MIT 报告指出 AI 正在侵蚀 office hours、学习小组与师生信任

The report, published in June 2026, documents how AI is chipping away at core parts of the college experience. Students are already using it across the board, and the effects are showing: fewer show up to office hours, online discussion participation is down, and study groups in dorms and libraries are thinning out. For faculty, it's getting harder and harder to gauge what students are actually learning.

Faculty and students are losing trust in each other

Faculty say that policing unauthorized AI use is damaging their relationships with students. AI text detection software is unreliable and often flags work by non-native speakers or neurodivergent students as AI-generated, so the committee explicitly advises against using it. These tools also risk an arms race with so-called "AI humanizers," programs that make AI-generated text look human-written.

For theses and dissertations, the committee wants a different approach: every paper must disclose its AI use, and AI may never be listed as a co-author.

Students, meanwhile, fear false accusations and see a double standard when faculty use AI for slides, feedback, or grading while restricting student use. The committee recommends transparency rules for faculty too.

Lockdown browsers, exam software that locks down and monitors computers during tests, don't solve the problem either. The panel says the current generation is buggy and feels like surveillance.

Getting the right answer from a chatbot fakes real learning

The report builds on a guiding principle of "augmentation not automation": AI should extend human abilities, not replace them. Getting the right answer from a chatbot creates an illusion of learning and leads to intellectual surrender, with students falling back on AI at the first sign of difficulty.

That dynamic is now threatening one of MIT's most important programs. Faculty members are starting to consider AI agents instead of students as research assistants, which would hit the Undergraduate Research Opportunities Program (UROP), a pillar of the MIT experience: 93 percent of the class of 2025 participated at least once, and 58 percent of faculty served as mentors. The program exists to train students, not to provide cheap research labor, so replacing them with AI would gut its purpose, the panel argues.

Learning goals should come before AI policies

Rather than blanket rules, the committee says each course needs its own AI policy. The process should start with learning goals, move to assessment design, and only then address what AI use to allow. A poetry seminar has a completely different relationship to AI than a course on mathematical proofs, so a single institute-wide policy would be too loose for some contexts and too tight for others.

The report recommends a shift toward oral exams, semester portfolios, in-person discussions, and project-based work. Every course should spell out its AI policy in the syllabus with clear reasoning behind it.

Unequal access to AI tools widens performance gaps

Some students can afford premium AI access, others can't, and the committee warns that gap could drive real performance differences. MIT already offers all members access to various AI models through its Parley platform, with faculty and graduate students getting $30 per month in free credits. But premium subscriptions from OpenAI, Google, and Anthropic cost several hundred dollars a month, putting them out of reach for many students.

Broader research backs up the MIT report

The widespread AI use the MIT report describes matches what other schools are seeing. At Harvard, about 87.5 percent of respondents in 2024 said they used AI, with nearly half using it at least every other day. Around 25 percent said AI led them to visit office hours less, ask instructors for help less, and skip assigned readings. In the UK, usage among full-time students hit 95 percent by the end of 2025, and students from wealthier households used AI more often.

Several studies back up the concern about outsourced thinking. Anthropic found that students offloaded higher-order thinking like analysis and creation to Claude in nearly half of all conversations analyzed. A Chinese long-term study with more than 26,000 students found that AI use raised homework grades by 18 percent, but exam scores dropped 20 percent after six months.

A UC Berkeley study covering more than 500,000 grades showed that the share of A grades in writing- and coding-heavy courses rose by 13 percentage points since ChatGPT launched. The effect was concentrated in courses with a high homework share, pointing away from real learning gains. At Brown, average scores dropped from 96 percent on a take-home exam to 48.6 percent on the follow-up in-person test.

A two-year study at Vrije Universiteit Amsterdam by legal scholar Thibault Schrepel produced a different result. He randomly split students into three groups: no AI, AI without guidance, and AI with training. The no-AI group finished last both years. Even the group using AI without guidance did better, despite often accepting AI suggestions without question in class. Schrepel had expected the opposite and ended up rejecting his own starting assumption that AI only helps when paired with structured training.