Although much behind Friday’s unexpected announcement remains unknown, some things are certain: Altman was the critical figure in generative AI’s cultural breakthrough. As the CEO of the start-up incubator Y Combinator, he could have slid into the corner office of nearly any tech start-up. He chose OpenAI. He was unusually candid about his strengths (management, mission-focus) and weaknesses (better at tech than finished products, and not exactly in touch with the concerns of the common man). This was both a charming and shrewd way to do a high-profile job.
The most visible thing Altman did as CEO was … be visible. For most of the past year he performed a one-man traveling show around the world, but D.C. was his most important venue. Technology can be hard to understand, especially for members of Congress; Altman became the face of AI in much the same way Steve Jobs was once the face of personal computing. Altman was unfailingly patient and polite in sworn testimony before lawmakers, and behind the scenes he conducted demos for individuals and groups that reached more than 100 members of Congress. If the White House wanted him for a ceremonial occasion, he showed up. If Sen. Charles E. Schumer (D-N.Y.) needed Altman to bolster the credibility of his AI working groups, he didn’t need to ask twice. It’s worth stating how highly unusual this level of collaboration between tech and government had become in the previous decade.
But Altman was not merely AI’s hood ornament. He made two critical product decisions at OpenAI. The first was driving the design and user interface of ChatGPT, with that elegant little pause that makes it seem like magic is happening in response to your prompt — not just scale computing. As Altman told me over the summer, “I’m not a product guy. I’m like a tech guy. But in this particular case, I did have some strong opinions about what it was going to look like and how it ought to work for a user. It’s sometimes nice when one person’s got a clear vision. Doesn’t happen often, but in this case …”
He also decided to bring ChatGPT-3 to the general public in the fall of 2022, before much of its capabilities were known. It was risky. GPT-3 could gather and synthesize information, concepts and languages with a breadth, depth and speed beyond the capacity of human beings. It also hallucinated wildly and sometimes flunked basic math.
Altman’s decision to release GPT to the public allowed many of OpenAI’s competitors to reach for the moral high ground and claim that they would never do something so irresponsible. There’s some truth in that, but it’s just as true that their products did not have the goods. ChatGPT-3 did, and its arrival shook the planet, creating more curiosity about the future than all but a few technologies ever have before.
Sam Altman did that. We’ll find out soon enough what else he may have done — and what he wants to do next. Money is already lining up at his doorstep. Whatever he does next, Altman’s place in the history of AI is meaningful.