AI Deskilling Emerges as Doctors and Programmers Lose Skills Through Over-Reliance on AI Tools

Is AI ruining our skills? Early results are in -- and they're not good

As more professionals begin to rely on artificial-intelligence tools in their work, could their hard-earned skills atrophy? That possibility is a growing concern for medical specialists, computer scientists and other workers. Seventy percent of nurses and 77% of physicians, for example, are worried about losing their skills because of over-reliance on AI systems, according to a survey of US health-care workers published earlier this month. Their fear might be justified. Evidence suggests that AI-driven 'deskilling' is starting to happen in medicine, computer science and other fields. Researchers are now discussing how to preserve important human expertise in the age of AI. "Just being aware that this phenomenon exists hopefully provokes some self-reflection about which skills people want to maintain and which they're willing to outsource" to AI tools, says Kevin Crowston, an information scientist at Syracuse University in New York. Spoiled by AI? A study of physicians in Poland who specialize in endoscopy -- the use of flexible probes to examine the inside of the human body -- shows how quickly AI tools can erode human abilities. The physicians, who had all performed at least 2,000 colonoscopies during their careers, were given access to an AI system that analyses colonoscopy images in real time and flags a type of precancerous intestinal lesion called an adenoma. The tool was available to the specialists on some days but not on others. Once physicians began using it, their performance dropped significantly whenever the system was unavailable. During the three-month period before the AI tool was introduced, the specialists found at least one adenoma during 28.4% of colonoscopies. During the three-month period after the tool was introduced, the adenoma detection rate for colonoscopies performed without AI assistance decreased to 22.4%. The findings, published last October in The Lancet Gastroenterology and Hepatology, suggest that even highly skilled professionals might get worse at tasks that their job requires as they become more dependent on AI tools, says Robert Wachter, a physician at the University of California, San Francisco, who is the author of a book on how AI tools are transforming health care. The study authors say that continuous exposure to such tools can cause clinicians to become "less motivated, less focused, and less responsible when making cognitive decisions without AI assistance". Co-author Yuichi Mori, a physician-researcher at the University of Oslo, says that more studies are needed to confirm the phenomenon. But people who use AI tools should be aware that they risk losing some of their skills, he adds. "There is no established solution against deskilling right now. It should be a very hot research topic in the next decade." No lesson learnt To investigate whether skills are being lost in the field of computer science, researchers at the AI firm Anthropic in San Francisco, California, designed a randomized controlled trial in which 52 software engineers were asked to perform a basic coding task. During the exercise, all 52 participants could search the web and access instructions on how to do the task. Half of the participants were prompted to use an AI assistant as well.

AI in Medicine: What Happens When Clinical Skills Fade?

