A five-month MIT study has found that generative AI is undermining student confidence, eroding foundational learning practices, and creating what researchers call "cognitive surrender" among students who lean on chatbots at the first sign of difficulty. The findings, released by an Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training, arrive as schools and universities nationwide struggle to set coherent policies for a technology that is advancing faster than institutional safeguards.
The committee gathered input from students and staff across the institute. Its report documents a campus where AI use is widespread but inconsistent - some professors encourage experimentation, others practice what the report terms "AI refusal" - and where the tools have disrupted everything from take-home exams to study groups and office hours.
How AI is changing what students know and know how to do
The report pulls no punches about the downsides. It describes a weakening of "decades of distinctive MIT community norms and values about rigor, the creative friction required for learning, collaborative problem solving, and personal integrity." Students, it found, are increasingly isolated and the longstanding social contract between learners and instructors is fraying.
"Getting the right answer from a chatbot can create the illusion of learning - but it can also trigger 'cognitive surrender', where students fall back on AI at the first hint of struggle," the report said. That dependence, it noted, is directly undermining both mastery and confidence. Assessment has become harder, too, as instructors lose visibility into what students actually understand versus what a tool produced for them.
Students themselves reported confusion and concern about "a lack of clarity, consistency and justification about the use of AI, within a given subject and across the curriculum." The tools are popular - used for everything from creative inspiration to anxiety-driven last-resort help - but the guardrails are not keeping pace.
What instructors told the committee
Not every finding was negative. Many instructors told the committee that AI tools help them build customized, interactive learning materials that let students explore subject content with greater depth. The report acknowledged that "the potential of these technologies to augment work across campus is immense." Still, the immediate priority is fixing the assessment and integrity gaps that AI has widened.
The committee recommended a campus-wide review of what students need to learn in an AI-saturated world, with departments and instructors redesigning classroom activities, assignments, and assessments from the ground up. It called for dedicated resources - including additional teaching assistants - and physical spaces explicitly designed for experimentation, community-building, and AI-free interaction and testing.
Building communities of practice
Rather than leaving individual faculty to solve these problems alone, the report urged MIT to create communities of practice where educators can share challenges, findings, tools, and techniques. The goal is to move beyond ad hoc, course-by-course policies toward a coherent institutional response. The committee also stressed that any solution must be broad - addressing not just academic integrity but the social and cognitive effects the tools are already producing.
For educators looking to build practical skills around these challenges, structured training paths such as the AI for Teachers Learning Path offer frameworks for integrating AI into the classroom without sacrificing the rigor and human connection that MIT's findings show are at risk. Broader resources covering policy, pedagogy, and tool selection are available through dedicated AI for Education training content.
Why this matters for educators
The MIT study confirms what many teachers and administrators are already seeing: students are using AI, often without clear permission or guidance, and the inconsistency is hurting learning outcomes. The report's recommendation for campus-wide review and redesigned assessments is a concrete signal that piecemeal syllabus policies are not enough. For K-12 and higher-ed professionals alike, the takeaway is that institutions need to address not only when and how students may use AI, but also how to redesign assignments so that the thinking work remains visible - and assessable.
Your membership also unlocks:
