Universities across Spain are fundamentally redesigning how they test student knowledge, moving away from written assignments and toward real-time oral exams and live project defenses. The shift reflects a stark reality: with 89% of Spanish university students using artificial intelligence tools like ChatGPT regularly in their studies, traditional assessment methods no longer reliably measure what students have actually learned.

The problem is straightforward but urgent. When a student submits polished, error-free work, instructors can no longer assume the student wrote it. ChatGPT and similar tools can generate grammatically correct, well-structured answers in seconds. Many AI-detection systems are unreliable, making it nearly impossible to catch sophisticated use without direct evidence. The challenge has become so widespread that university ombudsman offices, which protect student and staff rights, are now fielding what one administrator called a "flood of enquiries" about AI-related academic conflicts.

Why Traditional Exams Are No Longer Trustworthy?

The University of Almería recently hosted a conference on artificial intelligence after receiving numerous complaints about conflicts caused by AI use in assessments. During the meeting, Vice-Rector Maribel Ramírez warned that AI "is rapidly transforming multiple areas of university life" and pointed to challenges including data protection, transparency, and algorithmic bias.

"Universities are already receiving many requests for intervention related to AI in assessment and other academic areas, which makes it necessary to look for a common response," explained Bernardo Claros, university ombudsman.

Bernardo Claros, University Ombudsman

The debate has expanded far beyond simple cheating concerns. It now encompasses ethical, legal, and institutional questions about how to fairly evaluate student competency in an age where machines can produce academically credible work instantly. Most Spanish universities have already detected students submitting AI-generated answers even during supervised exams, according to a report by the CYD Foundation.

How Universities Are Redesigning Assessment Methods

In response, educational institutions are implementing several new assessment strategies:

• Oral Exams: Students must explain concepts, argue their positions, and answer follow-up questions in real time without external support, making it nearly impossible to rely on pre-generated AI content.
• Live Project Defenses: Submitting a document is no longer sufficient; students must justify their work, explain their reasoning, and defend their conclusions against direct questioning.
• Increased In-Person Testing: Many universities are raising the weight of supervised classroom tests to reduce reliance on take-home assignments where AI use is hardest to monitor.
• Real-Time Problem Solving: Assessments increasingly require students to solve novel problems on the spot, demonstrating critical thinking rather than knowledge retrieval.

These changes represent more than a tactical adjustment to exams. They signal a structural shift in how education defines learning itself. For decades, the system valued memorization and the ability to write polished assignments. Today, both of those tasks can be performed by artificial intelligence.

What Skills Are Becoming More Important?

As written work becomes less reliable as a measure of knowledge, universities are placing greater emphasis on skills that machines cannot easily replicate or simulate in real time. These include critical thinking, the ability to articulate and defend ideas verbally, and solving complex problems under pressure. Students must now demonstrate not just what they know, but how they think.

Interestingly, students themselves recognize both the risks and benefits of AI in education. A majority of Spanish university students believe ChatGPT can improve their academic performance, according to the CYD Foundation report. Universities also see potential in using AI to personalize teaching, facilitate research, and improve lesson preparation.

However, there is a darker concern: students may become overly dependent on technology, reduce their own effort, or develop only superficial understanding of material. The challenge for educators is to harness AI's benefits while preventing it from becoming a substitute for genuine learning.

The Spanish experience reflects a global pattern. As generative AI tools become more sophisticated and accessible, educational institutions worldwide face the same fundamental question: how do you assess knowledge when a machine can produce correct answers instantly? The answer emerging from Spanish universities suggests that the future of education may depend less on what students can write and more on what they can think, explain, and defend in real time.