INSIGHTS / ANALYSIS
AI is rapidly transforming the production of strategic analysis. But the ability to process information and generate sophisticated assessments is not the same as understanding what that information means. This analysis examines the enduring role of experience, context and human judgment in an increasingly AI-driven analytical environment.
Artificial intelligence is changing the production of analysis.
These Systems capable of processing vast quantities of information, identifying patterns, summarising complex material and generating structured assessments can now perform tasks that once required substantial amounts of human time. A competent analyst can use AI to research faster, test alternative hypotheses, identify contradictions and produce a first analytical framework in a fraction of the time previously required.
This development should not be underestimated.
But it also raises a more fundamental question: if artificial intelligence can produce sophisticated analysis, what remains the distinctive value of the human analyst?
The answer is not simply experience, Nor is it an argument that humans are inherently more intelligent than machines. The more important distinction concerns judgment.
AI can increasingly generate analysis. The experienced analyst must determine whether that analysis actually makes sense.
Much of what is described as analysis is, in practice, information processing.
Information is collected, organised, compared and summarised. Political statements are examined alongside economic indicators, historical developments and security trends. Frameworks like PESTLE, SWOT, DIME or PMESII can then be applied to structure the findings.
Artificial intelligence is exceptionally capable in this environment, because
It can process more information than an human analyst, work across multiple sources simultaneously and rapidly generate competing interpretations. It can identify relationships that may not be immediately apparent and expose gaps in an initial assessment.
These capabilities represent a substantial improvement in analytical efficiency.
But efficiency is not synonymous with accuracy.
A report can contain extensive information, sophisticated terminology and a logically structured argument while still reaching the wrong conclusion.
The problem begins when analytical sophistication is mistaken for analytical validity.
One of the more important risks associated with AI-generated analysis is not that it produces obviously poor work, but is that it can produce plausible work that is wrong.
An AI system can construct a convincing explanation of political instability. It can identify institutional fragmentation, economic pressures, elite competition, public dissatisfaction and external influence. Each individual observation may be correct, Yet the central interpretation can still be mistaken.
Perhaps the apparent political crisis is largely performative.
Perhaps an economic decision that appears commercially motivated is actually intended to influence a foreign government.
Perhaps a government's public position is designed for domestic consumption rather than international signalling.
Perhaps the actor that appears most influential is not the actor making the important decisions.
These distinctions matter because geopolitical events rarely have a single observable meaning.
The analytical challenge is therefore not simply to identify what is happening. It is to establish why it is happening, what is driving it and which interpretation best explains the available evidence.
That requires judgment.
Strategic analysis operates within environments where formal structures and actual behaviour do not always correspond.
A government may possess extensive formal authority but limited practical capacity. An institution may appear central to a decision-making process while having little influence over the outcome. A political leader may publicly support a particular policy while privately attempting to constrain it.
Political language is particularly difficult to interpret without context.
A statement expressing support for “constructive dialogue” may represent genuine cooperation. It may also be diplomatic damage control, an attempt to buy time, or a signal directed toward another actor entirely.
The words themselves are observable.
Their strategic meaning is not.
This is where accumulated experience becomes relevant.
An experienced analyst develops an understanding of how institutions, governments and political actors behave over time. This does not eliminate analytical bias—experience can create biases of its own—but it provides a deeper reference framework against which new developments can be evaluated.
The analyst does not simply ask:
What does the information say?
The more important question is:
Does this interpretation correspond with how these actors have historically behaved?
Not all knowledge is explicit.
Some knowledge exists in legislation, statistics, official documents, academic literature and databases. It can be collected, searched and processed.
Other knowledge is tacit.
It includes an understanding of institutional culture, political incentives, informal power structures, bureaucratic behaviour and the difference between what an institution is formally supposed to do and what it actually does.
This distinction is particularly important in geopolitical analysis.
Two analysts can examine exactly the same information and reach different conclusions because they interpret the surrounding context differently.
One may see a new policy as an isolated development.
Another may recognise it as part of a longer sequence of decisions.
One may focus on what a government says.
Another may focus on what it has consistently done.
One may interpret an announcement as the beginning of a policy.
Another may recognise it as the culmination of a process that began months or years earlier.
The difference is not necessarily access to information.
It is contextual understanding.
There is an important qualification.
Experience alone does not produce good analysis.
An experienced analyst can be wrong. Experience can generate confirmation bias, excessive confidence and an attachment to established interpretations. A person who has correctly anticipated events in the past may become overly confident in their ability to predict the future.
AI, in some circumstances, can actually help mitigate these weaknesses.
It can challenge an analyst's assumptions, generate alternative hypotheses, search for contradictory evidence and construct scenarios that the analyst may initially have overlooked.
