BCG: Half of Managers See AI Eroding Their Employees' Skills

Half of the managers surveyed by Boston Consulting Group already notice that employees are losing professional skills because of AI. In BCG's latest study, published in 2026, the situation is described as "requiring attention." The author of an analysis on Habr, who has worked with AI for about two years, offers a practical response: ten habits, five of which are worth dropping and five of which, conversely, are worth keeping even though they look ambiguous.

The cost that doesn't show up on your token bill

When people talk about the cost of AI, they usually count subscriptions and tokens. The BCG study adds a line item that never appears in the accounting books: deskilling. The term describes a gradual weakening of professional skills that happens when the core work is handed to a model again and again.

The author of the original text describes a scenario many will recognize. At first you feel like a wizard: "summarize this study," "give me the three main takeaways," "prepare a presentation." Each of these commands saves hours. But in a single chain of prompts, a person hands off reading, analysis, and structuring all at once. These are precisely the skills for which an analyst, manager, or editor is valued on the market.

Ten habits: the logic of the breakdown

The author splits the practice into two equal halves. The first contains five habits he advises giving up, even though they deliver a noticeable productivity gain. The second contains five habits that seem questionable at first glance but make both the individual and their team stronger. The full list is given in the original on Habr; the announcement reveals only the principle behind the split.

The principle is this: a habit is judged not by how quickly it produces a result, but by what happens to the person's competence. If the model replaces thinking, the skill weakens. If it works as a sparring partner that checks and critiques, the skill grows. That is why a useful habit may well slow work down, and that is exactly why it is easy to mistake for a wrong one.

Key facts from the material:

• Data source: Boston Consulting Group study, 2026.
• Share of concerned managers: half of the surveyed managers see a loss of skills among employees.
• Term: deskilling, the weakening of skills when work is constantly handed off to AI.
• Format of the analysis: ten habits, five harmful and five useful.
• Author's experience: about two years of daily work with AI.

How to check your own habits

What follows are my editorial conclusions, not a retelling of the author's list. You can apply them right now, without waiting for the full text.

• Identify the skills that form the core of your profession: writing, data analysis, code review, negotiation. These are the most dangerous ones to hand over to a model entirely.
• Split your tasks into two types: delegable (routine work, formatting, initial search) and trainable (whatever builds your expertise).
• Periodically solve a typical task without AI and compare the result with your earlier one. That shows whether the skill is growing or has already slipped.
• Use the model as a critic: write a draft yourself first, then ask it to find the weak spots. You keep authorship, and AI strengthens the result.
• For teams: discuss not only "how many hours we saved" but also "what we have stopped being able to do ourselves."

The main question is who in the pairing is responsible for the thinking. As long as it is the human, AI works for their qualifications. When the roles swap, the time savings are gradually paid for with a loss of competence, and managers, judging by BCG, are already starting to notice.