For generations, people built workplace expertise through repetition: doing simpler tasks, making mistakes, and slowly taking on harder problems as they proved they were ready. That path is becoming less linear, and as AI takes on routine execution, employers expect people to bring skills like stronger judgment, AI delegation, and leadership to the forefront of their roles.

The question isn’t whether expertise still matters. It definitely does. The real questions are how people will develop expertise faster, and how employers will recognize those skills once they have them. That second question is harder to answer than it sounds because most companies simply don’t have good visibility into what their people can actually do.

That’s the gap a practice I call skillspotting is meant to close. Skillspotting is the practice of identifying what people can do today, including capabilities they’ve developed beyond the roles they were hired for. It’s about looking past static titles and credentials to understand real, demonstrated skills.

But knowing what to look for starts with understanding how AI is changing which skills matter, and how people build them.

AI is Changing What it Means to Be Skilled

Deloitte has described how automation can disrupt the entry-level work that employees traditionally learn by doing. When AI handles simpler tasks, the on-the-job repetition that once built expertise disappears. Employees lose that natural path to practice and employers lose a natural way to observe who’s developing real capability.

That loss makes skillspotting even more important, because the old signals people relied on to judge capability are getting less reliable too. A polished answer used to be a decent proxy for skill. Now AI can produce one instantly, so a good output no longer tells you much about the person who produced it.

What matters instead is what happens around the output: whether someone can direct the work, recognize when an answer is wrong, and explain the reasoning behind it. That’s really a test of judgment, not knowledge, and it shows up most clearly in how someone uses AI itself. When you break it down, using AI well is a form of delegation. You need to know what to ask for, what to challenge, and where the output can’t be trusted, and that requires enough grounding in the work to catch errors before they become expensive.

‍

Read full article here