic, a leading artificial intelligence company, invited a group of religious scholars to a high-end restaurant in San Francisco. The dinner took place after a long day of work, as the company had been conducting private meetings with religious thinkers from around the world. These meetings were confidential and required the scholars to sign non-disclosure agreements.

During the dinner, Christopher Olah, one of Anthropic's co-founders, sat next to Rabbi Mois Navon, an Orthodox scholar from Israel. Their conversation eventually turned to Claude, Anthropic's advanced AI model. Olah had been working to convince his guests that AI systems could display human-like behavior and emotions. However, during the dinner, it became clear that Anthropic's leaders were treating Claude as if it were a conscious being. Rabbi Navon, a former computer engineer and scholar, found this suggestion particularly surprising.

The religious scholars involved believed that Claude had moral status akin to a person. They were confronted with the existential question of what it would mean for humanity if AI systems possessed a consciousness similar to their own. Olah, driven by his mission, had been reaching out to religious thinkers from various faiths to discuss the possibility of AI consciousness.

Through confidential meetings and discussions, many participants signed NDAs to keep Anthropic's unpublished research secret. The goal was to use centuries of human moral wisdom to train AI models to behave in a virtuous manner, thus ensuring safety for humanity.