A question I started asking after an AI model disproved a math conjecture and what it revealed about the most expensive blind spot in my business.
There was a moment this past week that stopped me.
OpenAI announced that one of its models disproved a central conjecture in discrete geometry. A problem that human mathematicians had worked on without a solution — solved by an AI that was not expected to succeed.
I am not a mathematician. That is not the part that stopped me.
What stopped me was the implication. If AI can find a solution in mathematics that the experts in the field could not, then it can find the things in my business that my own expertise is preventing me from seeing.
I sat with that for a while. Because I have a lot of expertise. And I know what expertise does to a person. It gives you depth, which is a gift. It also gives you patterns, which can be a limitation. You start to see everything through the lens of what you already know. The new information gets filtered through the old frameworks.
That is not always wrong. But it is always incomplete.
Key Takeaways
- An OpenAI model disproved a central conjecture in discrete geometry — demonstrating that frontier AI is capable of original reasoning, not just pattern-matching and production tasks.
- Expertise creates valuable depth but also creates cognitive filters that make some opportunities and threats invisible to the expert.
- The most productive use of AI for entrepreneurs is not replacing execution — it is challenging the assumptions that expertise has made invisible.
- The question “what am I not able to see because I am too close to this?” is one that AI, prompted correctly, can answer with unusual honesty.
- Building a regular practice of AI-assisted strategic inquiry is more valuable long-term than any single AI output.
The Uncomfortable Thing About Being Good at Something
I have spent years learning how to use AI for business. I have built frameworks, trained teams, written about it, spoken about it, and deployed it in my own business in more ways than I can easily count.
That expertise is genuinely useful. It helps me see things other people miss. It helps me make faster, more grounded decisions about AI adoption and strategy.
And here is the uncomfortable thing: it also creates blind spots that other people can see more easily than I can.
This is not a unique problem to me. It is a universal feature of expertise. The deeper you go into any domain, the more automatic your pattern recognition becomes, and the more those automatic patterns filter your perception. You stop questioning things you “already know.” You stop considering possibilities that do not fit your established models.
That is the filter. And the filter is expensive.
Every entrepreneur I work with has it. The longtime real estate investor who cannot see that his underwriting assumptions are based on a market that no longer exists. The marketing expert who cannot see that the channel strategy she has refined for fifteen years is being disrupted by a behavior shift she is too close to notice. The CEO who cannot see that his strongest conviction about his competitive advantage is the thing his best competitor is directly targeting.
Expertise protects you from beginner mistakes. It does not protect you from expert blind spots. Those require a different kind of help.
What the Math Discovery Is Actually Saying
The mathematics result is significant not because it means AI is smarter than mathematicians. It is significant because it demonstrates that AI can reach valid conclusions that human experts — working within their established frameworks — cannot.
The difference is not intelligence. The difference is that AI does not share the expert’s conceptual commitments. It does not have a professional identity built around certain ways of approaching problems. It does not have a career invested in certain beliefs being true. It does not have the cognitive fatigue that comes from working on the same hard problem for years.
None of those human things are bad. But all of them add up to a filter.
When I ask Claude to audit my business strategy, it does not care whether I have been doing something for five years. It does not filter its response through respect for my experience. It looks at what I describe and applies reasoning without the history that makes certain things invisible to me.
That is not a threat to expertise. It is a complement to it. The expert brings the depth and the context. AI brings the filter-free reasoning. Together, that combination can see things that neither could alone.
How I Started Using This in Practice
I have built a quarterly practice that I want to describe honestly, because I think a lot of entrepreneurs are sitting on the same potential without knowing exactly how to access it.
Once a quarter, I give Claude a full description of what my business is doing — the products, the positioning, the audience, the channels, the results. Then I ask it to do something that I have come to think of as the expert audit: challenge the five most foundational assumptions I am making about my market, my competitive position, and my strategy.
