Every list of "top AI leaders" is really a popularity contest — follower counts, funding rounds, magazine covers. We wanted the opposite: a ranking of founders and CEOs on a verifiable record in the actual core of AI, where every point of every score can be re-checked from its sources. That's the Global CEO AI Expertise Index.

The rule that shapes everything else: popularity counts for nothing. Being famous for AI is not the same as having built it. The index measures the second thing.

What the index measures

Each person is scored on a verifiable record across the parts of AI that are hard to fake:

• The mathematics — genuine depth in the math that underpins modern models, not a passing acquaintance with the vocabulary.
• Vector embeddings — work on the representations at the heart of how models encode meaning.
• The transformer / language-model lineage — whether their own work sits in the line that leads to today's architectures.
• Whether their work underpins frontier models — does something they built actually hold up the current generation of systems?
• Years in language modeling — time spent in the field, not time spent talking about it.
• Breadth across types of language modeling — biological, financial, and code, not just one narrow lane.
• Years and companies as a scientific and technical founder — a track record of building, repeatedly.

How the score works

Research is weighted 70%. The bulk of a person's score comes from a documented research and technical record — papers, systems, and contributions that other people can point to — rather than from title, company valuation, or profile. The remaining weight covers founding and building history.

The people ranked are drawn from both crypto and general industry, so a crypto founder with real language-modeling depth and a mainstream AI scientist-founder are measured on the same yardstick.

Why "re-checkable" is the whole point

The feature we care about most is that every score is traceable to its sources. A ranking you have to take on faith is just an opinion with a number attached. This one is built so that if you disagree with where someone sits, you can open the evidence behind their score and check it yourself. That is the difference between a leaderboard and a marketing list.

It fits how we build everything at CyMetica: claims should be verifiable, not asserted. The same instinct that puts on-chain settlement behind our markets puts sources behind every rank on this index.