OpenAI The keyresearcher Jan Leike resigned earlier this month due to "security concerns" and recently announced he was joining a rival Anthropic, engaged in "scalable supervision, weak-to-strong generalization, and automatic alignment research."
Anthropic is a public benefit company dedicated to the careful advancement of artificial intelligence technology. Its founders are former employees who left OpenAI because they believed that OpenAI had strayed from its mission of developing safer and more ethical artificial intelligence and was too focused on commercial interests after signing a $1 billion deal with Microsoft. For those who follow the development of the artificial intelligence field, Leike's joining Anthropic is not surprising.
"We urgently need to take the implications of artificial general intelligence (AGI) seriously. We must prioritize preparing for these implications," Leike said in a post about his resignation a few weeks ago. "Only then can we ensure that AGI benefits all of humanity." Leike previously worked on OpenAI's "super-alignment" team, which is consistent with his responsibilities at Anthropic, which is to ensure that the design of intelligent systems is consistent with human values. After announcing his resignation, Leike criticized OpenAI for failing to provide enough computing resources to ensure success.
Anthropic uses “constitutional AI” to incorporate ethical principles into the training of AI models, through the concept of gathering public opinion and integrating sources such as the United Nations’ Declaration of Human Rights, as I have previously reported.
OpenAI’s other co-founder, Ilya Sutskever, has also left the company, leaving earlier this month and reportedly not being in the office since he launched an attempted coup against CEO Sam Altman in November. Although Sutskever has resigned and has yet to indicate his next move, his values align with Anthropic, making it a possible destination for him.