Speaker
Péter Kosztyó
(Eötvös Loránd University)
Description
Artificial intelligence, including large language models (LLMs), is transforming education and daily life. Many online texts are AI-generated without public awareness, and students increasingly use tools like ChatGPT. Our study explored whether 372 high school students could identify if a scientific text was AI- or human-written and examined traits aiding recognition. Results showed students struggled but performed slightly above chance. Factors like gender, prior knowledge, language proficiency, or science skills didn’t improve accuracy. The only positive correlation was familiarity with ChatGPT’s language. This highlights the need to educate students on recognizing AI-generated content.
| Education level | Age 15-18 (Secondary education) |
|---|---|
| Physics topic | Other |
| Research focus | Artifical Intelligence |
| Research method | Analytic Physics Education Research (Quantitative research) |
| Organizing preference criteria | Research focus |
Author
Péter Kosztyó
(Eötvös Loránd University)
