What is “de-skilling” and what does it have to do with AI The consequences of widespread chatbot use on people’s cognitive abilities are increasingly being debated — but there’s a catch

The impact of artificial intelligence on people’s cognitive abilities has been discussed for some time, but lately – thanks to the growing spread of systems that use these technologies – the debate has accelerated. Online artificial intelligence systems have become widespread in recent years, mainly due to the success of ChatGPT, fostering the development of new forms of interaction based on a conversational approach; however, although they have proven surprising in some cases, these systems are still rather rudimentary compared to the technology’s potential. Moreover, most people use them to perform fairly simple tasks, but – as the Wall Street Journal pointed out – using these tools to summarize an email, for instance, among the many operations they can perform, is like taking a Lamborghini to deliver a pizza. Studies on the use of chatbots by the masses are still few and inconclusive: for now, it is difficult to predict how artificial intelligence might have concrete effects on individuals’ cognitive spheres, just as it remains complicated to understand this for technologies that have existed for much longer – such as television or social networks.

However, as the Atlantic recently noted, these concerns in the educational field are real: although there is virtually no concrete evidence yet, people are increasingly talking about “de-skilling”, that is, the gradual loss of skills among students (and others) due to the heavy use of chatbots. It is equally true, however, that since artificial intelligence systems became widespread, they have changed – perhaps forever – the way people approach studying and homework. In many subjects, in fact, tools like ChatGPT – thanks to their ability to write texts, solve exercises, and provide explanations based on given input – play an increasingly important role as a form of support.

How school has changed with AI

One of the consequences of this shift is that homework has become less suitable for assessing students’ learning levels – for this reason, more and more teachers are choosing to stop grading traditional take-home assignments or checking whether they are completed. Increasingly, professors also believe that ChatGPT or similar tools should somehow be integrated into teaching, in order to teach students how to use them consciously and intelligently.

Using AI to study, however, doesn’t necessarily mean putting in less effort. Many high school students, for instance, turn to ChatGPT for their Greek and Latin translations: besides performing the translations – which are also easily found online – the chatbot often helps understand specific passages in the texts. Others use ChatGPT’s voice interaction mode to become more fluent in English or another language and thus prepare for oral exams. The same approach, however, does not apply to scientific subjects, where ChatGPT often provides incorrect answers: although it may seem counterintuitive, many artificial intelligence systems are unable to correctly perform calculations, as they do not reason based on logical-mathematical principles but instead generate responses based on probabilistic criteria.

Nevertheless, the widespread use of chatbots has created some distortions in certain fields – mostly professional ones. For example, according to Business Insider, in the United States the adoption of ChatGPT as a productivity tool has led to some disparity in certain companies: those who fully integrate it into their work, treating it as a true assistant, actually gain advantages in terms of performance and income (albeit with all the related challenges), while those who use it only for very simple tasks do not seem to reap such significant long-term benefits.

It should also be noted that tools like ChatGPT are increasingly used for reasons related to the personal and private sphere rather than work or school, as acknowledged by OpenAI itself. Recently, much has been said about how many people, especially young people, use AI as psychologists or therapists – so much so that even Sam Altman has expressed doubts about this kind of interaction.