From Turing Test to the Green Campus – Current Developments in AI in Education

We’ve compiled the ten most interesting news stories into a single, continuous narrative. We invite you to explore this story, which connects (sometimes seemingly unrelated) events.

Imagine walking into a lecture hall on a Monday morning and finding unusual news waiting for you there: GPT-4.5 fooled as many as 73% of trained evaluators in an authentic Turing test, who mistook it for a human. The University of San Diego study isn’t just a media sensation; it provides solid confirmation that artificial intelligence can successfully mimic human behavior to the point of confusing human interlocutors. Of course, this is only the case when it is carefully guided by appropriate instructions (so-called “persona prompts”). However, this study also raises the question of how to distinguish genuine student work from a skillfully crafted text generated by artificial intelligence.

News of this kind spreads quickly. British experts suggest that in future research and academic work, AI algorithms will also be able to help read through thousands of reports and co-create research content, where AI tools will provide us with greater linguistic confidence, faster writing, and clearer argumentation. But the very news about the Turing test shows that blind trust in the models’ judgment can be a double-edged sword: what saves us time can also conceal biases.

Meanwhile, a heated debate is raging in the halls of universities and colleges, as as many as 42% of students already use generative AI on a daily or weekly basis, according to the Time for Class of 2025 report. And so it is time to consider how to integrate this technology into the educational process at the systemic level while ensuring its ethical, safe, and unbiased use.

A study in the journal Nature adds another piece to the mosaic of this story: moderate and guided use of tools such as ChatGPT can contribute to a statistical increase in grades and the development of higher-order cognitive skills, while uncritical reliance on such tools in the learning process offers no benefits.

EDUCASE also responds to this in its article “In the Room Where It Happens”, urging universities to include students, educators, and IT experts when drafting guidelines for the use of AI in education. After all, this is the only way we’ll get guidelines that are understandable and will actually be adopted in practice.

Of course, educators aren’t waiting for bureaucracy :). EdSurge reports on a trend where, instead of traditional assignments like writing essays, students create prototypes, video journals, or solutions to real-life problems. They are engaged in activities where students’ contextual knowledge cannot be effectively replaced by the use of genAI, as is also advocated at XJTLU. In other words, tasks in which AI is not a shortcut to the result, but rather a tool for co-creation.

In one of the recent studies, we also see a concrete example of designing learning activities that combine the use of AI and virtual reality environments. In a simulation where virtual mentors adapt the scenario in real time, students solve ethical dilemmas related to sustainable development and immediately receive personalized feedback. This example encourages us to consider how we, too, can push the boundaries of teaching and implement activities we hadn’t even imagined until now.

These optimistic educational stories are somewhat overshadowed by this news, which predicts that the growth of data centers—primarily due to the development of AI—will drive electricity consumption to a historic high in the coming years. Scientists therefore caution that we must also consider this aspect of AI tool usage and carefully select the situations in which to use them. To conclude our weekly AI story, here’s some encouraging news from Germany, where researchers are already working on a prototype that could reduce data center energy consumption by up to 90%.

We’ve seen that, in the past week alone, AI has fooled people, improved learning outcomes, and transformed teaching and assessment methods—all while placing a burden on our planet. The common thread running through all the news in our story is the need for responsible co-creation.

• Let’s reexamine existing methods of assessing knowledge.
• Let’s involve students in shaping policies.
• Let’s redesign assignments so that AI adds value without providing the final product.
• Let’s measure our carbon footprint and support the development of more sustainable models.

If we coordinate these steps, the coming weeks will bring stories in which AI enhances university excellence without losing our human touch or harming the planet.

Authors: Sanja Jedrinović, Mateja Bevčič, Eva Kern Nanut, Eva Škraba, University of Ljubljana Center for the use of ICT in pedagogical process