What the Balogun Red Card Can Teach Us About AI and Judgment
On the night of July 1, in a World Cup Round of 32 match in Santa Clara, the United States men’s national team lived through two very different kinds of decisions within half an hour. You probably already know one of them, but the difference between that moment and another from earlier in the game is one of the most useful things leaders could learn about artificial intelligence this year.
Early in the first half, U.S. striker Folarin Balogun seemed to have scored. He hadn’t: Semi-automated offside technology showed he was fractions of a second ahead of the last defender, and the goal was chalked off. Nobody argued. Nobody demanded the referee weigh context, intent, or the run of play. A question that used to depend on a linesman’s eye had been answered by a combination of sensors, cameras, and AI.
Half an hour later, Balogun’s trailing boot scraped down the leg and onto the ankle of Bosnian defender Tarik Muharemović. A Video Assisted Referee, or VAR, system flagged the contact. Referee Raphael Claus walked to the sideline monitor, watched a slow-motion replay, and sent Balogun off for serious foul play. Even after the U.S. finished the match with 10 men and won anyway, the decision became one of the tournament’s most argued-over moments.
Put aside for a moment whether the call was correct, and also ignore the unexpectedly political aftermath. The case highlights an important point: In the technology-rich environment that soccer has become, the expertise and decision-making skills of the human referee became more, not less, important.
One call was instant and uncontested. The other led to a debate over days that involved world leaders.
The difference had nothing to do with how much data was available, and everything to do with what kind of question each call actually was. That distinction is important to anyone thinking about the role of decision-making in the age of AI.
One class of decisions is straightforward and measurable, often binary, and no judgment is needed. Offside is a line-crossing problem, for example, while goal-line technology is one of physics. Preserving a ceremonial human role in this kind of call usually adds nothing but delay and the possibility of error.
The business world is full of similarly routine decisions, which can become automated workflows: Routing an inbound request to the right queue, serving the best ad for a given user, flagging a transaction that matches every marker of fraud. These decisions can be handed to a machine without nostalgia. Humans in the loop generally do not supply wisdom here; they merely add friction.
Indeed, one survey of C-Suite executives last year found that 44% would override a decision they had already planned to make based on AI insights.
But whether AI impacts a decision – for a referee or for a C-Suite executive – should really depend on what the 44% decisions are. Are they an offside on goal-line decisions? That should be a no-brainer. Are they foul decisions? That’s a more nuanced case with different consequences for the human decision maker.
The decisions that matter, increasingly, are the kind that require judgment and experience to interpret facts.
Leaders assume that more data shrinks the territory where human judgment is needed. In actuality, it does the opposite: It shrinks the territory of easy calls and subjects the hard ones to scrutiny.
Whether a challenge was reckless or unlucky, whether force was proportionate, whether intent should count at all — no additional data resolves these questions. More data can only make the situation murkier, as everyone stares at masses of evidence while disagreeing about what it means.
The calls that survive automation are, by definition, the ones that automation could not and should not settle. Making these calls requires human experience, ability, and decision-making skills.
So the charge for leaders is not to keep humans hovering over every decision AI touches. Instead, those in charge have two key responsibilities:
- Identify the decisions that are truly settled once the data arrives; and then
- Accept that the remaining decisions will be more contested, more visible, and more dependent on judgment than ever.
In the age of AI, decision-making becomes more, not less, important, because everything left to decide is precisely the part that no dataset can decide for you. Those judgment calls cannot be delegated to a machine. They are concentrated in a leader.
Business leaders and referees alike need to balance the cognitive friction between human task evaluation and analytic precision. They must establish a trust model to avoid automation bias (over-reliance on systems) and algorithm aversion (under-reliance on AI). Striking the balance between intuition and information requires intentionally designing the collaboration protocol around who is responsible for making the decision and its consequences.
At the 2026 World Cup, the referees’ job hasn’t gotten easier as the technology around them has improved. It’s gotten narrower and, on the calls still left to them, requires considerably more expertise, judgment and decision-making skill. Leaders adopting AI should expect the same trade-off.
The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.
Oded Netzer, Christopher Frank, and Paul Magnone are business professors at Columbia Business School. They are co-authors of Decisions Over Decimals: Striking the Balance between Intuition and Information.
