The value of any advisor, AI vendor, or market-anointed thought leader is not in being trusted outright. It is in whether the board can think more sharply after the engagement than before it, including about the advisor’s own advice.
Why this matters right now
Watching Anthropic and OpenAI pledge restraint on 12 September and then ship major capability jumps within hours of each other ten days later is a useful lesson, and I already wrote about what it means for reading vendor claims and building real governance around them. There is a third lesson underneath both, and it applies as much to advisors as it does to AI labs. Boards that defer entirely to any single voice, a vendor’s safety essay, a celebrity forecaster’s prediction, or an advisor’s recommendation, have quietly handed away the thing a board actually exists to do.
The IBM Watson lesson, told honestly
I founded the IBM Watson AI Lab for Cancer at the Oslo Cancer Cluster during the years when Watson was being presented publicly as a breakthrough in oncology. I was inside that story, building real work with real clinicians, genuinely proud of what our lab produced. I was not aware, at the time, of what was happening in parallel elsewhere in the same company. The Watson for Oncology partnership with MD Anderson was later found, in a University of Texas system audit, to have spent sixty two million dollars against procedural violations before the project was shelved. I learned the fuller picture only years afterward, from the outside, the way most people learn it.
I tell this story not as an achievement but as an admission. Being inside the rise of something does not give you full visibility into its failures happening at the same time, even when you are genuinely close to the work. If that was true for me, sitting inside one of the most closely watched AI programs of its era, it is true for any advisor a board brings in today, including me now. Nobody inside a fast-moving story sees all of it. The only real protection a board has is its own independent capacity to ask hard questions, not the credentials of whoever is speaking.
What unchallenged deference actually costs
CEOs are already asking ChatGPT who to call before they ask their own advisors, and the instinct is understandable. A confident, articulate source is easier to defer to than the slower work of building your own judgment. But deference to a chatbot and deference to a celebrated consultant fail the same way. Neither one can be cross-examined by the board that is relying on it, and neither one is accountable when the recommendation turns out to be wrong.
Unchallenged advice is just dependency with better credentials.
The credentials make the dependency feel like due diligence. A well-known advisor, a famous lab, a highly cited essay, all of these carry the appearance of rigor. None of them substitute for a board that has built its own capacity to test what it is being told.
How I actually work with boards
The methodology I use with clients is built specifically against this failure mode, including the version of it where a board defers to me. It has three parts.
Mapping how the board has historically made decisions. Before any AI question comes up, I want to know how this specific board actually resolves disagreement, who gets deferred to by habit, and where decisions have historically been made by consensus versus by one loud voice in the room.
Building the board’s own AI hype literacy. Most people forming an opinion about AI have never properly used it, which means most boards are evaluating AI claims from headlines rather than hands-on experience. Part of the work is closing that gap directly, so the board can read a vendor’s pacing pledge or a competitor’s AI announcement with real judgment instead of secondhand anxiety.
Socratic training toward independent judgment. Rather than handing the board my conclusions, I work through the questions with them until they reach their own answer, and I test whether that answer holds up when I argue the other side. The most common failure in hiring a strategic AI advisor is settling for one celebrated name that no one else on the board can independently evaluate, which recreates the exact problem the hire was meant to solve.
What this means for your next AI decision
Trust between a board and any outside advisor has to be built, not assumed, and it has to be built in a way that leaves the board stronger rather than more dependent. Before your board accepts the next confident recommendation, whether it comes from a vendor’s essay, a market-anointed expert, or from me, ask the question that actually protects you. Does this advice make our own judgment sharper, or does it just give us a credentialed reason to stop thinking.
The goal of good advice is to make itself less necessary over time, not more. Any advisor who cannot say that plainly is asking for the wrong kind of trust.
If you are leading your organization through this, this is the kind of work I do with a limited number of senior leaders each quarter. → https://anglero.com/the-reset/
Strategic AI Advisor, MerkabaPhi AS, Oslo. 450+ keynotes across 30+ countries.
Listen to the Clarity at the Top podcast on Spotify.
Inquiries: anglero.com