Deskilling is the process of gradual loss or erosion of professional skills. It occurs when a human activity is delegated to technological tools, software, or AI systems, leading to the "unlearning" of technical skills and critical thinking. This phenomenon has already been observed in many fields characterized by heavy use of technology. For example, drivers who rely too heavily on global positioning system may gradually lose their sense of direction, pilots who use autopilot systems may have greater difficulty handling sudden emergencies, and workers who use automated tools tend to use their manual skills less. Similarly, students who make extensive use of automated calculation tools may develop a reduced ability to perform calculations on their own. Deskilling Risk In medicine, this phenomenon occurs every time new automated procedures have been introduced into diagnostic and care processes, as has been the case with the numerous diagnostic tools used in clinical practice today. With the advent of AI, however, this concern has become even more pronounced. Excessive reliance on algorithms and a view of medicine as a purely technical activity can reduce healthcare professionals' ability to make independent judgments and decisions. Is this merely a theoretical hypothesis -- albeit one based on experience gained in this and other fields -- or is there empirical evidence to support its validity? This is the question posed by a study recently published in The Lancet Gastroenterology & Hepatology. The researchers analyzed more than 1000 patients who underwent colonoscopy to evaluate the effects of the systematic use of AI-based diagnostic support tools to identify the presence of adenomas. For this purpose, the researchers compared the results of exams performed by endoscopists up to 3 months before the introduction of AI-assisted diagnostic tools at the study's participating centers with those performed 3 months after the introduction of these systems, but without using them during the exam. Starting on the implementation date, colonoscopies were performed with or without AI support according to a randomization procedure on the basis of the date of the procedure. The results revealed a particularly important and unexpected finding. The adenoma detection rate for standard colonoscopy decreased significantly from 28.4% (226 of 795) prior to the introduction of AI-based support to 22.4% (145 of 648) after AI exposure. Multivariate logistic regression analysis confirmed that exposure to AI was an independent factor significantly and negatively associated with adenoma detection (odds ratio, 0.69; 95% CI, 0.53-0.89). In brief, exposure to AI tools designed to assist in adenoma identification made endoscopists less capable of correctly detecting adenomas once that technological support was removed. Broader Risks The study has significant methodological limitations -- it is a retrospective, observational study conducted at a limited number of centers, with potential biases among the patients and endoscopists involved -- which are partially mitigated by using robust statistical methods such as multivariate analysis. Nonetheless, it represents one of the first pieces of scientific evidence in the literature to highlight the danger of losing clinical expertise in procedures that are delegated to automated support. This may not be the only risk. Excessive use of AI -- particularly generative AI -- could lead to a leveling off of skills, limiting or even hindering learning processes. Uninformed use of these tools risks not only eroding individual skills but also increasing dependence on these technological systems. In a recent systematic review, Federico Cabitza , PhD, associate professor of computer science from the University of Milano-Bicocca in Milan, Italy, highlighted potential vulnerabilities associated with AI use in physical examination, differential diagnosis, and clinical judgment. Cabitza has spent many years studying the causes and consequences of deskilling in medicine. Moreover, when AI proposes predetermined solutions, human input tends to be limited to what is often a superficial review. This can lead to a gradual reduction in the exercise of professional judgment and a shift in the perception of responsibility when errors occur. Building Resilience One possible strategy is to promote reskilling and upskilling programs. Increasingly, organizations -- including those in the healthcare sector -- are investing in continuing education to help professionals use AI as a decision support tool rather than as a substitute for their own skills. Another approach involves expanding and strengthening, among physicians, the skills that algorithms are unable to fully replicate, such as creativity, critical thinking, emotional intelligence, and ethical sensitivity. Some experts also suggest a balanced approach to AI use, one that involves using it selectively rather than continuously. This allows healthcare professionals to maintain their focus, decision-making autonomy, and judgment. Another potential benefit is the explainability of AI systems. This feature is also emphasized by regulations governing the use of AI in healthcare, including the EU AI Act, and may help physicians better understand the information and criteria used to support clinical decisions. According to a recent article in the British Journal of Psychiatry, which recently addressed these issues, it is imperative that physicians recognize the limitations of AI tools and continue to rely on and grow their knowledge, experience, and clinical skills. AI tools may assist with basic, administrative tasks in medicine but cannot replace the uniquely human interpersonal and reasoning skills of physicians. Physicians must prevent deskilling and learn how best to use AI tools in clinical medicine. In conclusion, the widespread adoption of AI in clinical practice, if not accompanied by appropriate training and oversight, could lead to increasing reliance on technology and the risk of compromising certain diagnostic skills that are essential for disease prevention and patient care. Action is needed now to ensure that these critical competencies are preserved as AI becomes more deeply integrated into healthcare. Eugenio Santoro reported being a senior researcher in digital health at the Mario Negri Institute for Pharmacological Research IRCCS (Istituto di Ricerche Farmacologiche Mario Negri IRCCS), where he directs the Unit for Digital Health Research and Digital Therapies. He also reported being a member of the ICT group of the National Federation of Italian Medical Associations (Federazione Nazionale degli Ordini dei Medici Chirurghi e degli Odontoiatri; FNOMCeO), the Technical-Scientific Committee of the Parliamentary Intergroup on Digital Health and Digital Therapies, the Working Group on Artificial Intelligence in Healthcare of the Emilia-Romagna Region, and the Control Committee of the Istituto di Autodisciplina Pubblicitaria (Institute for Advertising Self-Regulation). He reported having authored several books and scientific and popular articles on digital health and, in 2021, having edited the entry "Digital Health" for the X Appendix of the Treccani Encyclopedia, dedicated to the words of the 21st century. He reported teaching in several university master's programs. This story was translated from Univadis Italy, part of the Medscape Professional Network.

