Good judgment used to come from grunt work. Now AI is doing a lot of the grunt work, and some Gen Zers prompting LLMs are struggling to explain or defend what the AI hands them. That results in a growing backlog of work that bosses are stuck checking themselves.
Enter the “judgment bottleneck,” outlined by McKinsey Senior Partner Alex D’Amico and his coauthors in a new article. One executive described it this way: “I used to be able to give one of my teams guidance on a Monday and have a full week before they needed my input again. Now I hear back from them within a few hours, asking me to review what they’ve done. I don’t have the bandwidth to review it all.” (A tortoise-and-hare story for the AI age: The hare finishes first, then waits while someone retraces its steps.)
This isn’t Gen Z’s fault. Zoomers, like every other colleague, are increasingly expected to use AI to move faster. But without the right training, what they gain in speed comes at the expense of practice. People build judgment through repetition, feedback, and low-stakes mistakes, the authors note, not by receiving finished answers.
None of this is news to Gen Z workers. In one survey, entry-level employees said they worried more about AI eroding their thinking skills than about AI taking their jobs. In another, 62 percent of Gen Zers reported using it as a crutch.
The risks are both near and long term. When juniors can’t vet AI output and bosses lack time to check it, errors can slip into real decisions. Meanwhile, when junior colleagues don’t get the training and development they need, a company’s leadership bench thins out, too.
To address these risks, employers may need to rethink their AI training programs. Most AI onboarding covers how to use the tools and keep data secure. Fewer programs teach junior staff to interrogate AI’s analysis—and know when to trust or challenge it.
One approach: Let the junior employees take a first pass on their own, then compare it with AI’s output. Here, a manager helps their direct report spot what they missed, what AI missed, and how to make the final product better than either version. Baking accountability into the process also matters: Whoever hands in AI-assisted work should be ready to explain and defend it. Employers may also need to make peace with a slightly slower clock. Building in time to review and challenge AI’s work costs hours up front, but it catches problems before they compound and teaches junior staff how to check their own work. That’s two wins for one delay.
The goal isn’t to use less AI, but to use it differently. McKinsey Senior Partner Kate Smaje says her product teams work by a simple mantra: “Not AI that does it for me, but AI that does it with me.” In practice, that means spending more time on how teams operate and what skills people need, not just on which tools they use.
The upside for Zoomers: As an AI-native generation, they have a head start on colleagues still getting comfortable with the tech. Pair that fluency with the judgment to question AI’s answers, and the advantage compounds. “There’s a premium on those capabilities now,” Smaje says.
Turns out being a little judgmental about your AI’s work might be the most valuable habit a 23-year-old can build. |