This is why the most productive relationship between AI and strategic analysis is not necessarily adversarial.
It is complementary.
The question is not:
AI or analyst?
It is:
What can an analyst accomplish with AI that an analyst working without it cannot?
The strongest analyst of the future may not be the person who competes with artificial intelligence.
It may be the person who knows how to use it without surrendering judgment to it.
AI can perform much of the analytical preparation:
The analyst then performs the functions that remain fundamentally strategic:
This changes the role of the analyst.
The analyst becomes less of an information processor and more of an interpreter, challenger and adviser.
There is another consequence of widespread AI adoption that deserves attention.
If organisations increasingly rely on similar AI systems, similar datasets and similar analytical prompts, their assessments may begin to converge.
This could create a form of analytical homogenisation.
Different organisations may independently produce reports containing similar assumptions, terminology, scenarios and conclusions. The resulting consensus may appear reassuring because multiple assessments seem to confirm one another.
But consensus does not necessarily mean accuracy.
If the same underlying assumptions are reproduced across multiple analytical products, apparent agreement may simply represent the replication of a common analytical model.
Independent analysis therefore retains value precisely because it can challenge the prevailing interpretation.
A good analyst should not ask only:
What is the consensus?
The analyst should also ask:
What if the consensus is wrong?
And more importantly:
What evidence would demonstrate that it is wrong?
AI is particularly effective at constructing internally coherent scenarios.
But geopolitical reality is rarely internally coherent.
States pursue contradictory objectives. Governments make mistakes. Bureaucracies protect institutional interests. Political leaders respond to domestic pressures. Economic actors react emotionally as well as rationally. Historical grievances influence contemporary decisions. Actors sometimes choose options that appear irrational from the outside because those options are politically necessary from within.
Consequently, the theoretically optimal course of action is not necessarily the most probable.
Strategic analysis must therefore account not only for what an actor could do, but what that actor is capable of doing, incentivised to do and politically able to do.
That distinction can fundamentally alter a forecast.
An AI system may construct an elegant scenario.
The analyst must ask:
Would the actors involved actually behave this way?
The ultimate purpose of strategic analysis is not to produce a sophisticated document.
It is to improve a decision.
There is a hierarchy here:
What is happening?
Why is it happening?
What is likely to happen next?
What are the implications?
What should the decision-maker do?
The first questions are increasingly accessible to artificial intelligence.
The final question is considerably more difficult.
A theoretically optimal recommendation may be impossible to implement. Political constraints, institutional capacity, legal considerations, reputational exposure, financial resources and the client's own risk tolerance can all alter what constitutes a viable course of action.
The best recommendation is therefore not necessarily the one that looks best on paper.
It is the one that can survive contact with reality.
Artificial intelligence will inevitably change the economics of analytical work.
The production of conventional research, summaries and basic assessments will become faster and cheaper. Some forms of analysis that once required significant human effort will increasingly become automated.
That is not necessarily a threat to serious analysts.
It may instead expose a distinction that was already present but often overlooked.
If the principal value of an analyst is the ability to collect information and produce a report, AI represents a significant disruption.
If the principal value is the ability to interpret uncertainty, challenge assumptions, understand political behaviour and provide realistic strategic judgment, the role remains considerably more difficult to automate.
In this environment, access to information becomes less scarce.
Judgment becomes more valuable.
The future competitive advantage may therefore belong neither to humans who reject AI nor to organisations that attempt to replace analysts entirely with machines.
It will belong to those capable of combining technological capability with human judgment.
Artificial intelligence can already produce analysis of a surprisingly high standard. Its capabilities will continue to expand.
The appropriate response is not to dismiss those capabilities, Nor is it to assume that increasingly sophisticated machine-generated analysis makes human analysts redundant.
The central issue is the difference between producing an analytical product and exercising analytical judgment.
AI can process information at extraordinary speed.
It can identify patterns, construct scenarios and challenge assumptions.
But strategic analysis ultimately depends on understanding context, causality, incentives, behaviour and uncertainty—and on recognising when an apparently convincing explanation does not correspond with reality.
The analyst's task is therefore changing.
It is no longer sufficient to know more information than everyone else.
The valuable analyst is the one who can determine which information matters, what it means, what it does not mean, what may happen next, and what a decision-maker should realistically do about it.
AI may make analysis faster.
It may make it more accessible.
It may even make mediocre analysis much easier to produce.
But in an environment saturated with machine-generated information and increasingly sophisticated automated assessments, the ability to exercise independent judgment may become more—not less—important.
**AI can generate the analysis.
The analyst must determine whether it is worth believing.**