AI May Be Creating a Dangerous Problem No One Saw Coming: Why Doctors and Programmers Could End Up Trapped by It

As artificial intelligence becomes a routine part of work in healthcare, software development and other professions, researchers are warning that heavy reliance on these tools could weaken important human skills. Studies involving doctors and software engineers found that while AI can improve performance and efficiency, users may struggle more when working without it and retain less knowledge over time. Experts say the growing risk of AI-driven "deskilling" highlights the need to balance technological assistance with continued development of human expertise. Artificial intelligence has become a regular part of work across industries. From helping doctors detect medical conditions to assisting software engineers with coding tasks, AI tools are making work faster and often more efficient. Companies and organizations continue to invest heavily in these technologies, seeing them as a way to boost productivity and reduce workloads. However, as AI becomes more deeply embedded in daily work, researchers are beginning to ask whether constant reliance on these tools could come at a cost. A growing number of studies suggest that depending too heavily on AI may gradually weaken the skills people once developed through years of training and experience. Experts say the issue, often described as "deskilling," is emerging in fields ranging from healthcare to software development. "Just being aware that this phenomenon exists hopefully provokes some self-reflection about which skills people want to maintain and which they're willing to outsource" to AI tools, said Kevin Crowston, an information scientist at Syracuse University. Signs of Skill Erosion in HealthcareConcerns about deskilling are particularly noticeable in medicine, where AI tools are increasingly being used to support clinical decisions. A recent survey of US healthcare workers found that 70% of nurses and 77% of physicians worry about losing professional skills because of over-reliance on AI systems. Those concerns are backed by research published in The Lancet Gastroenterology and Hepatology. The study involved experienced physicians in Poland who used an AI system that could identify adenomas, a type of precancerous intestinal lesion, during colonoscopies. Before the AI tool was introduced, doctors detected at least one adenoma in 28.4% of procedures. After becoming accustomed to the technology, their detection rate during procedures performed without AI assistance fell to 22.4%. The researchers suggested that continuous exposure to AI tools can leave clinicians "less motivated, less focused, and less responsible when making cognitive decisions without AI assistance". Yuichi Mori, a physician-researcher at the University of Oslo and co-author of the study, said, "There is no established solution against deskilling right now. It should be a very hot research topic in the next decade." Similar Trends Emerging in CodingResearchers are seeing comparable patterns in software development. A study conducted by AI company Anthropic asked 52 software engineers to complete a coding task. While all participants could access online resources, only half were encouraged to use an AI assistant. When later tested on what they had learned, those who used AI scored significantly lower. The AI-assisted group averaged 50%, compared with 67% among participants who completed the task without AI help. The findings suggested that while AI users were able to complete the work successfully, they struggled more with questions that required them to diagnose coding errors and explain underlying concepts. According to Crowston, "Now you have this very odd disconnect between performance and learning. People can perform at a pretty high level, because they're basically borrowing skills from the AI, but they are not developing those skills themselves." Balancing Efficiency and ExpertiseExperts note that technology has replaced certain human skills before. GPS navigation, for example, reduced people's need to memorize routes. But researchers argue that generative AI is different because it automates tasks involving reasoning, interpretation and decision-making. Tapani Rinta-Kahila, an information systems researcher at the University of Queensland, believes the long-term impact could be significant if younger professionals miss out on foundational learning experiences. "Next generations of programmers may not understand the foundations of coding that well at all, if they lack the hands-on experience," he said. "The same goes for many other knowledge-intensive professions, such as accounting and law."