A few weeks ago, one of the hardest problems in mathematics was apparently solved by AI.
The Navier–Stokes problem is one of seven Millennium Prize Problems. It had resisted mathematicians for decades. Solve it and there’s a $1 million prize waiting for you.
OpenAI reportedly threw around 10,000 AI agents at it.
Within days, they had an answer.
There’s now a controversy around what happened. Questions about who deserves credit. Whether the AI could have been influenced by the unpublished work of mathematicians already working on the problem. And what happens when an individual researcher can spend years pursuing an idea, only for an organisation with enormous amounts of compute to point thousands of AI agents at the same problem.
At first, I thought this was another story about AI replacing humans.
But the more I thought about it, the more it took me somewhere else.
Back to why I became a physicist in the first place.
I studied physics because I wanted to understand things.
That was pretty much it.
I was fascinated by how the world worked. I loved learning. I loved that feeling of following a question and discovering that underneath it was another question, and underneath that another one.
Eventually I did a PhD in atomic physics.
And that’s when I got to see more closely how science actually gets done.
There was still curiosity, of course.
But there were other forces at work too.
Funding.
Careers.
Recognition.
Politics.
Status.
Egos.
Who got credit for what.
Which research was worth pursuing.
Which research someone would actually pay you to pursue.
At one point we got money for doing nanotech while my research was about bouncing atoms off videotape.
None of this should have surprised me. Scientists have mortgages and families and careers like everyone else. Research requires equipment, institutions, salaries and time.
But I think I had a rather naive relationship with science.
I wanted it to be about curiosity.
And suddenly I could see all this other stuff wrapped around it.
At the time I didn’t have particularly good ways of dealing with that. I avoided confrontation. I wanted people to be happy. Politics and competing agendas made me uncomfortable.
I remember looking at the path ahead and thinking:
I don’t think I want this.
If I was going to do this for curiosity, I didn’t want to have to worry about the money. And if I was going to do things for money, there were more lucrative and less challenging ways to do that.
It’s probably part of the reason I didn’t go into corporate life either. I knew that in competitive environments like that my soul would slowly shrivel.
And, strangely, this story about AI solving a maths problem has brought me back to that feeling.
Because underneath the argument about AI and mathematics, I wonder if there are three different things getting tangled together.
Curiosity. Economics. Status.
Curiosity says:
I want to know.
Status says:
I want to be the person who discovers it.
Economics says:
I need the resources and time that allow me to keep discovering things.
Of course, they’re not neatly separable.
If I’m honest, I’m not sure my own curiosity about physics was ever completely free of status. There was something appealing about being a physicist. About being good at something difficult. About knowing things other people didn’t know.
And economics makes curiosity possible. Someone has to pay for the laboratory, the equipment and the time to sit around thinking about atoms.
We’ve built an enormous institutional system around all three.
Universities. Research grants. Academic journals. Professorships. Patents. Prizes.
Even the $1 million Millennium Prize is interesting in this context.
The Clay Mathematics Institute says part of the purpose of the prizes is to draw attention to the frontier of mathematics and emphasise the importance of tackling extraordinarily difficult problems.
In other words, we’re using economic value to signal intellectual value.
This question is worth a million dollars. Pay attention.
And then AI arrives.
Suddenly the economics change.
A human mathematician might spend years developing an intuition about a problem.
An organisation with enough capital can potentially deploy thousands of AI agents to explore it simultaneously.
That doesn’t necessarily damage curiosity.
In fact, from the perspective of pure curiosity, it’s extraordinary.
We wanted to know the answer.
Now we know it.
And knowing it creates more questions.
That’s how science has always worked.
We stand on the shoulders of giants so we can see a little further.
What happens if the giant suddenly grows incredibly quickly?
Maybe we get to see much further.
But AI potentially disrupts the other two forces.
If I spent ten years trying to solve something and an AI solves it tomorrow, perhaps part of my disappointment isn’t:
Damn. Now humanity knows the answer.
Perhaps it’s:
Damn. I wanted to be the one who found it.
That’s status.
And perhaps another part is:
If machines can do this, what happens to my career?
That’s economics.
Neither response is shameful. They’re deeply human. People need livelihoods. Recognition matters to us.
But I think separating them from curiosity is useful.
Because it leads to a thought experiment I find strangely hopeful.
Imagine mathematicians didn’t have to worry about earning a living.
Imagine their status didn’t depend upon being first to publish.
And then an AI solved the problem they’d been working on for ten years.
What would a curious scientist do?
I suspect they’d look at the answer and say:
That’s fascinating.
How did it do that?
Does that technique work somewhere else?
What happens if we change this assumption?
Does this tell us anything about the physical world?
What question does this allow us to ask next?
The inquiry doesn’t end.
It expands.
And maybe that’s one of the more interesting possibilities AI presents.
We’ve spent a lot of time worrying about what happens when AI has better answers.
Perhaps we should be thinking more about what happens when answers become cheap.
Because if answers become abundant, something else becomes more valuable.
Questions.
Knowing what is worth asking.
Knowing which answers matter.
Knowing when an answer is mathematically correct but physically meaningless.
Connecting something discovered in one field with something happening somewhere completely different.
Being discerning about which possibilities we pursue.
And ultimately deciding what we do with all this new knowledge.
AI might massively increase our ability to answer questions.
But it can’t make the question of what humanity should care about disappear.
If anything, it makes that question more important.
And this takes me back to the young physicist I was.
Maybe what I loved wasn’t physics itself.
Maybe what I loved was inquiry.
Following curiosity wherever it went.
Trying to understand something I didn’t understand yesterday.
I walked away from academic science partly because I discovered that curiosity didn’t operate in a vacuum. Economics, politics, ego and status were there too, and at that point in my life I didn’t particularly know how to navigate them.
Thirty years later, AI is exposing those same forces in a completely new way.
And I’m surprisingly heartened by it.
Not because I want machines to replace scientists.
And not simply because we’ll be able to find answers faster.
But because perhaps it forces us to become clearer about what the human contribution really is.
Maybe our greatest value was never having all the answers.
Maybe it was our capacity to wonder.
To notice.
To question.
To choose what deserves our attention.
And, when the answers arrive, to decide what they’re for.
If AI gives us the ability to stand on the shoulders of giants taller than anything we’ve previously imagined, the most important question might not be how high they can take us.
It might be:
Where do we want to look from up there?