Artificial intelligence can write computer code, analyze spreadsheets, summarize documents, create presentations, explain complicated subjects and answer questions that once required hours of research.

And it is getting better remarkably quickly.

It would therefore seem reasonable to conclude that expertise is becoming less valuable.

I think the opposite may be happening.

The easier technology makes it to produce an answer, the more important it becomes to have someone who knows whether the answer makes sense.

Information Is Becoming Cheap

I have spent much of my career working with technology and teaching people how to use it.

For years, acquiring technical knowledge required considerable effort. If you wanted to learn a programming language, database system or complicated piece of business software, you bought books, attended classes, read documentation and spent countless hours experimenting.

Even finding the answer to a relatively simple technical question could consume an afternoon.

Today, I can describe a problem to an AI system and receive a plausible solution in seconds.

That is an extraordinary achievement.

But there is a difference between having an answer and understanding a problem.

And that difference is becoming increasingly important.

The Question Matters More Than It Used To

Imagine asking an AI system:

How can we make our customer-service department more efficient?

It might provide twenty excellent suggestions.

Automate routine inquiries. Build a knowledge base. Analyze response times. Introduce self-service tools. Categorize support requests. Use AI to draft responses.

None of those ideas is necessarily wrong.

But which problem are we actually trying to solve?

Perhaps customers aren't receiving answers quickly enough.

Perhaps employees are answering the same questions repeatedly.

Perhaps the company's software is creating the problems that generate the support calls in the first place.

Perhaps management is measuring the wrong things.

Perhaps the department is already efficient and customers are unhappy for an entirely different reason.

AI can help investigate all of these possibilities.

But somebody still has to recognize which questions are worth asking.

That requires judgment.

I've Seen the Same Thing in Technical Training

One of the interesting things I've observed through teaching technology is that the most valuable part of a class is often not the material in the book.

A student asks a question.

Then another student says, "We have that problem too."

Suddenly we're no longer talking about a software feature.

We're talking about how their organization actually works.

Why does this process require six approvals?

Why does one department enter information differently from another?

Why are employees maintaining spreadsheets containing information that already exists somewhere else?

Why does everyone know a process is broken, yet nobody changes it?

Those conversations require technology knowledge, but technology knowledge alone isn't enough.

You need enough experience to recognize what you're looking at.

AI doesn't eliminate that requirement.

It may actually expose how important it is.

Knowing How to Use a Tool Is Not the Same as Knowing What to Do

This isn't unique to artificial intelligence.

A spreadsheet can perform millions of calculations perfectly and still produce a meaningless financial model.

A database can return exactly the records requested by a poorly designed query.

A programmer can build precisely what was specified even though the organization needed something entirely different.

A musician can play every note correctly and still leave an audience unmoved.

Tools execute.

People decide what is worth executing.

AI dramatically increases what an individual can produce, which makes the quality of those decisions even more consequential.

Experience Is More Than Stored Knowledge

This is where I think we sometimes misunderstand expertise.

We tend to think of an expert as someone who knows more facts than everybody else.

If that were the definition, AI would present a formidable problem. No individual can compete with a system capable of accessing and synthesizing enormous amounts of information.

But much of expertise isn't knowing facts.

It's recognizing patterns.

It's remembering that the last three times someone proposed a particular solution, something unexpected happened.

It's noticing the question nobody in the meeting is asking.

It's understanding when the technically elegant solution will never survive contact with the people expected to use it.

It's knowing when to follow the accepted procedure and when circumstances justify questioning it.

And sometimes expertise is simply having made enough mistakes to recognize one before making it again.

That kind of knowledge is difficult to capture in a manual.

AI Can Make an Expert Much More Capable

This is the part of the AI discussion that interests me most.

I don't think the important comparison is:

AI versus the expert.

A much more interesting comparison is:

The expert without AI versus the expert using AI.

Give an experienced accountant an AI system and that person can potentially analyze information faster, investigate anomalies more efficiently and communicate findings more clearly.

Give it to an experienced engineer and the engineer can explore alternatives, generate prototypes and investigate unfamiliar technologies much faster.

Give it to an educator and lessons, exercises and explanations can be adapted to the needs of a particular group of students.

Give it to a musician and it can assist with transcription, practice materials, research, promotion and countless administrative tasks that previously consumed time better spent making music.

In each case, AI supplies leverage.

The person supplies direction.

This Changes What We Should Be Teaching

For decades, much of technical education has concentrated on procedures:

Click here.

Enter this.

Run this command.

Use this function.

Memorize this terminology.

Those skills still matter, but AI is rapidly reducing the value of memorizing procedures that a machine can explain whenever they're needed.

That should allow education to move toward something much more interesting.

Give people problems.

Ask them to investigate.

Make them explain why they chose one approach rather than another.

Let them discover that several technically correct solutions can produce very different business outcomes.

Teach them to challenge assumptions.

Teach them to recognize when the question itself is wrong.

In other words, teach judgment alongside technology.

The People Who Know Something May Matter More, Not Less

There is understandable anxiety about what artificial intelligence will do to employment.

Some jobs will disappear. Others will change substantially. New ones will emerge. We've seen versions of this transition with previous technologies, although AI may accelerate the process considerably.

But I would be reluctant to conclude that human knowledge therefore becomes irrelevant.

Something more subtle may be occurring.

When almost everyone has access to extraordinarily capable technology, simply having the technology is no longer much of an advantage.

The advantage moves elsewhere.

It moves toward knowing your industry.

Knowing your customers.

Knowing your craft.

Understanding people.

Recognizing patterns.

Asking better questions.

Knowing when something doesn't look right.

And having enough experience to understand why.

AI can give millions of people access to remarkably powerful tools.

What it cannot give all of them simultaneously is twenty years of experience using those tools to solve real problems.

That may turn out to be one of the great paradoxes of artificial intelligence.

As machines become more capable, genuinely knowledgeable people may become more valuable—not because they know everything, but because they know what to do with what the machines know.