AI can speed up work while eroding the skills behind it

Keep the work that teaches you

Skill loss happens when people hand tasks to AI often enough that they become less able to do them on their own. That does not mean every task should remain manual. AI can take on repetitive, administrative and formatting work, and help people learn new things. The line to watch is whether the task itself is how you learn.

If you are preparing to speak or discuss a subject in a meeting, you need to understand the material and be able to recall it without AI. For routine work, automation may make sense. For work that develops a skill you will need later, replacing your own effort is a worse trade.

The distinction matters because the skills most vulnerable to neglect are also the ones that make people useful beyond producing a quick answer:

Frequent AI use can also reduce motivation, engagement and a sense of responsibility for work. It may make it harder to remember how a task was done. Those costs are not limited to the quality of the output; they affect the worker’s ability to take the next task on independently.

Use AI without handing over your judgment

A practical approach is to do the first round of thinking yourself. Assess the problem, sketch possible solutions or connect ideas, then use AI for feedback, questions or alternative scenarios. For writing, simple messages are a good place to keep your own voice; for a harder task, AI can draft, but the worker should revise it.

The same principle applies to learning. AI can explain a topic, offer examples or act as a tutor. It can also quiz you, role-play a situation, challenge your ideas or give feedback. These uses keep the user involved; asking for a finished answer does not.

People early in their careers may be particularly exposed because they are still building foundational skills. They can ask AI to help improve their work or serve as a mentor, rather than asking it to do the work for them.

Workers also need to check what AI produces, adapt it and take responsibility for the result. That is a quality-control requirement, but it also preserves a link between the work and the person accountable for it.

The skills employers say they need

The World Economic Forum’s “Future of Jobs” report lists critical thinking, problem-solving, creativity, communication and writing, and resilience among skills gaining importance as AI changes work. The list also includes AI and big-data skills, networks and cybersecurity, technological literacy, continuous learning, leadership, talent management, analytical thinking and environmental responsibility.

That mix of technical and human capabilities is the point: learning to use AI does not replace the need to reason, communicate or adapt. A PWC report says salaries in AI-related professions are 56% higher than in professions that do not use AI. E&Y is investing $100 million to reward and recognize employees who show human qualities, including business acumen, judgment and adaptability.

For employers, the challenge is to make those expectations concrete. They should set clear rules for when staff may use AI and when a person must be involved, explain where the company does and does not use it, and provide opportunities for employees to develop their skills. The source also points to growing technology stress as AI use increases, including uncertainty and fear of job loss.

My view is that “use AI responsibly” is too vague to protect anyone’s skills. The useful question is which parts of a workflow employees must still be able to perform themselves. Without that distinction, a company can encourage AI adoption while quietly removing the practice through which its people learn.

A confidence problem, not just a skills problem

New data published by the University of California suggest AI can increase dependence and reduce confidence, but can also create false confidence, in part because it often responds approvingly to a user’s reasoning. Workers therefore need to assess their own strengths and weaknesses rather than treating an encouraging response as proof that their judgment is sound.

The broader case for deliberate use is straightforward: AI can help people learn, but it can also make it easy to skip the effort that learning requires. The tension for employers is whether they can gain the speed of automation without stripping away the practice that produces capable, accountable workers.