Massachusetts Institute faces a decline in students' mental friction: a crucial test for teaching models in the region
Listen to this article
Read by Anchor
At the editorial office we examined the recently issued internal report from the Massachusetts Institute of Technology, which observed a tangible decline in students' attendance at faculty office hours and a reduction in time allocated to solving complex problems in interactive work groups. The report warned that generative AI tools have begun to cause "the erosion of creative friction, which is the direct debate and discussion between students and faculty members and their assistants that has shaped the institute’s educational identity for decades. This observation places universities and technical training centers in the Gulf, Egypt and the Levant before a structural dilemma, as the challenge is not limited to purchasing smart model licenses for students but extends to the risk of eroding the analytical discussion environment that builds the actual skill of solving complex problems.
To address this shift in academic behavior, the Massachusetts Institute’s Human Insights Initiative, launched in 2024 with total funding of $5.5 million to support 67 projects across 31 units, allocated about a third of its budget to examine the relationship between artificial intelligence and humanity and society. The initiative operates in partnership with the College of Humanities, Arts and Social Sciences and the Initiative for Social and Ethical Responsibilities in Computing. Heather Baxter, professor of anthropology and co-dean of the college, explained that examining the phenomenon does not stop at individual ethical decisions or workforce retraining, but focuses on how AI reshapes social bonds, modes of individual interaction and collective thinking within classrooms.
The academic risks are precisely embodied in the joint research of Eric Klobfer, professor of comparative media studies, and Bawan Sinha, professor of brain and cognition sciences, who lead a project measuring AI’s impact on mental processes. The researchers pose a fundamental question of whether generative models drive increased information consumption at the expense ofoffloading cognitive functionsand delegating thinking to the machine, or whether it can enhance cognitive capacities when integrated into genuine innovation processes. This concern intersects with findings from the Parallel Project in political science, involving professors Adam Berinsky, Charles Stewart and Chara Budimata, whose study of the 2026 Alaska primary elections showed that voters’ reliance on Anthropic’s Claude model gave them incomplete information about the candidates, underscoring the danger of automated mediation in knowledge formation.
Institute actions did not stop at warning; they moved to redesign curricula to resist automated indoctrination. Anthropology professor Graham Jones and computer science scholar Arvind Satyanarian launched a course on human-centered user experience design, in which students revealed a genuine tension between building a system that serves humans and a system that mimics humans and pushes toward dangerous dependency. In a related move, associate professor Duway Banerjee revived the computing-and-society track across 11 academic units after a period of stagnation, while Michael Tres led intensive workshops to train faculty in guiding university writing assignments based on a survey conducted by the University of Illinois in 2026. Seth Monakin, co-chair of the faculty board in the initiative, sees the current challenge as comparable to the moment when the institute helped shape the reality of World War II in 1941, with a crucial difference: AI questions cannot be answered by technical capabilities alone, apart from the humanities.
What This Means for the Region
Universities and higher-education councils in Saudi Arabia, the United Arab Emirates, Egypt and Jordan rushed to embed AI tools in curricula and academic programmes, viewing rapid adoption as a sign of institutional modernization. MIT’s findings demonstrate that limiting achievement to digital usage figures hides a silent decline in mental friction between students and faculty. The academic decision required in the region now is to reassess measurement methods and examinations, shifting from ready-made homework to interactive oral discussions and direct analysis sessions that protect independent thinking from erosion.
At the business and tech-sector level, education outcomes that rely entirely on automated generation produce graduates who can formulate commands but lack the muscles of critical thinking and deep code auditing. When a student is relieved of the effort of trial and error, employers in the region inherit a workforce that is fast at superficial production yet fragile when confronting atypical technical problems. Your step as a training or hiring manager calls for testing a candidate’s ability to perform abstract logical deconstruction without resorting to models, to ensure building competencies that possess original solutions rather than mere electronic replication.