The Most Valuable Paper I Took at University
By Sarah Williams
I rarely think about most of the papers I took during my undergraduate degree. Some are no longer relevant, and others simply don't apply to the world I work in today. The paper on strategy and change probably gave me some of the foundations, and certainly some of the confidence, for the work I do now. I can also tell you far more than you ever wanted to know about the international banking system, and somewhere in the recesses of my brain sits the knowledge required to optimise an investment portfolio. Useful at the time. Less so in my day-to-day life now.
But there is one paper I'm forever grateful I chose, because I've used what it taught me almost every day since: critical analysis.
We were given articles, research, and other pieces of writing and asked to pull them apart. What perspective was the author taking, and why might they have taken it? What assumptions sat underneath the argument? What evidence supported it? What was fact, what was interpretation, and what was opinion? And ultimately, what was actually relevant and useful from the information in front of us?
Itdidn'tfeel revolutionary at the time. Nearly three decades later, I think it might be one of the most important capabilities a leader can develop.
AI raises the stakes for critical thinking
AI and large language models are incredible tools. They can synthesise enormous amounts of information, identify patterns, challenge our thinking and compress hours of work into minutes. But the quality of an answer still depends on the information available to the model, the context we've given it, the assumptions it makes and, increasingly, the sources it can access.
That last point deserves more attention than it gets. A 2025 study looking at access for AI crawlers found that 60% of the reputable news sites it examined blocked at least one AI crawler. Among misinformation sites, that number was just 9.1%. Sit with the implications of that for a moment. Some of the most reputable sources of information are actively restricting AI access while significantly more questionable sources remain wide open.
AI remains enormously useful. The caution is simpler than that: an incredibly articulate answer is not automatically an accurate one.
There'sanother layer too. When we ask AI a question, we rarely give it everything it needs, so it fills the gaps. It infers our intent and makes assumptions based on the patterns it has learned. Sometimes those assumptions are exactly right. Sometimes they subtly skew the entire answer, and unless we're paying attention, we may never notice.
The real risk is unquestioning acceptance
This is where that university paper has become more relevant to me than ever. When AI gives me an answer, "does this sound good?" is the wrong test. The questions that matter are the same ones I learned in that classroom:
What assumptions has it made?
What perspective is this coming from?
What information might be missing?
What is evidence, and what is interpretation?
Where did this information come from?
Does it actually make sense in the context I'm working in?
That last question carries the most commercial weight, because AI can give you an answer that is perfectly logical and completely inappropriate for your situation. And as the technology improves, those answers become harder to catch. The writing is polished, the reasoning sounds plausible, and the confidence is convincing, so we stop questioning.
Then we create more AI slop
We already have a name for the result. Generic articles, recycled opinions, poorly researched claims, content that looks credible but adds very little. And I worry about what happens as more of it gets published, because AI-generated material becomes part of the information environment that future systems learn from. Poor information gets repeated, repetition lends it the appearance of credibility, and gradually it becomes harder to tell where the original evidence ended and the recycling of someone else's interpretation began.
Which brings me back to that classroom. Knowing how to use AI is fast becoming table stakes. The capability that will separate leaders and organisations is knowing how to think alongside it: to question it, challenge it, check it, recognise where assumptions have been made, ask for sources when the evidence matters, and decide, as the human in the lead, what deserves to be carried forward.
AI can help us analyse an extraordinary amount of information, but we still have a job to do. We decide what is fac, what is opinion, what is relevant, what is questionable, and what we are prepared to put our own name behind.
Nearly three decades on, critical analysis might just turn out to be the most valuable paper of my entire degree.
