Sam Altman says he correctly predicted how AI will develop. But there is one prediction that he was not entirely correct.
“We feel very correct in technical predictions, and then we thought society would feel much different if we actually broadcast it than before,” Openai CEO Altman said in a recent episode of “No Cap with Jack Altman.” “But I am not. It's not even clear that it's a bad thing.”
Altman believes Openai has “cracked” reasoning in the model, and states that O3 LLM in particular is equivalent to a person with a “doctoral degree.” On many subjects. The trajectory of technology is almost predictable, but Altman said it is not as proportionately as people expected.
“Models now allow us to make a kind of reasoning in a particular domain where we expect to be able to have a PhD in that field,” he said. “In a way, I think, 'Oh, AIS is like the top competitive programmer in the world right now'' or 'AIS can be like the top score in the most difficult mathematics competition in the world' or 'AIS can prefer to be able to expect a doctorate of experts in my field.'
Although AI usage is on the rise, society has not yet changed at major leaps and boundaries. AI is already affecting businesses, and many companies are adopting AI tools, and in some cases they use them to enhance or replace the human workforce.
Altman thinks the response to this technology is relatively overwhelming when he endures what he considers as a future potential.
“If I told you in 2020, we'd say, “We'd make something like ChatGpt and be as smart as PhD students in most fields. And we'll unfold it. “Maybe you'd have believed it, maybe you don't have it.”
“But conditioned by that, I would say, 'Okay, if that happens, the world looks so different than it is now',” he added.
Altman acknowledges that AI is the most useful now as a kind of “co-pilot,” but foresees major changes if it can act autonomously, especially considering potential applications in science.
“You already hear the scientists say they're faster with AI,” he said. “There is no AI that does science autonomously, but if human scientists are triple productive with O3, that's still quite a big deal. And it continues, and AI can do science autonomously and find new physics…”
From a risk perspective, Altman said he wasn't too worried despite other AI leaders like humanity's Dario Amodei and Demis Hassavis of Deepmind.
“I don't know there's a high risk. I think the ability to make biological weapons, or to defeat the entire country's grid, can be very harmful without physical stuff,” he said. “It's a silly way to be risky. For example, I'm afraid of walking someone walking around my house, unless I really trust them.”
Openai did not immediately respond to a request for comment by Business Insider.
For now, according to Altman, life remains relatively constant. But if things start to snowball, and he believes they will – he doesn't think specifically about what the world will look like.
“I think we'll reach a very smart and capable model. We can discover important new ideas and automate a huge amount of work,” he said. “But if that happens, I feel completely confused about what society will look like. So, I seem to be most interested in the ability question, but at this point, how do more people ensure that society will gain value from now on?”
