AI just cracked an 80 year old maths problem that humans have been trying to solve.

May 30, 2026 1:59 PM

“The math world is losing its mind over the new solution to an Erdős problem. This is what AI found, how we missed it—and why it matters.”
“ OpenAI announced that one of its models resolved a puzzle known as the unit distance problem without the help of any humans scribbling a bunch of equations on chalkboards. “
“….. , a former OpenAI researcher did some back-of-the-envelope math and estimated it took less than 32 hours….” Well, ….fuck.

https://apple.news/AY-v4q2_JQVqe4gP5s-HmHA

computers have been solving math since they were created. big whoop.

1 month ago | Likes 2 Dislikes 12

its what AI was originally invented for not making fanfics and editing people out of historic images

1 month ago | Likes 7 Dislikes 0

To be clearer, this is a good thing, this is like the use of AI for protein folding. The speed and efficiency is the reason for the “Well,..fuck.”

1 month ago | Likes 9 Dislikes 2

1 month ago | Likes 1 Dislikes 1

The combine mathematics achievements of every mathematician up to this point all jellied together solved it.

“ai” is not a little robot that looks like young Haley Joel Osment.

1 month ago | Likes 1 Dislikes 0

No one cares. Fuck AI

1 month ago | Likes 3 Dislikes 14

This is exactly what AI should be used for, and it takes far far far less resources to do this than it does to be data harvesters and embedded into every damn new device. Look at the work done on blood types. Look at the DNA work. We could remove many inherited illnesses in the next 5~10 years. AI /does not have to be the massive crapstain/ that most people think of.

1 month ago | Likes 36 Dislikes 1

This. There is most certainly uses for AI that could benefit a lot of areas which doesnt require massive burdening infrastructure. The problem is, we live in a greed fueled society ran by a handful of selfish elites looking to exploit people in anyway possible, and the data harvesting is more profitable to them than things like solving these types of problems.

1 month ago | Likes 6 Dislikes 0

We've also been doing stuff like this WITHOUT AI for a long time. Projects like Folding@Home, BOINC, the World Community Grid and GIMPS have been using distributed computing to solve massive mathematical and biological problems for at least a decade if not more. And that doesn't include "normal" super computing methods like using CRAY, ASICs, Gravity Pipe, etc.

1 month ago | Likes 3 Dislikes 0

True but all those projects used far more resources. The Cray-1 doesnt even compare to a Pi (about 4x faster). Edu's did beowolf clusters in the 90's and 00's that were less powerful than a pair of 4090's but were world changing at the time. F@H etc are cpu/gpu capable, use far more resources than a dedicated facility, its just distributed and they dont pay for the hardware because people donate it (they do have some of their own)

1 month ago | Likes 3 Dislikes 0

Yes, it does have to be that crap stain, because that’s how the Tech Bros will make money. They sort of research is not particularly profitable.

1 month ago | Likes 2 Dislikes 0

No it doesn’t because there are open models that no one makes real money off of that anyone can use and has nothing to do with “tech bros” making money

1 month ago | Likes 1 Dislikes 0

Do you think the people with data centers in their communities who can't afford electricity and are losing their water supply give a fuck about solving math problems?

1 month ago | Likes 1 Dislikes 2

Well, I would hope that those people would realize that those data centers are for “for profit” enterprises and not scientific research, as there’s very little financial return on solving old math problems.

1 month ago | Likes 2 Dislikes 1

The setups that solve math problems and do the dns and blood work use around 2 homes worth of electricity and zero water. The issues with the DC you are talking about are not even a thing for the type of setups we were talking about. Not all AI is the same, stop lumping it all as such.

1 month ago | Likes 3 Dislikes 1

Ai is unethical bc the creators/owners/users are unethical. This is amazing, but if people cannot continue to be taught paid to teach, and work on complex maths, then it's an ill gotten gain.

1 month ago | Likes 11 Dislikes 3

Something a person with a wisdom of about 19 would say.

1 month ago | Likes 5 Dislikes 2

Ooo edgy

1 month ago | Likes 4 Dislikes 1

But 19 is very high?

1 month ago | Likes 3 Dislikes 0

Wait, are we talking ab age or level?

1 month ago | Likes 2 Dislikes 0

We are talking in DnD stats.

1 month ago | Likes 2 Dislikes 0

This. If there's no one who understands the result, what good is it?

1 month ago | Likes 6 Dislikes 2

Yup! Or know if the result is a correct and valid one!! I forgot where I saw it (probably here) but they gave the best conceptualization, something along the lines of, "ai should be treated like a calculator, a tool to aid us in solving a problem, to take the tedium out of tasks we already know how to do".

1 month ago | Likes 2 Dislikes 0

And as you shift to simpler and simpler use cases to minimize hallucinations, a smart person starts to think, "Is an LLM actually faster than just writing a script to do this?"

1 month ago | Likes 3 Dislikes 1

::Gasps in CEO::

"What?!...and pay some dev to build it???"

1 month ago | Likes 2 Dislikes 0