When we leave the writing to artificial intelligence and large language models, our thinking suffers.
Photo: Sturlason
New technology has always had its enthusiastic supporters and its concerned sceptics, as was the case when the printing press was introduced in the 15th century and when the first personal computers appeared in the 1970s. Today, it is artificial intelligence (AI) and large language models that dominate the conversation. This technology will continue to shape all areas of society, including health care, research and education.
Using tools like ChatGPT and similar programs for text generation and language assistance can simplify writing processes. Well-formulated texts can be generated within seconds on the requested topics. In just a few years, AI has transformed research and academic writing (1) . In healthcare services, AI programs are already being used to produce accurate and comprehensive written records of medical histories from nothing more than audio recordings of conversations between patient and doctor (2) . Large language models can make diagnoses and treatment decisions faster and more precisely than doctors (3) . ChatGPT has even been shown to reduce anxiety and depression in cancer patients (4) . Efficiency and creativity are increasing as a result of AI.
But what are we losing in the process? This topic has been widely discussed, both within and beyond the pages of the Journal of the Norwegian Medical Association (5–8) . Will AI gradually destroy our ability to process information, to reflect and to reason? Can researchers simply leave the writing of scientific articles and theses entirely to AI? And what value-based choices will AI make on our behalf?
Will AI gradually destroy our ability to process information, to reflect and to reason?
Writing is closely linked to thinking. It is a way of thinking, a learning process that helps us better understand reality, grapple with conflicting viewpoints and clarify our own thoughts. As someone has stated: it is only when I write something myself with the intention of communicating it to others that I understand what I truly think. When we leave the writing to ChatGPT, we are less engaged in the thinking process. But it is convenient, less time-consuming and therefore tempting. My generation and those before us learned to write through years of schooling, being pushed, encouraged and guided to do our own writing. Will future generations receive the same educational grounding and, in doing so, develop stronger and more nuanced thinking skills? Or will they just learn to hand over the writing to AI?
An experienced author and professor of writing in the United States has vividly described what happens to students when they rely on AI in the writing process (7) . For her, writing is a joy: hard work, but a joy. She believes that much of that joy disappears when AI is used as a writing aid. Her concern can be summarised as follows: what happens to students who will never get to experience the reward of pushing themselves toward a difficult thought that seems obvious when written down?
In an experiment at the Massachusetts Institute of Technology, published as a preprint, participants who wrote an essay with the help of a large language model displayed lower EEG brain activity during writing, recalled less of what they had written afterward and demonstrated less ownership of the text than those who wrote without AI support (8) . What are the implications of these findings for the future of teaching and learning?
Changes to course design and assessment formats in medical education and other professional study programmes are inevitable. Students need to be trained and tested more rigorously in reflection and reasoning, both orally and in writing, than current home exams and multiple-choice exams allow. Home exams can no longer be used as an evaluation method because they practically invite extensive use of AI. Researchers must read and adhere to AI usage guidelines devised by leading scientific journals and international editorial groups (9) . We must all remain aware of the capabilities and limitations of AI.
It is often said that the future belongs to those who know how to use AI effectively. Those who do not will be left behind. Another prediction warns that AI could create a new class divide – between those who have learned to write without AI and those without extensive practice in writing their own texts. Reflection, scepticism of prevailing opinions and the ability to articulate new ideas are prerequisites for progress. This applies to all areas of society, including medicine.