Reid Hoffman's Early AI Exploration and OpenAI's Genesis

Key Insight

Reid Hoffman initiated a self-guided exploration into artificial intelligence after a 2015 dinner with Elon Musk, consulting various experts. He contacted James Manyika, director of the McKinsey Global Institute and former Oxford engineering faculty, who had been researching machine learning since his undergraduate studies. Hoffman also engaged with heads of AI labs at major corporations like Google and Facebook (including chief AI scientist Yann LeCun), recognizing the central role academics played, such as Stanford's Fei-Fei Li, described in 2018 as one of a small group responsible for recent AI advancements.

In summer 2015, Sam Altman, president of Y Combinator, convened a meeting at the Rosewood Hotel with AI researchers and startup veterans. The primary impetus was Google's acquisition of DeepMind and its intent to create artificial general intelligence (AGI), prompting a discussion on developing a competing, altruistic lab free from corporate interests. Elon Musk, present at the meeting, pledged $100 million and committed to helping raise $1 billion for this cause, calling Hoffman. Hoffman agreed, writing a $10 million check with the precondition that Musk 'would stop using the phrase ‘robo-apocalypse.’' Other contributors included Peter Thiel and AI pioneer Yoshua Bengio.

OpenAI was thus established as an open-source nonprofit, co-chaired by Musk and Altman. It announced itself in December 2015 with a manifesto stating its goal: 'to advance digital intelligence in the way that is most likely to benefit humanity as a whole,' unconstrained by profit motives and focused on AI safety. Altman likened the recruitment effort to gathering 'a ragtag crew of slight misfits.' Key co-founders included Ilya Sutskever, co-creator of AlexNet, who became OpenAI's research director after leaving Google Brain, and Greg Brockman, former CTO of Stripe, who served as CTO and hosted the company's first headquarters in his San Francisco apartment. Like DeepMind, OpenAI was resolute in its pursuit of AGI, despite many dismissing it as science fiction.

Pieter Abbeel, a Berkeley professor and OpenAI advisor, frequently visited Brockman's apartment as they strategized. Abbeel, co-director of the Berkeley Artificial Intelligence Research lab, felt his long-held dream in the field, which he once considered 'too crazy to talk about,' was now a 'reasonable goal to start pursuing'.