AI Is Making Knowledge Cheaper. Judgment Is Becoming More Valuable.

Welcome to Issue #8 of Skills to Solo Income. Each week, I help employed professionals turn what they already know into a practical path to independent income, without burning the career they're still in.

Twenty years ago, difficult answers were expensive. If you wanted to understand a technical subject properly, you bought the book, enrolled in the course, found someone who had done it before, or spent enough years inside the profession to learn what couldn’t be easily written down. Expertise was partly valuable because access to the knowledge behind it was scarce.

That scarcity is disappearing remarkably quickly.

Ask an AI tool for a marketing plan, and it will give you one. Ask it to build a financial model, draft an SOP, explain a legal concept, design a training programme, compare business strategies, or write a piece of code, and it can usually produce something plausible enough to get you moving. You can have ten options before you have finished your morning coffee.

This is an extraordinary expansion of access to knowledge. It is also changing where professional value sits. Because once everyone can get an answer, having the answer is no longer quite the advantage it used to be.

The valuable question becomes: Which answer actually matters here? And that is a different skill altogether.

For most of my working life, professional development followed a fairly simple logic. Learn more, gain experience, accumulate qualifications, become more knowledgeable than the person who entered the profession yesterday.

That still matters. But AI has broken something important in the economics of that model.

A beginner can now sit down with a tool and produce work that, on the surface, looks surprisingly competent. They can ask for a market analysis without knowing much about marketing. They can generate a cash-flow forecast without being particularly good at finance. They can ask for a launch plan, a sales script, or a competitor analysis without having spent years learning the underlying disciplines.

The information gap between beginner and expert has narrowed. The judgment gap has not.

Suppose AI gives you five ways to improve sales. One recommends lowering the price. Another recommends increasing advertising. A third suggests adding features. A fourth proposes changing the positioning. A fifth suggests targeting a different customer.

All five could be sensible. None of them matters if the actual problem is that hardly anyone knows the product exists.

The difficulty was never generating another possible answer. The difficulty was identifying the variable that mattered enough to act on.

That distinction becomes more important as AI becomes better, because better AI gives us more plausible options, not fewer. A weak tool gives you one bad answer. A strong tool might give you ten good ones.

I spent a large part of my career teaching professional accountancy students. The interesting thing, looking back, is that almost none of the information they needed was secret.

The syllabus was available. The textbooks were available. Past examination papers were available. Examiner reports and technical material were available. Students could buy all the knowledge they needed.

Yet access to all that information did not automatically translate into examination performance.

A student could spend three hours studying something that was technically important but strategically low-value. Another could understand a concept perfectly and still misread what an examination question was asking. Someone else could work harder than almost everyone in the class and still allocate their limited study time badly.

So I stopped treating teaching as an exercise in covering material uniformly. I studied historical papers. I looked for patterns in what was examined. I paid attention to examiner comments and the rationale behind questions. I concentrated more teaching effort where I expected it to have the greatest effect on the actual objective: performance in the examination.

That difference now seems much bigger than an education problem. It is the same problem AI is placing in front of every professional.

There is a phrase people use about experienced professionals: They've seen this before. That sounds almost trivial, but it explains a great deal.

Give the same report to two people. One sees information. The other notices that the sales increase came entirely from one customer.

Give them the same project plan. One sees a schedule. The other notices that everything depends on an approval that routinely arrives three weeks late.

Give them the same AI-generated business strategy. One sees five convincing recommendations. The other immediately rejects four because they depend on assumptions that do not survive contact with the real world.

That is what useful experience often becomes: a collection of compressed decision rules.

You know what deserves a second look. You know which number is probably hiding the problem. You know when someone is answering the question that was asked rather than the question that actually matters. You know which apparent crisis is routine noise and which small anomaly is worth investigating immediately.

None of this means experienced people are automatically right. Experience can fossilise bad habits just as easily as it can improve judgment. Someone can spend twenty years repeating the same mistake and become extremely experienced at making it.

The advantage appears when experience has been examined, tested, and converted into better decisions.

AI can help answer every one of them. It still cannot take responsibility for the decision.

This creates an uncomfortable implication for anyone whose value has largely come from knowing things other people do not know. Some of that value is being commoditised.

The junior employee can now ask AI how to construct the spreadsheet formula. The manager can ask it to explain a financial ratio. The entrepreneur can ask it for a first draft of the marketing plan. The programmer can ask it to generate routine code.

The knowledge has not become worthless. The cost of accessing it has fallen.

So the more interesting question is where a professional should move next. One answer is upstream.

This is not an argument that everyone must become a strategist or stop doing technical work. Someone still has to understand accounting before exercising sound accounting judgment. Someone still needs to understand software before deciding whether an AI-generated architecture makes sense.

But the centre of gravity changes.

That is a much harder capability to commoditise.

This is also why I find many conversations about AI replacing professionals slightly misplaced.

The question is often framed as a contest: AI versus accountant. AI versus consultant. AI versus marketer. AI versus teacher.

I think the more useful comparison is between two professionals: one who uses AI simply to produce more information, and another who uses it to improve the quality and speed of decisions.

The first person asks AI for twenty ideas.

The second asks it for twenty ideas, challenges the assumptions underneath them, compares the options, identifies what evidence is missing, and decides which cheap experiment could tell them something useful.

This is why my own view of AI is neither particularly utopian nor particularly defensive. AI has dramatically reduced the cost of research, explanation, analysis and technical assistance. That gives ordinary people access to capabilities that once required considerably more money, education or specialist help.

But cheaper assistance does not remove the need for judgment, execution or responsibility. If anything, it increases it.

When information was scarce, finding the answer was difficult. When information becomes abundant, deciding what deserves attention becomes difficult.

Most professionals do not describe themselves as having judgment. They describe what they know.

“I know finance.” “I know procurement.” “I know project management.” “I know HR.”

But if you have been doing something for long enough, the more valuable asset may be hiding one level below the subject matter.

Ask yourself what people come to you for when the answer is not obvious.

Perhaps they ask you to look at the numbers because you can usually see where something does not add up. Perhaps you are the person who can tell whether a customer's complaint is routine frustration or the beginning of a serious account problem. Perhaps you can look at a project and immediately see the dependency everyone else has ignored. Perhaps you know which piece of a complicated process deserves attention first.

Those are not merely things you know.

AI makes that question more urgent, not less.

Knowledge is not disappearing. We are simply moving from a world where acquiring it was difficult to one where filtering it may become the harder problem.

That changes what I would invest in.

I would still learn the fundamentals of my field. Without them, judgment has nothing solid to stand on. But I would pay much more attention to the invisible layer underneath my work.

Those are not questions AI makes obsolete. They are questions AI makes more valuable.

Because the machine can increasingly supply the possibilities. Someone still has to decide which possibility deserves to become reality.