Not to give me tips or improvements. Not to validate what is working. To identify what I am assuming is true that might not be.
The first time I did this, I got an answer that made me uncomfortable for three days. Not because it was wrong. Because it was right about something I had been avoiding looking at directly.
That discomfort was worth more than six months of confirmation.
I am not saying AI is always right about these challenges. It is reasoning from the information I give it, and that information is incomplete. Sometimes the challenges are wrong or beside the point. But sometimes — often enough to make the practice valuable — they are pointing at exactly the thing I had been circling without seeing.
That is the application of what happened in mathematics. Not “let AI do the work.” Let AI see what you cannot.
The Practice
If you want to build this into your own rhythm, here is what I would suggest.
Start with one question: “What am I assuming about my business that I have not questioned recently?”
Ask it seriously. Ask AI to help you answer it. Describe your business model, your competitive assumptions, your customer understanding, and your strategic direction honestly. Then ask AI to identify the assumptions underneath those descriptions and challenge the ones that seem most foundational.
Then do the hardest thing: be willing to hear an answer that makes you uncomfortable.
The insight that requires you to reconsider something you built is more valuable than the insight that confirms something you already believed. The mathematics result was valuable because it disproved a conjecture, not because it confirmed one.
Your business holds the same potential. The question is whether you are willing to let AI be honest with you about what it sees when the filter of your expertise is not doing the screening.
Frequently Asked Questions
Is this saying that AI is smarter than human experts?
Not at all. It is saying that AI can reason without certain cognitive filters that expertise creates — and that this filter-free reasoning is a valuable complement to expert knowledge. The best outcomes come from combining expert depth with AI’s absence of conceptual commitments.
How do I get useful answers when I ask AI to challenge my assumptions?
Be specific and honest about the assumptions you are working with. The more clearly you describe your beliefs about your market, your customers, and your strategy, the more targeted and useful the challenge will be. Vague descriptions produce generic challenges. Specific descriptions produce specific insights.
What if AI’s challenges are just wrong?
That happens. Sometimes the challenge will miss the mark because AI is working from incomplete information. The value is not in accepting every challenge but in the discipline of encountering challenges regularly. Even a challenge you ultimately reject can clarify why you believe what you believe and strengthen your thinking.
How often should I do this kind of strategic inquiry?
Quarterly is the rhythm that tends to work for most entrepreneurs. It is frequent enough to catch emerging blind spots and infrequent enough that you have had time to execute on what you learned in the previous session.
Is this different from a regular business strategy review?
Yes. A strategy review typically evaluates performance against goals and adjusts tactics. The kind of inquiry I am describing challenges the goals and assumptions underneath the tactics. It is a level deeper. It is asking not “how are we doing?” but “are we doing the right thing, and how do we know?”
The Close
The math conjecture fell this week. It fell because AI applied reasoning without the filters that kept human experts from seeing the solution.
Your business has the equivalent of unsolved conjectures. Assumptions that feel like facts. Blind spots dressed up as convictions. Risks that look like strengths because you are too close to see them clearly.
That is not a weakness. It is a feature of being human and being good at something.
The invitation is to use AI for what it is uniquely positioned to offer: an honest look at your business without the filter of your experience, your identity, and your professional commitments.
Bring your expertise to the table. Let AI look at it without the bias you cannot help but carry.
What you find in that conversation might be the most valuable output you have ever gotten from a tool.
I believe the best thinking comes from being willing to be wrong. And I believe the best use of any intelligence — human or artificial — is the kind that helps you see more clearly, not the kind that tells you what you want to hear.
About Jonathan Mast: Jonathan Mast is an AI strategist, speaker, and the founder of White Beard Strategies — a community of 75,000+ entrepreneurs learning to build and grow with AI. He is the author of The AI-First Entrepreneur and the creator of the Perfect Prompt Framework. He follows Jesus, loves his family, and believes the best conversations are the ones that leave you thinking for days.
