AI Is Not Replacing Expertise. It Is Replacing Isolated Expertise.

We keep hearing that AI will replace domain experts, especially in fields that take years of training to enter.

I read it differently. What AI displaces is not expertise itself. It is the habit of using your expertise inside one domain only.

I have spent 20 years in marketing, but I did not start there. I studied fine art, made animation and film, worked on digital humans and VR, spent years between content and technology, and now work where AI and marketing meet.

Looking back, the biggest asset in my career was never any single specialty. It was the ability to connect different things into one judgment. AI is making that ability far more valuable right now.

When execution gets easy, deciding gets hard

Until recently, execution was the expensive part. You needed someone who could write the copy, make the design, ship the code, cut the film. So depth in one field was scarce.

That has changed. Ask, and a draft appears.

The hard part comes next. Should we ship this? Does it fit the brand? Is it what the customer actually wants? Is now the right time? Do we revise, or throw it away?

As execution got faster, judgment got heavier.

I wrote about this in my book, Augmented Experts. An augmented expert is not someone who uses a lot of AI. It is someone who can decide, out of everything AI produces, what is worth keeping.

Why I keep returning to the idea of the fusion expert

By fusion I do not mean a generalist who knows a bit of everything. Almost the opposite.

I mean someone with a clear specialty who can translate it into the language of another domain. Someone who reads design in the language of the business. Someone who reads technology in the language of customer experience.

Such a person may not be as deep as the specialist in any one field. What they can do is notice the misalignment where two specialties meet.

Beautiful but unsellable. Technically possible but commercially pointless. Good for the customer but wrong for the brand. Buildable, but not worth building now.

Noticing that gap is, I think, one of the important forms of expertise in the AI era.

Netflix is saying something similar

On July 19, Elizabeth Stone appeared on Lenny’s Podcast. Stone is Chief Product and Technology Officer at Netflix, overseeing engineering, product, and design.

What caught my attention was the part about hiring. Netflix is no longer looking only for depth in a single domain. It increasingly looks for people who can move across areas and see the whole structure of a problem. The episode title carries the contrast: betting on systems thinkers, not specialists.

One passage stayed with me. Juniors used to build craft by doing repetitive execution themselves. People entering now start in an environment where AI handles much of that repetition.

Which creates a new problem. Who decides whether the output is any good?

Producing an output and recognizing a good one are different skills. Stone notes that responsibility for the quality of what ships does not go away. I would add that it has grown heavier, simply because far more gets produced.

Stone calls the ability to recognize a good outcome a very scarce skill. It is not quickly acquired. It takes experience, and it takes learning from people who already have it. I would add one more thing: the hours spent comparing many outputs until you have your own standard.

The easier AI makes execution, the scarcer that ability becomes.

Adoption is not the same as results

This is not just my impression.

In Atlassian’s State of Teams 2026, released in April, 89 percent of executives said AI increased speed. Only 6 percent said they had clear examples of organization-wide AI ROI.

McKinsey’s State of AI 2025 found 88 percent enterprise adoption, while 39 percent reported an impact on EBIT at the enterprise level.

Using AI is no longer a differentiator. Deciding what to do with the same tools is.

In the same Atlassian study, the 14 percent of teams that did crack AI ROI had one thing in common. It was not a new tool. It was a new way of working.

So is there an opening for people with long careers

I think there is. In fact, a rather good one.

With one condition. Experience by itself is not yet an asset. Twenty years of work is a record. It becomes an asset only when what you saw, what you chose, what you discarded, and where you changed your mind comes out into the open.

That is why I am starting a series called Experience to Meaning. About not leaving experience as a résumé line, but translating it into meaning, and then into standards.

The smallest way to start

Write down three decisions you made this week.

Not what you chose, but why. Not what you made, but what you threw away. And if you can, what you were looking at when you decided.

It will look trivial at first. After a few entries, patterns appear. What you keep treating as important. What others walk past and you catch first. What everyone calls essential and you quietly skip.

That is the judgment standard hiding inside your expertise. It is probably also the part AI finds hardest to copy.

Twenty years in one field is not the point. What you found in there, and how it connects to somewhere else, is.

Execution is getting cheaper. Judgment is not.

---

Sources

Lenny’s Podcast, “Why Netflix is betting on systems thinkers—not specialists—in the AI era | Elizabeth Stone (CPTO)”, July 19, 2026.

Atlassian, “The State of Teams 2026”, April 27, 2026.

McKinsey & Company, “The State of AI in 2025: Agents, Innovation, and Transformation”, November 4, 2025.