Analysis is cheap. Judgement is the advantage.

For much of the history of management, analysis has been expensive.

Answering an important strategic question meant assembling data, commissioning research, building a model, testing assumptions and producing a paper. The cost imposed a natural constraint. Leadership teams could ask for more analysis, but doing so consumed time and resources. Eventually, someone had to decide.

AI is removing that constraint.

A leadership team can now generate another scenario, interrogate another assumption, construct another argument or ask for another assessment at negligible marginal cost. Questions that once required days of analytical work can increasingly be explored in minutes.

That is clearly useful. But it also changes the economics of executive decision making. When analysis becomes abundant, the temptation to keep analysing becomes much harder to resist.

And the capability that limits the quality of the decision is no longer analysis. It is judgement.

When working the problem harder becomes almost free

We have argued elsewhere that leadership teams respond predictably when faced with a difficult decision: more analysis, a broader group, another option, a deferral until more information arrives.

But there is a point beyond which additional analysis produces diminishing returns. Until recently, there was at least some friction in continuing to work the problem harder. Another analysis had to be commissioned. Someone had to produce it.

But imagine a leadership team today considering whether to withdraw from a commercially attractive market because of concerns about the conduct of the government. Generative AI can help analyse the financial consequences of withdrawal, identify relevant legal obligations, map stakeholder positions, develop scenarios for competitor behaviour, summarise historical precedents and construct arguments for and against different courses of action.

Having done all of that, the team can ask it to do more.

What have we missed? Challenge our assumptions. Make the strongest case for staying. Now make the strongest case for leaving. Add a five-year horizon. Weight the reputational risks differently. What would an investor say? What would an employee say?

Each iteration may add something. But at some point, the remaining problem is no longer analytical. The team has to decide how much commercial value it is prepared to sacrifice, what responsibilities it accepts, which risks it is willing to carry and what its values mean in this particular set of circumstances.

Abundance changes the problem

This matters because scarcity has always played an underappreciated role in decision-making.

When information was costly, leaders had to decide when they knew enough. Imperfectly, the cost of further analysis created a stopping rule. AI weakens that rule.

The diminishing-returns threshold has not disappeared. But one of the signals that told us we had crossed it — the cost of continuing — is disappearing.

This creates a paradox. Organizations may become substantially better at analysing decisions while becoming no better at making them. There is already some evidence of the gap. In Board Intelligence's 2026 Board Value Index, which surveyed more than 400 non-executive directors, chief executives and finance directors, 86% said that rigid or inconsistent decision-making frameworks had contributed to delayed, rushed or poor decisions in the previous six months.

Abundant analysis can also make the weakness harder to see. A request for more analysis looks like diligence, but it is often not an attempt to understand the problem better. It is an attempt to defer the moment at which judgement has to be exercised.

The source of advantage is shifting

Superior analytical capability was long a source of advantage. But as AI makes many forms of analysis widely accessible, the advantage moves downstream. The question becomes what a leadership team can do with it.

·       Can leaders distinguish between additional information that changes the decision and analysis that merely adds detail?

·       Can people challenge an apparently authoritative analysis?

·       Can the team recognise considerations that are important precisely because they cannot readily be quantified?

·       Can the team determine what its values require when several defensible courses of action remain?

·       And can it eventually stop analysing and choose?

These are questions of collective judgement.

That is why AI increases, rather than diminishes, the importance of decision capital: the stock of clarity, trust, and judgement that enables a leadership team to make confident, values-driven decisions together.

Start with clarity. AI is exceptionally good at expanding the analytical surface of a problem: more scenarios, more considerations, more framings, all at very low cost. But more ways of looking at a decision do not produce greater clarity about the intent of the decision itself: what are we solving for? Without that clarity, analytical abundance becomes analytical sprawl.

Then trust. AI-generated analysis arrives with a different kind of authority: comprehensive, polished and apparently neutral. But it is not neutral: choices have been made — implicitly and explicitly — about the question posed, the information provided, and the weight given to different considerations. And people who would hesitate to challenge a senior colleague may hesitate just as much to challenge an analysis presented as the product of a powerful technical system. The question is not whether people trust AI, but whether they trust one another enough to question how it is being used.

Finally, judgement. Used well, AI improves the material on which judgement operates. But no neutral weighting resolves a choice between financial return and employee trust, or between the benefits of innovation and harms whose probability is uncertain. Those choices turn on significance, responsibility, and acceptable trade-offs. AI can help leaders see the choice more clearly. It cannot make the choice cease to be a choice.

Collective judgement becomes the competitive advantage

None of this is an argument for using less AI in decision making. If analysis becomes cheaper and more widely available, organizations should exploit it. But collective judgement, unlike access to technology, cannot be acquired instantly.

It is built through repeated experience in learning how to surface disagreement, understanding how colleagues reason, and examining past decisions. That is decision capital.

AI can contribute to its accumulation. It can act as a challenger, exposing missing perspectives and testing the consistency of reasoning. A team that has never surfaced disagreement effectively may find that a model does it more reliably than its own members.

But a model challenges at no cost to itself. It has no standing to lose and no relationship to protect, which is what makes its challenge easy to accept and equally easy to ignore. Trust is what makes people willing to take that risk with one another, and it is that willingness which carries over to the most complex decisions. A leadership team can be extensively challenged and still never have learned to disagree. AI contributes to erosion too, whenever analysis is used to avoid disagreement, obscure responsibility, or defer a difficult choice.

The technology is therefore not outside the question of decision capital. It places an existing capability under a new load. Boards will rightly set rules for AI's use and determine where human oversight is required, but there is another set of questions worth asking.

·       Where has AI materially improved the quality of our decisions, rather than simply increasing the volume of analysis available to us?

·       How do we know when further analysis is likely to change a decision?

·       Who challenges the assumptions behind AI-generated analysis?

·       Which considerations in our most important decisions cannot sensibly be reduced to a metric?

·       When reasonable people still disagree after examining the evidence, how do we decide what should carry weight?

·       And are we using AI to strengthen our collective judgement — or to postpone exercising it?

The competitive question is unlikely to be who can produce the most comprehensive analysis. Increasingly, everyone will be able to do that.

The harder advantage to replicate will be a leadership team capable of knowing what matters, disagreeing productively, and reaching a decision when the analysis runs out of answers.

Ithaka works with boards and executive teams on the quality of their collective decision-making, examining and strengthening how consequential decisions are made, and how trust, clarity and judgement are being built over time. If your organization is investing in analytical capability faster than the capacity to use it, get in touch.