The application of Artificial Intelligence in Education (AIED) is experiencing widespread interest among students, educators, researchers, and policymakers. AIED is expected, among other things, to enhance learning environments in the higher education system. However, in line with the general trends, there are also increasing concerns about possible negative and collateral effects. The consequent social impact cannot be currently assessed in depth. Balancing benefits with social considerations according to a socio-technical approach is essential for harnessing the true power of AI in a responsible and trustworthy context. This study proposes a semi-systematic literature review of the available knowledge on the adoption of artificial intelligence (AI) in the higher education system. It presents a stakeholder-centric analysis to explore multiple perspectives, including pedagogical, managerial, technological, governmental, external, and social ones. The main goal is to identify and discuss major gaps and challenges in context, looking at the existing body of knowledge and momentum. AIED should encompass pedagogical, ethical, and social dimensions to be properly addressed. This review highlights a not-always-explicit socio-technical perspective. Additionally, this study reveals a significant lack of empirical systematic evaluation of added value and institutional readiness. Because of the broad scope of the study and the intense ongoing debate on the topic, an exhaustive identification of the current body of knowledge is probably unrealistic, so this study aims mainly to identify the mainstream and major trends by looking at the most recent contributions.

Artificial Intelligence in education (AIED), particularly following the emergence of generative AI, has sparked increasing debate not only for increasing efficiency and productivity in academia but also its broader social implications. While discussions on AIED often focus on its potential benefits, increasing attention is being given to broader social concerns, including the risk of dehumanization, the development of skills, and overall well-being. These aspects remain insufficiently addressed in current literature. To explore these aspects and address this gap, the study proposes a research model that utilizes Stimulus-Organism-Response (SOR) as a structural framework and introduces a new conceptual model – Dehumanization-Skills-Wellbeing (DSW). The model captures the complex dynamics of AIED with these key social pillars. Drawing on survey-based data collected from students and instructors across higher education institutions in Saudi Arabia, findings indicate that higher levels of AIED use, often including generative AI tools, are positively correlated with empathy, support, and belonging that collectively reflect lower perceived dehumanization. These human factors are further associated with lower levels of anxiety, depression, and social isolation. Additionally, while cognitive skills development such as critical, technical and creative show a significant correlation with AI use, the impact on non-cognitive skills such as socio-emotional and collaborative skills remains unsupported and warrants further investigation. By examining these interconnected aspects, the study offers deeper insights into the ways AI may shape the social and emotional fabric of educational environments and guides the development of more human-centered and socially responsible AI adoption in higher education.