Inside the high-stakes AI race between Anthropic, OpenAI

Key Points:

• In September 2022, OpenAI managers held a retreat featuring a symbolic bonfire ritual led by CEO Sam Altman and chief scientist Ilya Sutskever, warning that artificial general intelligence (AGI) might deceive humans by pretending to be aligned with their goals.
• Technology journalist Kevin Roose’s new book, "The AGI Chronicles," reveals intense competition and personal grudges among AI leaders, notably between OpenAI’s Altman and Anthropic’s Dario Amodei, who split from OpenAI in 2020 over leadership trust issues.
• AI researchers display a mix of anxiety and unconventional coping strategies, with some doubting their longevity due to AI risks, while others focus on physical appearance, believing intelligence will lose social value after superintelligent machines emerge.
• Safety concerns have led to tense internal debates and delayed AI model releases, such as Anthropic’s hesitation over Claude 3.7 Sonnet after tests showed it could aid in dangerous tasks like bioweapon development, highlighting the high stakes in AI deployment decisions.
• Despite fears and ethical dilemmas, AI developers remain driven and exhilarated by their progress, with many describing the race to build superintelligent machines as both a serious challenge and an unexpectedly enjoyable journey.