Output Is Not an Effect

For most of the past three years, I misunderstood what artificial intelligence was for in our writing. I treated it as a way to produce more, when our actual requirement was to produce better. Those are different objectives, and they call for different uses of the same tool. For an organization that does not use social media, publishes papers and studies, and posts its essays on Substack, the distinction should have been obvious from the start. It was not, and that was my error.

This is not an argument against artificial intelligence. AI is a good tool, and tools that check structure, catch errors, and test whether an argument holds together have a legitimate place in serious work. It is also not an argument against publishing regularly. The problem was the question I began with. I asked how we could keep content moving, rather than asking what our writing was supposed to accomplish and what it would take to accomplish it. Using AI to improve the work is sound practice; relying on it to produce content is a misuse of it.

Had I started with the right question, the answer would have been clear. Social media rewards presence and frequency, because its requirement is attention. Our requirement is different. At the time, our readers were primarily people working in the women’s game: coaches, sporting directors, analysts, and others making decisions about players, programs, and investment with real consequences for their clubs. They came to our work for information they could use, reasoning they could test, and a method they could apply in their own environment. Since then, our work has extended beyond women’s football, and our readership has extended with it. The audience is broader, but the requirement has not changed. Frequency does nothing for that reader, whoever they are. A piece that arrives on schedule but adds nothing to their understanding is not a contribution; it is a cost to their time and to our credibility.

In practice, I used ChatGPT to check my work and, over time, to help keep the work coming. Some of our earlier white papers passed through it, and so did a number of the essays we published between December 2025 and February 2026. The writing was always mine. The arguments, the methods, and the conclusions came from our work, not from the tool. What the tool did was dilute them. It smoothed, shortened, and dramatized. It inserted one-line paragraphs, slogans, and staccato phrasing that read well at a glance and sounded nothing like me. It added vocabulary I would never use and certainty I would have qualified. Each change was small. Taken together, they produced writing that moved faster and said less.

This is where the distinction between outputs and outcomes matters. Volume is an output. It is easy to count, easy to report, and easy to mistake for progress. More content does not equal better content, and it does not produce a different outcome. Ten pieces carrying the same thinking leave the reader where one piece would have left them, and a decision does not improve because more was written about it. The outcome we actually need is a reader who understands the problem more clearly after reading than before, and who trusts that what we have written reflects our own thinking. When a tool reshapes the phrasing that carries the reasoning, that trust is harder to earn. For a firm whose value rests on the quality of its information and the soundness of its method, that is not a stylistic issue. It is a degradation of the capability itself.

There was a second cost, less visible but just as real. Time spent producing more was time not spent thinking harder. Quality information comes from observation, analysis, and the willingness to separate what we know from what we suspect. None of that is accelerated by generating text. It is accelerated, if at all, by having fewer things to say and saying them properly.

So we are changing how we work. We will publish when we have something worth reading, not because a calendar says it is time. The writing has always been mine, and it will now read that way, in my own words and in my own structure, without dilution. Tools will be used to check the work, and any suggested change will be one I accept or reject deliberately rather than one inserted on my behalf. We will judge each piece by whether it improves a reader’s understanding or decision, not by how recently the last one appeared.

The lesson is the same one we apply to every organization we work with. Understand what you are trying to accomplish before you decide what resources you need. I reached for a tool that increased output before I had confirmed that output was the requirement. It was not. Our requirement is quality information, clearly reasoned and honestly presented. AI can help us check that work, but it cannot substitute for it, and producing more of it was never the point.

Disclaimer

This publication provides independent, apolitical analysis of historical developments, current events, geopolitics, national security, and strategic challenges. We maintain strict political neutrality, aligning with neither the left nor the right. Examining or criticizing political actors, institutions, or policies does not constitute political advocacy, opposition, or endorsement.

Our assessments are grounded in historical facts, available evidence, established analytical methodologies, and evaluations of probability, consequence, and uncertainty. Hypothetical scenarios are illustrative and do not imply actual or planned events, intentions, or operations.

Our purpose is to identify, define, and understand problems—not to prescribe solutions or advocate outcomes. Conclusions reflect the available evidence, not political preferences. Decisions and actions remain the responsibility of others.