Jensen Huang says it doesn't matter if kids forget long division
Nvidia's chief executive was handed research showing that AI helps students finish homework without learning the material. His answer was that the skills in question were never worth much — and offered his own life as proof.
Jensen Huang's first long interview since Nvidia became the load-bearing company of the AI boom is a case for not mourning what the technology erases. In the New York Times's Ezra Klein Show, published on 23 September, Klein put a Chinese study to him: AI tools let students get homework done faster, but the skills the work was meant to teach don't stick. Huang agreed the retention problem is real and likely connected to the tools, then said it doesn't matter. "Try to get a kid to do long division right now," he said. "The multiplication table is starting to be forgotten. Doing square roots, my goodness. Basic math is being forgotten. Does it matter? I don't think it does. I don't think it does."
Klein pushed back — surely some set of skills matters — and Huang answered that "there are a lot of skills that don't matter," then volunteered his evidence. He does not know his own address. He once panicked at a gas station when the pump asked for his ZIP code. He does not know his phone number either. "I forget these things. I can live with it." He also told Klein that AI will produce a net gain in jobs.
The interview is worth taking seriously because the person making the claim is not a commentator. Nvidia's market capitalisation sits around $5.4 trillion, and the Times has reported that 15 cents of every dollar the US stock market has returned since 2023 came from Nvidia stock — so Huang is the single largest beneficiary of the assumption that AI capability keeps compounding. When he says the skills our institutions are built to teach can be written off, he is describing the world his own revenue depends on.
Here is the gap. The retention problem he conceded is not speculative, and the direction of travel is negative: we covered the OECD finding that students who lean on AI for schoolwork score a year and a half behind, a result that echoed a study of 26,811 Chinese secondary students where short-term homework scores rose about 18% while term exams fell 20% six months later. MIT's own committee report reached for the phrase "cognitive surrender" — MIT report warns of 'cognitive surrender' in AI-era classrooms — and its prescription was to move assessment toward oral exams and process work, precisely because the artifact no longer proves anything about the student.
Huang's answer is that new skills will be discovered to replace the old ones. That is a forecast, not a finding — and note what his own example actually demonstrates. Not knowing your ZIP code is not a skill rendered obsolete by intelligence; it is a skill rendered unnecessary by money. Someone else handles the errands, and the abstraction is the point. That is a fine argument for his household and a bad one for a school system deciding what to teach.
The framing that made the round of commentary worth reading came from the technology critic known as tante, who argued the statements function less as arguments than as tests — The Test — loyalty checks where the cost of failing to nod along is the interview itself. Klein had the transcript in front of him and let "I don't know my address" pass as an anecdote about automation rather than an admission of insulation.
What to watch: whether the labs' own evaluation practice moves toward process evidence — the thing MIT recommended — or keeps scoring outputs. Whatever they choose, it is what the next generation of students gets graded on.
A CEO who can't be bothered to know his own ZIP code is telling you which side of the abstraction he lives on. Does that make his argument about your kid stronger or weaker? Tell us in the comments.
