Employee judgment erodes when AI takes over analytical work. New hires lose the task work that once built problem-solving skills.
Every company rolling out AI tracks adoption. Dashboards log how many employees use the tools, how often they log in, and how much time is saved. Almost none of them track what those tools are doing to the people using them.
In a recent survey of 70 C-suite leaders, Boston Consulting Group found that half are already watching judgment and problem-solving skills weaken inside their own organizations. More than 60% believe that loss will hurt the business within three to five years.
Judgment, problem framing, and creative thinking are the skills that matter most to long-term performance. They're also the ones leaders say AI is most likely to erode.
What happens when teams lean too hard on AI
Nearly 90% of the executives BCG surveyed said people at their companies have stopped checking AI's work. Drafts come back polished, analyses arrive with confident conclusions — and nobody asks whether the reasoning holds up.
Once people stop questioning the work, they stop taking ownership of it. More than half the leaders said employees are less willing to own their decisions than they were before AI arrived. Nearly half said their teams think more alike than they used to: everyone runs the same problem through the same tools and gets the same answer back. Forty-three percent said their people argue less about whether those answers are right — and when no one pushes back, mistakes go uncaught.
A separate BCG study of AI training programs found that companies tracking course completion often confuse it with genuine skill. Employees pass the assessments and demonstrate understanding in the classroom, then return to their desks and either fall into old habits or lean harder on AI to do the thinking for them.
New hires are developing more slowly, too. AI now handles the researching, drafting, debugging, and problem-solving that once trained them. When a leader decides which tasks to hand AI, they're really deciding which skills the next generation will never build. With simpler analytical tasks automated, new hires are now expected to do work that once required five to ten years of experience. Companies have eliminated the entry-level work that taught people how to do the job — and no one has figured out how to replace it.
The skills people lose when AI does the work
Senior employees aren't immune. When experienced professionals get busy, they let others make calls they used to make themselves. One leader called it the "autopilot" pattern: their work still looks sharp, so nobody notices they're making fewer real decisions on their own.
A third of the leaders said senior employees are teaching less than they used to. People go straight to a language model for answers instead of asking a colleague who already has context. Senior people used to explain not just *what* to do, but *why*. Those conversations happen less and less.
The science backs this up. A 2025 MIT study found that participants who used ChatGPT to write showed significantly lower brain activity than those who used web search or worked unaided. They also remembered less of what they'd written afterward. Every skill that goes unexercised atrophies — and judgment is no different.
Teaching people to think before they prompt
A handful of companies in BCG's research have started redesigning work to protect thinking skills. At Shell, junior employees now work through problems on their own before touching an AI tool. They frame the question, test their assumptions, and produce an initial analysis by hand — then use AI to refine it. Early results are promising: junior employees explain their reasoning more clearly and need less from managers.
Most companies still care more about how many employees use it than anything else. As AI gets cheaper, the temptation to outsource more work will only grow. But the ones that will ultimately pull ahead will be the ones whose employees can still think without it.