deskilling

Generative AI, and more precisely large language models, have nowadays rapidly become useful in education, professional work, and everyday tasks. While these systems offer notable benefits to make work efficient and improved accessibility, the recent evidence suggests that their usage has cognitive costs. This paper presents a structured review of 16 reviewed studies investigating the impact of Gen AI usage on cognitive offloading, metacognitive engagement, automation bias, and human skill. By studying experimental, survey-based, qualitative, and theoretical research, this review synthesizes several findings across several fields to identify recurring mechanisms through which GenAI usage reshapes cognition. Our analysis reveals a consistent pattern, where GenAI often improves short-term task performance, but it frequently minimizes effortful cognitive processing, diminishes self-regulation, and changes skill developments. Rather than constant de-skilling, this paper points out reconfiguration of skills, favoring efficiency-oriented competencies such as prompting and verification over generative reasoning and critical reflection. Our study further highlights these outcomes are not unavoidable but are strongly mediated by interaction design, educational practices, and socio-technical environments. By showing these effects as the “cost of convenience,” this paper contributes an integrative perspective on the cognitive effects of Gen AI and highlighting the ways to make AI design, education, and governance cognitively sustainable.

As conversational artificial intelligence models become ubiquitous fixtures in professional and personal workflows, the nature of human-tool interaction has the potential to gradually shift from mechanical utility to parasocial dependency. While existing literature frequently addresses adverse events related to data privacy, algorithmic bias, and intellectual deskilling, the pathopsychological sequelae of sustained AI integration remain largely unexamined. Here, we describe and preliminarily characterize the AI Assistant Syndrome, a human psychiatric construct encompassing acute chatbot-induced anxiety and mental exhaustion. This preliminary description will guide future systematic investigations to recognize, prevent, and treat this syndrome.

با فراگیر شدن مدل‌های هوش مصنوعی مکالمه‌ای در جریان‌های کاری حرفه‌ای و شخصی، ماهیت تعامل انسان و ابزار می‌تواند به تدریج از کارکرد مکانیکی به وابستگی شبه‌اجتماعی تغییر یابد. اگرچه پژوهش‌های موجود عمدتاً به رویدادهای نامطلوب مرتبط با حریم خصوصی داده‌ها، سوگیری الگوریتمی و کاهش مهارت‌های فکری می‌پردازند، اما پیامدهای آسیب‌شناختی روانی ناشی از ادغام پایدار هوش مصنوعی تا حد زیادی بررسی نشده‌اند. در اینجا، سندرم دستیار هوش مصنوعی را توصیف و به طور مقدماتی مشخص می‌کنیم؛ این سازه روان‌پزشکی انسانی شامل اضطراب حاد ناشی از چت‌بات و فرسودگی ذهنی است. این توصیف مقدماتی، راهنمای پژوهش‌های نظام‌مند آینده برای شناسایی، پیشگیری و درمان این سندرم خواهد بود.