Student Who Used AI for Graduation Thesis Released After Appeal

A student at the University of Amsterdam (UvA) had their graduation thesis invalidated and was ordered to rewrite it. The supervisor raised suspicions of AI use, and the examination committee accepted this conclusion. However, the student appealed with a lawyer present, and the disciplinary action was ultimately nullified due to procedural errors.

One Thursday morning in December 2024, the student sat in a conference room at the Maagdenhuis, UvA's main building, facing four examination committee members. Committee chair Han Peters stated there was "no room for doubt" and presented the following evidence:

• Poor English suddenly transformed into academically excellent English
• Student unable to properly explain their own thesis during oral examination
• Multiple paragraphs flagged as AI-generated by AI detection systems

However, as Peters himself acknowledged, AI use is "difficult to prove," and only "catching them red-handed behind a laptop" provides definitive evidence. After a 20-minute deliberation, the student was cleared due to procedural defects.

Four Out of Five University Students Using AI Chatbots

According to a survey of 1,633 students conducted by Utrecht University, over 80% of students have experience using AI chatbots like ChatGPT for their studies. Students are using AI for purposes such as:

• Creating summaries
• Answering questions
• Brainstorming
• Writing assistance

ChatGPT has become such a routine learning tool among students that it's affectionately called 'Chat.' The problem is that this technology is fundamentally different from calculators, spell checkers, and translation programs.

Generative AI doesn't simply process information—it generates ideas, constructs logic, and writes sentences independently. This directly threatens the core values of university education: developing 'independent thinking,' 'critical reasoning,' and 'creative problem-solving' abilities.

What Has Changed: Before vs. Now

Dutch Universities Face 'Dilemma' in Detecting AI Misconduct

The NRC media outlet conducted a survey targeting humanities and literature departments at the four major Dutch universities (University of Amsterdam, Utrecht University, Leiden University, and Radboud University). These departments were selected because writing is central to the curriculum.

The survey revealed universities facing the following dilemmas:

1. Limitations of Detection Technology AI detection software shows high false positive rates and is difficult to use as legal evidence. Some students have learned methods to evade detection by slightly modifying AI-generated text.

2. Difficulty of Burden of Proof Without "catching them red-handed behind a laptop," securing definitive evidence is difficult. When students claim "I didn't use AI," there's insufficient conclusive evidence to refute this.

3. Confusion in Educational Philosophy Some professors argue "we must change teaching methods to suit the AI era," while others counter that "traditional writing skills are still essential."

4. Crisis of Degree Value Credibility If diplomas cannot distinguish whether they certify students' own competence or their effective use of AI, the social value of degrees themselves may decline.

[AI Analysis] Complete Redesign of University Evaluation Systems Inevitable

The emergence of generative AI poses fundamental questions to university evaluation systems that have been maintained for over 300 years. Universities are likely to evolve in the following directions:

Short-term Response (1-2 years)

• Increase proportion of oral exams and real-time writing tests
• Establish guidelines clearly defining the scope of AI use
• Strengthen one-on-one face-to-face evaluations between professors and students

Medium-term Strategy (3-5 years)

• Include AI utilization ability as an evaluation criterion (e.g., "conducting deeper research using AI as a tool")
• Shift to process-centered evaluation (recording intermediate processes, not just final outputs)
• Introduce blockchain-based learning history tracking systems

Long-term Vision (5+ years)

• Redefine the meaning of university degrees: shift focus from knowledge acquisition to 'problem definition,' 'ethical judgment,' and 'collaboration' abilities
• Establish new academic competency standards for coexistence with AI
• Transition to lifelong learning systems (continuous competency verification rather than one-time degrees)

The Dutch universities' case has significant implications for Korean universities as well. Simply prohibiting AI use won't solve the problem; what's needed is rethinking what education fundamentally means and redesigning evaluation systems from the ground up. Universities in the AI era must evolve into institutions that assess not 'what you know' but 'what you can do' and 'how you judge.'