From our other sites
The story
A thread titled “Is there any point in learning math?” blew up on the Nanidemo Jikkyou (nanG) board. It kicked off with a post arguing that since LLMs now outperform mathematicians at math, the whole subject has lost its purpose — and that class time should be shifted to science instead. That drew a wave of pushback: math builds logical thinking, math is a language and a tool you need, and so on. The debate spiraled further into whether AI has actually solved one of the Millennium Prize Problems. In the end, the discussion never quite resolved, with posters talking past each other — some arguing about the ability to solve math itself, others about the ability to use math as a skill taught in school.
What people said
LLMs killed math.
Since LLMs are better at math than mathematicians now, I think the point of learning math has basically disappeared.
Why not just cut math and shift that time, scope, and exam weight into science instead?
Since raw math ability can never beat AI, it's pointless.
AI >>>>> mathematicians
That's just how it is.
Logical thinking is needed for literally any subject.
Even memorizing history requires logical thinking.
Huh?
Can you not read Japanese?
If you've got a counterargument, make it logically lol
You're supposedly good at this, right? lol
No matter how much math you study, you'll never beat AI, and mathematician as a profession is going away too.
Raw math ability genuinely means nothing anymore.
Thanks to AI, math's only remaining value is as the language of natural science.
What matters is 'the ability to use math as a language.'
That's why math should be cut and that time, scope, and exam weight shifted to science.
Nah, there is a downside.
No matter how hard you study, AI will still be better at math than any mathematician.
What you're describing is 'math as language/tool.'
That's exactly why I'm saying it should get more time — folded into science class.
My position is that 'math as a language/tool' should be emphasized even more within science than it is now.
Physics is a natural science.
Math is a formal science.
They're completely different disciplines.
LLMs are the end of formal sciences.
AI now outperforms any human at formal sciences in general, not just math — and unlike natural science, AI can even self-verify the results, so it's truly over for that field.
Sounds like the kind of guy who'd fillet a mebaru (a type of rockfish) lol (random derail — classic 5ch non-sequitur humor)
You can do that with science too, right?
If so, rather than wasting effort on pointless study, better to put that time into science instead.
No, like I said…
The point of learning math is training the ability to work out the correct answer from the materials you're given…
You can do that with science too.
So then science is what should get the emphasis.
Don't get me wrong though —
I'm not saying 'scrap math class!'
I'm saying 'scrap math class and shift that time into science!'
What's needed isn't 'raw math ability' but 'the ability to wield math as a language/tool.'
Since AI is becoming better than any mathematician at 'raw math ability,' that skill isn't needed anymore.
That's like saying weightlifting is pointless because excavators exist.
If it's going to be taught in science anyway, isn't it better to learn it systematically in math first?
Wait, do you seriously think compulsory and secondary education teaches math so kids can go prove new theorems!?
What even is 'math' outside of it being a language?
It's to scout future mathematicians, right?
But that profession is disappearing, so it's pointless.
'Math as a language/tool' gets built up through science class instead.
That's way too much of a leap.
With math, at the very least, mathematicians are becoming unnecessary.
An LLM apparently solved the Navier–Stokes Millennium Prize Problem in a few days, and they've announced 'substantial progress' on the other Millennium Problems too.
Rumor has it the Hodge conjecture is already on the verge of being proven.
That Navier–Stokes thing was just the AI ripping off an unpublished idea from a researcher and taking credit.
How exactly would it 'rip off' something unpublished?
For the other Millennium Problems too, it made 'substantial progress' within days and is close to a proof.
And honestly, even that researcher's Navier–Stokes idea, while groundbreaking, was still nowhere near the proof OpenAI actually produced.
Kinda funny when you think about it.
It's the other way around.
Logic itself ultimately leads you to mathematical thinking — that's what my symbolic logic professor said back in college.
Are you guys math majors or something?
Physics/engineering and medicine, actually.
I'm just a lowly engineer who thinks math is genuinely necessary to learn.
Couldn't that just fall under science? Engineering is science, isn't it?
I'm talking about math as taught in secondary education, up through high school.
Go ahead, try deriving Maxwell's equations without using math.
Whatever an ordinary person does in science or scholarship, AI can already do.
And who's going to come up with the questions in the first place?
You can't tell if someone's a genius unless you actually let them try, can you?
Natural science is a no-go.
Experiments and observation are absolutely required, and physics simulations are too computationally heavy — it's just not feasible.
Replacing that would require at least running AI on quantum computers.
By its very nature, natural science will be the last thing standing.
Background and Key Points of This Debate
The Millennium Prize Problems, math’s most famous unsolved challenges, remain officially unsolved as of 2026. The claim floated in the thread — that an LLM solved the Navier–Stokes problem in a matter of days — was itself disputed within the same thread, with one poster countering that the AI simply repackaged a researcher’s unpublished idea. In other words, this isn’t an established fact. What AI has actually demonstrated is the ability to solve International Mathematical Olympiad–level problems and assist with parts of proofs — there’s no confirmed case of it independently producing a complete, peer-reviewed proof.
The thread’s real fault line was the distinction between “the ability to solve math itself” and “the ability to use math as a tool or language.” Even granting that the former can be handed off to AI, many pushed back that secondary school math education is exactly where the latter gets built, and that science class alone can’t teach it systematically. Because opinions on the value of the discipline itself and opinions on math’s place in the school curriculum tend to get tangled together, it’s worth keeping track of which argument is actually being made at any given point.
*This article is compiled and summarized from the 5ch (Nanidemo Jikkyou board) thread ““Math” 👈 is there even a point learning this lolololololol.”
Leave a Reply