The AI shortcut: Are we sacrificing essential skills for convenience?

SINGAPORE: When Google first arrived on the scene, there were plenty of fears that the search engine would make us less intelligent.

As people began drifting away from libraries and encyclopedias, some worried that having information readily available would come at the expense of our ability to think for ourselves.

Those fears never quite materialised. A search engine could make finding information considerably easier, but it still left us with some work to do. We had to sift through pages of results, weigh up different sources, and decide which information was worth trusting.

The technology could help us find an answer, but the judgement remained ours.

AI has changed that balance. Rather than simply helping us find information, it can summarise long texts of information, explain it, solve problems, and even carry out the task itself.

People no longer need to do hard work. This is known as “cognitive offloading.” 

Although it is incredibly convenient, researchers warn that frequently depending on AI could weaken the skills we honed for years.

For instance, a 2025 study found that doctors trained to detect polyps during colonoscopies appeared to lose some of that ability after relying on AI assistance.

After three months of using AI, their polyp detection rate fell when they returned to working without the technology. The researchers suggested that the doctors had become less able to recognise the visual signs they had previously learned to spot.

The concern is not limited to medicine. Similar findings have emerged in education, where researchers are examining what happens when students become accustomed to having AI help them through problems.

Machine-learning researcher Grace Liu of Carnegie Mellon University in Pittsburgh and her colleagues reported in an April preprint on arXiv.org that people who used an AI assistant while practising SAT questions had more difficulty producing the correct answers once the tool was taken away.

They were also more likely to skip questions after losing access to the AI than participants who had never used it. 

Economist Alp Sungu of the Wharton School at the University of Pennsylvania and his colleagues found a similar effect among high-school students using AI tools such as ChatGPT to solve math problems.

The students performed better when the AI was available, but once it was removed, they performed worse than students who had never used the tool.

The alternative

Michael Gerlich, a professor at the Swiss Business School in Zurich who studies AI and cognitive offloading, recommends using AI merely as an “assistant” rather than depending on it entirely

Instead of asking AI to do everything, he suggests people should first work on the problem alone. 

When they need extra data or get stuck, they can ask targeted questions without disclosing details about the wider project.

Once they have arrived at an answer themselves, they can give the AI both the question and their solution and request feedback.

Gerlich says this approach is more helpful and does not erode people’s skills.

Read also: Man weighs new job offer with same salary but no WFH—Redditors urge him not to make the switch