Research
The Machine's Version
How AI Describes Philippine Business Leaders
The Machine’s Version measured what AI tools say about a business leader when someone checks that leader before a deal. Over two weeks in mid-August 2026, I asked three AI tools six questions a counterparty would ask about 40 Philippine business leaders. The tools were ChatGPT, run with and without web search, Google Gemini, and Perplexity. I split the leaders into four tiers of ten, ranked by how much public information exists about each one. I scored 894 answers on accuracy, currency, definition, source quality, and tone. Each answer was scored against a ground-truth file I built and verified for every subject before fieldwork began.
Seventy-three percent of the failures came from the public record. The tools repeated what the record said, and the record was wrong, stale, or incomplete. Prominence did not buy accuracy. Scores for the searching tools held between 217 and 224 of 270 in every tier. What changed with prominence was the kind of failure. The least prominent leaders got silence and invented detail. The most prominent got stale titles and weak citations. Almost all invention came from ChatGPT without search. Web search cut invention tenfold, but 13.5 percent of searched answers relied on weak sources, and three quarters of those came from one tool.
No subject was contacted. The report names no subject, and subject-level findings are sealed.
Download the report (PDF, 1.01 MB)
How to cite
Grgurovic, A. (2026). The Machine's Version: How AI Describes Philippine Business Leaders. Dalmatian Consulting. https://dalmatian-consulting.io/version/report/2026
Private briefings on the findings are available to family offices and principals.