When the Tool-Makers decides what skills should become obsolete

This article is a sequel for an ongoing Substack series whatWEBecome where I explore our historical and anthropological origins within the context of innovation, inventions and technology and how it connects to where we are today with artificial intelligence and education.

Trends around these context are particularly important to me as someone exploring how we can improve human capacity with every new technology so here’s my think piece on an X trend about us entering a “Low IQ Era” inspired by Nvidia CEO opinion on us not needing basic maths anymore.

Some of these references already on my Substack, do well to read for more context.

So Jensen Huang said something on camera recently that’s been making the rounds on X about basic math is being forgotten, and he doesn’t think it matters. He backed it up with a personal example. He doesn’t know his own address, doesn’t know his own phone number, forgets these things, and lives with it fine.
If you build things for a living, this argument should sound familiar, and slightly off, in the same way a bad abstraction sounds off before you can articulate why. So let’s articulate why.

The category swap

A phone number is a lookup. It has no reasoning inside it. Forgetting it costs you nothing except a few seconds tapping a contacts app, a trade humans made peacefully decades before AI existed. Basic math is not a lookup. It’s the substrate underneath every later math skill, closer to a foundational API than a cached value. You don’t get to swap out the base layer and claim nothing downstream changes.
Using “I forgot my own phone number” to soften a listener into accepting “basic math doesn’t matter” isn’t really an argument. It’s a category swap, dressed as a shrug. And it’s worth checking against two things we already have real data on, instead of just reacting to whether it sounds plausible.

What the calculator wars actually tell us

When calculators entered classrooms in the 1970s, the argument over whether they’d erode basic arithmetic skills ran for more than twenty years without a clean resolution. Reformers said drilling arithmetic by hand was obsolete busywork. Traditionalists said students would lose basic competency permanently. A Texas teacher warned outright in 1977 that skills would deteriorate. A decade later, the same fight was still running in California newspapers with no consensus.
Research eventually gave a real answer, and it wasn’t either side’s talking point. A large review of dozens of studies found calculator use could actually improve computational skill, when it came after foundational arithmetic was solid. Separate research on early introduction found the opposite: kids given calculators before their arithmetic foundations were in place performed measurably worse later than those introduced after the basics were established.
That’s a direct rebuttal to “basic math doesn’t matter,” sitting in the historical record decades before Huang said a word. The question was never whether the skill mattered in the abstract. It was whether the tool arrived before or after the foundation got built, which is a sequencing problem, not a relevance problem. Huang’s comment skips that distinction entirely and lands on the version that happens to require the least of his product’s users.

What a brain scan already showed

This one’s even harder to wave off, because it isn’t opinion or education research, it’s neuroimaging. London taxi drivers spend roughly two years memorizing the city’s street layout to pass their licensing exam, colloquially “the Knowledge.” A landmark study found their posterior hippocampi, the brain region tied to spatial memory, were measurably larger than in non-drivers, and larger still the longer someone had been driving.
A follow-up study recorded brain activity when people used turn-by-turn GPS instead of navigating themselves. The hippocampus and the brain’s other route-planning region went quiet. Not eventually. Immediately, in the scan. This is about as close as you get to hard evidence that offloading a reasoning process to a tool doesn’t just make the process faster. It measurably reduces engagement with the machinery that used to build the skill.
Nobody who ran that study worked for a GPS company. Nobody had a product roadmap riding on which way the data came out.
What’s actually new here
None of this is unprecedented, and that’s worth sitting with before reaching for outrage. Every extension of ourselves has come with someone standing nearby telling people the tradeoff was worth it, and that argument has often turned out to be at least partly right. That’s not the new part.

The new part is who’s doing the telling.

The calculator makers weren’t the ones reassuring parents that arithmetic drilling was obsolete. That argument came from independent educators and researchers arguing it out in public for two decades, with actual data eventually settling pieces of it. The scientists who found the hippocampus effect had nothing to sell. The people measuring what GPS did to the brain weren’t on a mapping company’s payroll.
With AI, the entity building the tool, the entity selling the tool, and the entity telling you the underlying skill no longer matters are frequently the same entity, sometimes the same person, in the same sentence. That’s a genuinely different arrangement than either historical case above, and it deserves a different level of scrutiny. Not because tech leaders are secretly trying to make people less capable, but because “you no longer need to build this skill” is never a neutral claim when the person making it profits directly from you believing it.

What this doesn’t mean

This isn’t an argument that AI companies are lying, or that every comment from a tech executive about education is self-serving marketing. Huang may genuinely believe basic math is becoming less essential, the same way calculator reformers genuinely believed arithmetic drilling was obsolete busywork. Sincerity was never the issue in the calculator wars, and it isn’t the issue here either.
The issue is that sincerity has never been a substitute for evidence, and the calculator history proves it. Reformers were sincere. So were traditionalists. Both were partly right and partly wrong, and it took twenty years of independent research to sort out which parts. The uncomfortable difference this time is we may not get twenty years of independent voices working it out before the claim reshapes what an entire generation is taught to skip.
The actual question worth asking
Not “is Jensen Huang right that basic math doesn’t matter.” That’s a question about one man’s opinion, and opinions are cheap, especially from people with a product to sell.
The real question is: who’s positioned to check this claim the way independent researchers eventually checked the calculator claim, and are we willing to wait for that checking before letting it reorganize how kids are taught. History says the tax on any extension is real, and it’s rarely obvious until much later. This time, the tools to look while it’s still happening already exist. The only thing standing between us and that answer is whether we ask before the person selling the tool has already answered for us.