In 2026, Ivy League universities such as Brown and Harvard returned to in-person exams to combat AI-assisted cheating. The result? A 50% drop in performance. What went wrong, and is this the only solution?
The Ivy League grading crisis: data that surprised everyone
In the 2026 summer semester, Brown University recorded a 47% drop in GPA compared to the previous semester. Harvard and Yale also reported similar trends – 42% and a record-low average of 2.9 on a 4.0 scale, respectively. These figures are not just statistics – they are evidence of a deep crisis that has shaken the American higher education system.
The sharpest declines hit the humanities, where written assignments were previously the dominant form of assessment. For example, at Brown University, English literature saw a 52% drop. STEM fields, while less affected, also felt the shift – computer science lost 38% of its points. What was behind these numbers?
"The introduction of in-person exams in the 2026 summer semester was intended to restore the integrity of the grading process. The average GPA for this semester was 2.8 (on a 4.0 scale), compared to 3.4 in the previous semester. This difference reflects both the effectiveness of the new rules and the need to adapt teaching methods to the requirements of independent work."
How were students cheating with AI?
Before 2026, AI-assisted cheating became widespread. Students used tools that allowed them to bypass traditional assessment methods:
- Essay generators: Tools like Jasper or Copy.AI created homework in minutes. The finished texts were then paraphrased using quillbot to avoid detection.
- AI in programming: In computer science departments, github Copilot and chatgpt helped solve algorithmic tasks. In 2025, it was discovered at Stanford that 60% of programming homework contained fragments generated by AI.
- Test solving: Tools like Wolfram Alpha or Socratic by Google were used for cheating during remote exams.
- Deepfakes and fake IDs: In 2025, cases were recorded where students hired individuals to take online exams using synthetic avatars.
Universities tried to combat this phenomenon by introducing anti-plagiarism software or stylometric analysis, but it proved insufficient. AI-generated texts were not copied from existing sources, which made them difficult to detect.
Why did the Ivy League return to in-person exams?
The decision to return to traditional exam formats was not accidental. In February 2026, the American Council on Education published a report titled "The AI Cheating Crisis," which revealed alarming data:
- 78% of students admitted to using AI for homework.
- 62% of faculty members reported that they were unable to distinguish between student-written work and AI-generated content.
Media and parental pressure also played a role. Articles such as "Harvard’s Honor Code in The Age of AI" drew attention to the credibility crisis of grades. Universities were faced with a choice: either accept the cheating or implement drastic changes.
Brown University, Harvard, and Yale chose the latter. Starting from the 2026 summer semester, final exams had to be taken in person, without access to electronic devices. Exam halls were monitored, and students were randomly assigned seats.
Academic community reactions: from protests to support
The changes were not met with universal acceptance. The academic community split into several camps:
Students: between strikes and support
At Brown University, students organized a strike in May 2026, arguing that the change was implemented too quickly and without consultation. A survey conducted by the Harvard Undergraduate Association showed that 65% of students felt unprepared for in-person exams after years of remote learning.
On the other hand, some students – especially in STEM fields – supported the changes. They argued that AI "kills the motivation to learn" and that returning to traditional methods is necessary.
Faculty: criticism and alternative solutions
Professor Jonathan Zimmerman of the University of Pennsylvania argued in the article "We Failed Our Students" that universities had ignored the problem for too long and were now punishing students for their own negligence. Conversely, Professor Stuart Russell of UC Berkeley stated that in-person exams are a "necessary step backward to move forward."
Some faculty members proposed alternative solutions, such as:
- Group projects instead of individual assignments.
- Practical exams, e.g., labs or simulations.
- Using AI for personalized learning, but not for assessment.
Educational experts: the future of assessment
UNESCO, in its report "The Future of Assessment in The AI Era," recommends moving away from assessment based on factual knowledge toward critical thinking and creativity skills. It proposes project-based exams and portfolios as an alternative to traditional tests.
Sal Khan, founder of Khan Academy, stated in an interview with edsurge:
"In-person exams are a band-aid on a wound. The real solution lies in rethinking what we truly expect from education. Do we want students to memorize facts, or to be able to analyze them and create something new?"
International comparisons: how are other countries coping?
The Ivy League crisis is not isolated, but reactions in other countries vary significantly:
- United Kingdom: Universities such as Oxford and Cambridge have been using in-person exams since 2023. In 2026, Oxford introduced mandatory AI ethics training for students.
- Germany: Technische Universität München is testing AI-assisted exams – students can use tools, but must document the creation process of their work.
- China: Universities use exams with cameras and AI monitoring, but stress levels among students have increased.
- Japan: The University of Tokyo introduced oral exams as a standard form of assessment in humanities departments.
In Poland, the Central Examination Board (CKE) has strengthened supervision over matriculation exams, but no drops in results comparable to the Ivy League have been recorded.
Long-term consequences: what’s next for education?
The Ivy League crisis could have far-reaching effects on the future of education:
Changes in curricula
- Brown University: Starting in 2027, it is introducing a mandatory course titled "Critical Thinking in the Age of AI."
- Harvard: Plans to reduce homework in favor of group projects and seminar discussions.
- MIT: Is testing "open-book" exams using AI – students can use tools but must explain the process of arriving at the solution.
New technologies in education
- Blockchain: The learningchain project, developed by Stanford and Berkeley, aims to verify the authenticity of student work.
- AI as a teacher: Tools like Khanmigo or Duolingo Max are being tested as learning assistants.
- Biometrics: Universities are considering handwriting analysis and voice recognition as methods for identity verification during online exams.
Alternative forms of education
- Growth in the popularity of coding bootcamps and practical courses.
- Micro-credentials: Certificates for specific skills are gaining importance at the expense of traditional degrees.
Limitations and challenges: what could have influenced the results?
The 50% drop in results may not be solely the result of AI-assisted cheating. Other factors that could have influenced this data include:
- Stress of returning to in-person learning: After years of remote learning, students may have had difficulty adapting to traditional exams.
- Curriculum changes: Some universities introduced new subjects or teaching methods in 2026.
- "Novelty" effect: Students may not have been accustomed to the new rules, which temporarily lowered results.
Comparative data is also missing – it is unknown what the results would have looked like if in-person exams had been introduced gradually rather than suddenly.
Summary: are traditional exams the only solution?
The Ivy League crisis shows that education is facing a serious challenge. The return to in-person exams was an attempt to restore grading integrity, but it also brought unexpected consequences. A 50% drop in results is a signal that the education system must change – but are traditional methods the only answer?
Alternative solutions, such as project-based exams, portfolios, or using AI for personalized learning, may prove more effective in the long run. One thing is certain: education in 2026 must find a balance between modern technologies and traditional methods to prepare students for the future.
If you are interested in how AI is changing other areas of life, read our post about breakthroughs in artificial intelligence in 2026.
Comments