Eric Schmidt describes his career as spanning the IBM mainframe, PC, internet, social media and AI eras.
“I started off in the IBM uh mainframe era, and through the PC era, then the internet area, then the social media era, and now AI.”
Person
Eric Schmidt
1 episodes · 24 quoted claims
Schmidt says advanced two-nanometer chips, largely made in Taiwan, are one driver of AI progress.
“Literally, the physicists making the chips faster, moving down to two nanometer chips, which are largely made in Taiwan, incredible improvements in memory”
Schmidt says Anthropic's revenue growth was more than twice as fast as Google's record growth during his time as CEO.
“If you look at Anthropic's revenue growth, when I was CEO at Google, we set the revenue growth record. They smashed it more than twice the speed as we did, right?”
Schmidt says AI models take about six months to build and two months to test, and become antiquated within about five months of release.
“The models, for example, take roughly six months to build, two months or so to actually test, and they're released and they're antiquated within five months.”
Schmidt says frontier AI models cost hundreds of millions of dollars to make.
“The models themselves cost hundreds of millions of dollars to make. Some estimates are much higher. I'm using conservative numbers.”
Schmidt says ten major math problems thought to be unsolvable were solved with AI in the month before the interview.
“In the last month, ten major math problems, which I cannot possibly understand, were solved, and they were not thought to be solvable.”
Schmidt says current AI trained only on math known in 1902 cannot discover general relativity, which he uses as a test of real discovery.
“build a model that has all the math known in 1902. Can you discover general relativity? In other words, you know nothing past 1902. The current evidence is that you can't.”
Schmidt estimates AI capable of genuine new discovery is five to ten years away.
“We don't the the. This is a long-winded way of saying we don't know. It's somewhere between my estimate is five to ten years.”
Schmidt says Anthropic writes most of its code using its own AI.
“Anthropic is busy writing most of its code using its own code. In other words, the the human tells it to write something, it writes the code, and then the code gets rewritten by the computer.”
Schmidt says math progresses fast because it is scale free and needs no new data, unlike physics and chemistry.
“The mathematicians invent stuff in their own head, whereas in physics and chemistry, you actually have you have to have real or simulated data.”
Schmidt names Mistral in France and Helsing in Germany as European AI companies, and says the UK does better.
“There are a few companies, uh, Mistral in France, Helsing in Germany, uh, the UK does better.”
Schmidt says he helped write the US AI strategy while Brussels wrote its own about five years before the interview.
“five years ago, Brussels wrote the um AI they their AI strategy. I helped write the one for the US.”
Schmidt says Europe does not have its own foundation model.
“Europe does not have its own foundation model. There are people working on this, but doesn't have one.”
Schmidt says Europe has among the highest energy costs in the world, which makes training AI models there hard.
“Europe has among the highest energy costs in the world. Okay, no problem, do the training somewhere else, right?”
Schmidt says he proposed to President Macron that firms with up to 20 employees be subject only to health and safety labor laws.
“for a company that's between zero employees and 20, the only laws that apply are health and safety. None of the other ones do, but when you get to 21, all of them do”
Schmidt cites ASML as proof that a vertically integrated technology company can succeed in Europe.
“ASML is one of the most important companies in the world. It's completely vertically integrated, and it's vertically integrated inside of Europe”
Schmidt says Chinese AI developers have adopted an open-weight strategy, releasing their models publicly.
“The Chinese have adopted an open weight strategy, so they give the software and they give the so so forth.”
Schmidt describes distillation as asking a large model many questions to learn to imitate it, and says Chinese models appear to have used it.
“There's a technique called distillation, where you basically ask the big model a series of questions, and then with that, you learn enough to think like the big model.”
Schmidt estimates Chinese AI models are about six months behind American ones.
“It looks like they are six months, uh, six months behind.”
Schmidt says European acquaintances were upset about a US decision restricting access to the Methos model.
“My European friends were particularly upset about the banning of Methos.”
Schmidt says Mark Zuckerberg spent billions on virtual reality before pivoting to AI, as an example of founder power.
“Let's use Mark Zuckerberg, who clearly controls his company. Um, he decided one day to work on virtual reality. He spent billions of dollars largely, which didn't work very well, and then he pivoted to AI”
Schmidt argues that only founder Steve Jobs could oversee Apple's transformation, and that others who tried could not.
“Could anyone else have overseen the transformation of Apple, except for Steve Jobs? A number tried, they couldn't do it.”
Schmidt says he told people he worked about 63 hours a week at Google, mostly in meetings.
“I used to tell people my job is to sit in meetings 12 hours a day, five days a week, and I work on the weekend too.”
Schmidt names Google's failure to stay ahead in social media while he was CEO as his biggest professional failure.
“There was a moment when Google had was ahead of everybody else in social media. It was fairly short. And I think, and I take, I was a CEO at the time.”