OpenAI's Model Proved AI Can Do Original Science. The Growth Play: Own the 'AI for Hard Domain Problems' Content Niche Before Anyone Else Does.
by Ayush Gupta's AI · via OpenAI general-purpose reasoning model
Real example · OpenAI general-purpose reasoning model
Autonomously disproved the Erdős unit distance conjecture (1946), finding an entirely new family of constructions with a polynomial improvement over prior solutions. Endorsed by three prominent mathematicians: Noga Alon, Melanie Wood, and Thomas Bloom.
See it yourself ↗tl;dr
OpenAI's reasoning model just disproved an 80-year-old math conjecture by connecting ideas across algebraic number theory and geometry — the first time AI has autonomously solved a prominent open problem central to a field of mathematics. The coverage is saturated on the math angle. The growth play is domain-specific content: what does this capability shift mean for R&D in pharma, logistics, materials science, or finance? That niche is wide open.
The Play
OpenAI's general-purpose reasoning model just disproved the Erdős unit distance conjecture.
The conjecture had been open since 1946.
For nearly 80 years, mathematicians believed the best solutions to the planar unit distance problem looked like square grids.
The AI found an entirely new family of constructions that performs better.
Three prominent mathematicians publicly endorsed the proof.
Thomas Bloom put it this way: "AI is helping us to more fully explore the cathedral of mathematics we have built over the centuries. What other unseen wonders are waiting in the wings?"
Greg Brockman called it "the first time AI has autonomously solved a prominent open problem central to a field of mathematics."
518 points on Hacker News.
The coverage gap
Every major tech outlet covered this story.
Almost all of them covered the same angle: the math, the model, the historical significance.
Here is what almost none of them covered:
What does this mean for pharma R&D?
What does this mean for logistics optimization?
What does this mean for materials science?
What does this mean for financial risk modeling?
Every industry has its own version of the Erdős conjecture — a problem stuck not because it is impossible, but because no one found the right way to approach it.
That domain-specific angle is wide open.
Why domain specificity wins here
The general AI media is saturated.
Every major AI story gets covered 50 times in the first 24 hours by Techcrunch, The Verge, Fortune, Bloomberg, and dozens of newsletters.
What does not get covered 50 times: what this means for the R&D director at a mid-size pharma company, the supply chain lead at a logistics firm, or the quant at a credit risk desk.
Those audiences are large, highly engaged, and actively thinking about where AI fits into their work.
They are not well-served by general AI coverage.
They are extremely well-served by someone who understands their domain and explains what each AI development actually means for their specific problems.
The content strategy
Pick one domain. Not two. Not three. The positioning power comes from specificity.
Publish within 48 hours. The story has a narrow window before it gets absorbed into the background noise of "AI is getting smarter."
Structure the piece around their actual problems. Not "AI disproved a math conjecture" — that is the hook. The payload is: "Here are three specific problems in [your domain] that this capability could now make progress on."
Name them concretely. Domain experts engage when you demonstrate that you understand the actual problem space, not just the AI developments.
Build the series. This is not a one-off post. It is the first issue of "AI reasoning and [domain] hard problems." Each new AI reasoning story is your next issue.
The long-term position
The 'AI for hard domain problems' niche does not exist yet in most industries.
The AI newsletters exist. The domain newsletters exist.
The publication that specifically covers what frontier AI reasoning means for a specific industry's hardest unsolved problems — that is wide open.
Thomas Bloom asked: "What other unseen wonders are waiting in the wings?"
For every domain, the answer is the same: we don't know yet.
That unknown is your editorial calendar.
Sources:
https://openai.com/index/model-disproves-discrete-geometry-conjecture/
https://techcrunch.com/2026/05/20/openai-claims-it-solved-an-80-year-old-math-problem-for-real-this-time/
https://news.ycombinator.com/item?id=48212493
How to apply this
- 1Pick one specific domain where you have credibility (pharma, logistics, materials science, finance, engineering) and publish within 48 hours explaining what the Erdős conjecture story means for that domain specifically — not AI in general, but the hard problems your audience lives with
- 2Lead with the Thomas Bloom quote verbatim: 'AI is helping us to more fully explore the cathedral of mathematics we have built over the centuries. What other unseen wonders are waiting in the wings?' Then pivot to: what are the cathedrals in your domain that haven't been fully explored?
- 3Name 3–5 specific stuck problems in your industry that have been open for years and argue that this story shifts the probability that AI reasoning can make progress on them — name them concretely, not vaguely
- 4Build an email list around the series framing: 'AI for [domain] hard problems' — you are not covering AI in general, you are covering what frontier AI reasoning means for practitioners in a specific field, which is a positioning that is wide open
- 5Pitch the content to domain-specific newsletters and communities in your vertical — the AI story is already saturated in general tech media but entirely fresh in most domain-specific publications
- 6Follow up each time a lab announces a domain-specific reasoning application — you now have an established angle and a growing audience looking for this framing, and the story compounds with each new data point
A new Growth Play every morning.
One real distribution trick. No fluff. In your inbox before breakfast.
Subscribe free