What Musk, Altman, Brin agree on: Students should pay attention to computer science & maths

There is simply a quiescent panic creeping done classrooms and coding labs. Artificial quality present writes codification faster than students tin larn it. It solves equations without showing its work.

It answers exam-style questions with unsettling confidence. For a procreation raised connected velocity and shortcuts, the decision feels obvious: possibly machine subject has peaked; possibly maths is yet optional.That decision is incorrect and the radical who cognize it champion are the ones who helped physique the systems causing the panic. Over the past year, an improbable statement has emerged among planetary tech leaders. They are not urging students to pursuit the newest programming connection oregon maestro the trendiest AI tool.

Instead, they are pointing insistently backwards to subjects galore students are anxious to escape: Mathematics, machine subject fundamentals, physics, and theoretical thinking.This is neither nostalgia, nor world romanticism. It is simply a informing astir wherever worth is draining retired of the system. AI is not eliminating intelligence, it is commodifying the mechanical parts of it. The archetypal skills to flatten are those based connected repetition: Syntax memorisation, model fluency, surface-level coding competence.

What survives, and successful information becomes rarer, is the quality to crushed from archetypal principles, to exemplary a occupation abstractly, to recognize wherefore a strategy behaves the mode it does erstwhile it fails.That is wherefore immoderate of the sharpest minds successful exertion are present sounding astir conservative. They are arguing that maths is not a hurdle but a filter; machine subject is not astir penning codification but astir structuring thought; and difficulty, acold from being a flaw successful education, whitethorn beryllium its past remaining prime check.In an property obsessed with shortcuts, these voices are making an unfashionable case: that the hardest subjects inactive matter—not contempt artificial intelligence, but due to the fact that of it. Here, we analyse wherefore immoderate of the world’s astir influential tech figures are pushing students backmost to maths and halfway machine subject astatine the precise infinitesimal galore are tempted to determination away.

Pavel Durov: Maths trains independence

The provocation began quietly. In mid-2025, Telegram laminitis Pavel Durov posted proposal aimed astatine students weighing their options successful an AI-dominated future.

“If you’re a pupil choosing what to absorption on, prime mathematics,” helium wrote. No emojis, nary caveats.Durov did not framework mathematics arsenic a warrant of employment. He framed it arsenic a subject that forces autarkic thinking. Maths, arsenic helium has often argued, does not let the luxury of imitation. You either recognize the problem, oregon you don’t. In a satellite wherever AI offers instant answers, that favoritism matters more, not less.

Durov’s subtext was unmistakable: reliance connected tools is rising; intelligence self-reliance is becoming scarce.

Elon Musk: First principles, oregon nothing

Durov’s station drew a effect from Elon Musk that became viral precisely due to the fact that of its brevity. Replying publically connected X, Musk wrote: “Physics (with math).”The 2 words were not a curriculum. They were a philosophy. For Musk, physics is the arena wherever first-principles reasoning is unavoidable, assumptions are tested against reality, not convenience.

Maths is the connection that makes that investigating precise. When Musk says “physics (with math)”, helium is rejecting aboveground competence. He is arguing that arsenic systems turn much complex—rockets, autonomous vehicles, large-scale AI—the punishment for shallow knowing becomes catastrophic.In the AI era, Musk’s connection was stark: tools volition alteration weekly; archetypal principles endure.

Sam Altman: High-leverage moment

By precocious 2025, a antithetic anxiousness had taken hold: that AI had made machine subject itself a mediocre world bet.

Speaking astatine Stanford University successful a nationalist speech with cryptography prof Dan Boneh, OpenAI CEO Sam Altman addressed that fearfulness head-on.“This is simply a truly chill clip to beryllium studying machine science,” Altman said. “It’s a high-leverage moment, particularly if you’re funny successful AI.”Altman’s phrasing was deliberate. “High leverage” does not mean casual returns. It means that knowing foundational systems present carries outsized interaction later.

AI, successful Altman’s telling, is not replacing machine science; it is concentrating powerfulness successful the hands of those who recognize however these systems are built, constrained, and deployed.For students, the accusation is uncomfortable but clear: shallow familiarity volition property badly. Structural knowing volition compound.

Demis Hassabis: The subject of difficulty

If Altman speaks arsenic a strategist, Demis Hassabis speaks arsenic a merchandise of world rigour. In a 2025 speech connected the Lex Fridman podcast, the Google DeepMind CEO reflected connected the formative interaction of his education.“I took immoderate precise hard mathematics and theoretical machine subject courses,” Hassabis said. “They taught maine however to deliberation profoundly and rigorously—and however to persist erstwhile things were hard.”He returned to the taxable aboriginal that twelvemonth astatine nationalist forums, cautioning students against abandoning maths and mentation simply due to the fact that AI tools look to marque them redundant. The existent worth of those subjects, Hassabis argued, lies not successful the contented itself but successful the cognitive grooming they impose: precision, patience, and the quality to wrestle with problems that defy speedy solutions.

In an epoch wherever answers get instantly, the capableness to beryllium with uncertainty becomes a competitory advantage.

Sergey Brin: Passion and caution successful the AI era

At a clip erstwhile students are proceeding 2 conflicting narratives — AI volition regenerate jobs and AI volition regenerate degrees — Google co-founder Sergey Brin offered 1 of the astir grounded responses successful January 2026 portion speaking to a caller procreation of engineers astatine Stanford University.His words were elemental but layered. “I chose machine subject due to the fact that I had a passionateness for it.

It was benignant of a no-brainer for me. I conjecture you could accidental I was besides fortunate due to the fact that I was besides successful specified a transformative field,” helium said. Brin’s accent connected curiosity alternatively than credential chasing was deliberate. He pointed retired that his ain travel — from a Stanford postgraduate pupil to co-architect of Google — was driven by interest, not fear-based vocation calculus. In an property of generative AI, wherever models specified arsenic Gemini and ChatGPT tin constitute and debug code, that favoritism matters much than ever.

Importantly, Brin did not halt astatine passion. He besides addressed the anxiousness astir automation head-on.

With diagnostic candour, helium quipped, “I wouldn’t spell disconnected and power to comparative lit due to the fact that you deliberation the AI is bully astatine coding. The AI is astir apt adjacent amended astatine comparative literature, conscionable to beryllium perfectly honorable anyway.”His constituent was twofold: Don’t fly STEM retired of fearfulness of automation, and don’t presume AI’s existent show undermines the worth of structured learning.

If you disregard the fame...

Fame is simply a distraction. Take the celebrities retired of it — Musk tossing “physics with maths” similar it is simply a mic-drop, Altman selling machine subject arsenic a “high-leverage” stake from a Stanford stage, Hassabis sounding similar the people topper who really enjoyed theoretical CS, Brin reminding everyone helium picked CS due to the fact that helium genuinely liked it — and the statement stops being glamorous. In fact, it starts being annoyingly sensible.

AI is not stealing intelligence, it is bulk-discounting the easier bits of it. The worldly that erstwhile passed arsenic accomplishment — regular coding, look application, template reasoning — is present a vending machine. What inactive refuses to automate is judgment: Spotting the atrocious assumption, knowing erstwhile an reply is plausible and erstwhile it is polished nonsense. That is wherefore they support returning to maths and halfway CS.