At the US Senate hearing on the opportunities and risks of generative AI, OpenAI CEO Sam Altman was asked about his worst nightmares about AI. But he didn't really want to answer like he had before.
He expects a "significant impact" on the job market, Altman said, but exactly what that impact will be is hard to predict. He suspects that there will be significantly more jobs "on the other side" of the AI revolution and that existing jobs will "get better."
Altman stressed that AI models are tools. "GPT-4 and things … other systems like it are good at doing tasks, not jobs," Altman said.
Still, Altman said, GPT-4 would likely fully automate away "some jobs," though he did not name specific occupations. OpenAI has published a study on the potential impact of GPT-4 on the job market, and there are early reports from the job market.
Altman said it is up to industry and policymakers, especially the latter, to address the potential problems of a changing job market. All in all, Altman is positive about our AI future: "I'm very optimistic about how great the jobs of the future will be." According to Altman, people are being "incredibly creative" in finding new jobs with new tools.
What is Altman's greatest fear?
With that answer, Altman initially dodged the senator's question about his worst nightmare. But after a friendly reminder from OpenAI critic Gary Marcus, who was also invited to the hearing, that Altman had not answered the question, the senator repeated his request.
But even on his second try, Altman would not really be specific. OpenAI, he said, had tried to be very clear about the risks of AI, which could cause "significant harm to the world" in "a lot of different ways." That's why OpenAI was founded, he said. "If this technology goes wrong, it can go quite wrong."
Altman was more explicit elsewhere: In an interview with "StrictlyVC" earlier this year, he called the extinction of humanity the worst-case scenario.
To end on a more heartwarming note: Marcus told the Senate that he felt Altman really cared about the dangers of AI, now that he had seen and felt him in person, standing next to him.