In our latest bestseller, “Breakpoint: The Crisis of the Middle Class & The Future of Work” we provide evidence that an Indian university degree neither increases the pay of an Indian graduate (in fact, it does the opposite) nor the probability of employment. In fact, the unemployment rate for graduates in India is 10x that of illiterates. Indian employers realised many years ago that most Indian degrees have low value. In this blog for the LSE, Marvin Starominski-Uehara says that the same fate might befall universities around the world.
Marvin Starominski-Uehara is an interdisciplinary scholar, AI literacy consultant and entrepreneur. He writes: “AI use among university students is widespread. A major international survey cited in Stanford University’s 2026 AI Index Report found that 80 per cent of students across 15 countries now use generative AI to help with their studies, double the figure from 2023….A survey by the Digital Education Council found that 58 per cent of students said they lacked sufficient knowledge and skills around AI.”
Dr Starominski-Uehara goes on to explains how AI is degrading the value of a university degree in three fundamental ways. Firstly, students who use AI to produce work and then get high grades (because their professors cannot detect that AI was used to produce that work) feel that they are learning when actually they aren’t. “When a student submits an assignment, they are typically given a marking rubric, a list of the criteria that will be used to grade their work. When a student feeds that rubric, along with all the assignment instructions, into an AI system, the system produces a response calibrated to every criterion on the list.
Research observing this in practice found that AI-generated submissions received an A grade or above and ranked first among submitted work in 80 per cent of cases. Graduate teaching assistants correctly identified only half of these submissions as AI-generated.
The work met the stated standards, and often exceeded them, while containing no original thinking, no engagement with the material and no learning.
For a student who completes this process and receives a high grade, there is no subjective experience of having done anything wrong. They followed the instructions, used an available tool and got a result that the institution accepted.”
Secondly, the faculty believe that they are providing learning and education to students when actually the students are not learning much because they are using AI to churn out assessments. “….many faculty, particularly in the humanities and social sciences, have limited familiarity with how AI language models work. A global survey found that while 61 percent of faculty had used AI in their teaching, 88 per cent did so only minimally. A faculty member who rarely uses these tools is not well-placed to spot their outputs.
A faculty member who cannot distinguish AI-generated from student-authored text is not able to assess whether submitted work reflects genuine intellectual effort….
In response to growing AI use, many universities have turned to automated detection tools, software that analyses text and flags it as potentially AI-generated. These tools are being used to make high-stakes decisions about students’ academic records. They are not reliable.
One study found a 94 percent probability that AI-generated submissions would pass through a real examination setting without being detected at all. A review of 14 detection tools found accuracy rates ranging from 33 to 81 percent, a range so wide that, at its lower end, it is no better than chance.”
Thirdly, in due course as employers discover that the student who has graduated from university has not learnt advanced thinking skills, not only will the university degree get degraded, the university itself loses its reputation. “Students leave university without the skills their degree is supposed to certify. Employers and graduate programmes who take those credentials at face value discover the gap. Over time, the reputation of the credential itself is diminished, not just for those who used AI dishonestly, but for everyone who holds a degree from that institution.
For teaching staff, the damage is more personal. Faculty who understood what was happening, and were given no institutional means to address it, are left defending a credentialing function the institution no longer effectively supports. Researchers describe this as an erosion not just of learning, but of the coherence of the credentialing system itself…”
If you want to read our other published material, please visit https://marcellus.in/blog/
Note: The above material is neither investment research, nor financial advice. Marcellus does not seek payment for or business from this publication in any shape or form. The information provided is intended for educational purposes only. Marcellus Investment Managers is regulated by the Securities and Exchange Board of India (SEBI) and is also an FME (Non-Retail) with the International Financial Services Centres Authority (IFSCA) as a provider of Portfolio Management Services. Additionally, Marcellus is also registered with US Securities and Exchange Commission (“US SEC”) as an Investment Advisor.
