“What if the greatest threat to a board is not disagreement, but unanimous agreement? What if management reaches a conclusion, AI independently validates it, every dashboard turns green, and every presentation reinforces the same message? Should directors feel reassured—or should this be the exact moment they become more sceptical?”
I have noticed a subtle but important shift in how strategic decisions are made. Earlier, when analysts presented a strategic report, management would challenge almost every number, question the assumptions behind the forecasts, and debate the conclusions. That scrutiny forced analysts to defend their thinking, improve their analysis, and uncover weak assumptions before decisions were made.
Today, AI has changed that dynamic. Instead of treating analysis as something to be challenged, many managers increasingly use AI to validate what they already believe. When both management and AI arrive at the same conclusion, there is often greater confidence—but not necessarily greater accuracy. Agreement can create the illusion that the analysis has been thoroughly tested, when in reality the same assumptions may simply have been reinforced by both.
This is where I believe the board’s role has become even more important. If management is no longer questioning assumptions with the same intensity as before, the responsibility shifts to the board. Directors should not ask whether AI agrees with management. They should ask whether both AI and management might be relying on the same flawed assumptions. In an age of AI-assisted decision-making, the board’s greatest value may no longer be approving recommendations, but challenging the thinking behind them.
When we challenge underlying assumptions, we may avoid expensive fall!
Silicon Valley Bank is perhaps one of the clearest examples. Before its collapse, the bank appeared financially healthy. Deposits had grown rapidly, key financial indicators looked reassuring and many quantitative models suggested limited immediate stress. Neither management nor traditional risk models anticipated the speed at which events would unfold. The problem was not faulty mathematics.
The problem was that historical relationships no longer applied. Rising interest rates significantly reduced the value of the bank’s investment portfolio. At the same time, a highly concentrated base of uninsured depositors withdrew billions of dollars within hours, amplified by social media and digital banking. Historical models had very little experience with bank runs occurring at digital speed. The assumptions embedded within the models failed because customer behaviour had fundamentally changed.
Commercial real estate provides another valuable lesson. For years, valuation models assumed relatively stable office occupancy, predictable refinancing conditions and consistent demand for commercial office space. AI trained on decades of historical market data would almost certainly have reached conclusions similar to those of many management teams.
Then hybrid work permanently altered the economics of office buildings.
The issue was not poor analytics. The issue was that both humans and machines were analysing yesterday’s world while the market had already entered a new one. Structural change invalidated the assumptions on which the entire analysis had been built. This is where the role of the board becomes even more important in the age of AI.
Directors should not attempt to compete with AI’s analytical capability. Machines will always process larger volumes of data faster than humans. Instead, boards must contribute something AI cannot easily replicate—critical thinking, independent judgment, practical experience and ethical reasoning. Directors should deliberately search for blind spots rather than confirmation. They should ask whether historical relationships still exist, whether structural shifts are underway, and what assumptions underpin every recommendation placed before them.
Perhaps the most valuable question a director can ask is also the simplest: “What if we are all wrong?”
That question forces management to defend its assumptions rather than merely present its conclusions. It encourages alternative scenarios, stress testing and discussions around uncertainty instead of certainty. Most importantly, it prevents boards from confusing consensus with truth.
As AI becomes deeply embedded in corporate decision-making, the responsibility of directors will not diminish. If anything, it will increase. Boards will create value not by accepting AI’s recommendations, but by challenging the assumptions behind them. The future belongs to directors who understand that AI excels at identifying patterns from the past, while boards are responsible for preparing organisations for futures that history has never experienced.