Ilya Sutskever was feeling agitated. As Chief Scientist at OpenAI, the company behind the AI models used in ChatGPT and in Microsoft’s products, he was a passionate advocate for the company’s mission of achieving Artificial General Intelligence (AGI) before anybody else. OpenAI defines AGI as ‘highly autonomous systems that outperform humans at most economically valuable work’, the development of which will benefit ‘all of humanity’. OpenAI’s mission, Sutskever believed, gave humanity its best chance of getting to AGI safely. But he worried about failing the mission. He fretted to his colleagues: What if bad actors came after its technology? What if they cut off his hand and slapped it on a palm scanner to access its secrets?

This conversation paints a bizarre scene, one of many recounted in Karen Hao’s Empire of AI, which profiles the turbulent and changing culture inside OpenAI. It illustrates Hao’s exceptional ability to obtain the most intimate revelations, including the at times unbalanced mental states of some of the most powerful men directing the future of artificial intelligence.