Researchers from Germany’s University of Passua have determined that essays generated by chatbots like ChatGPT were better than those written by human secondary school students.

To reach this conclusion, the researchers compared text produced by artificial intelligence (AI)-based language models — specifically ChatGPT 4 and its predecessor ChatGPT 3 — to essays written by secondary school students.

Source: University of PassauSource: University of Passau

Both sets of essays, according to the researchers, were measured according to criteria established by the Ministry of Education of Lower Saxony. Both sets of content were reportedly evaluated on criteria topic, completeness, logic, vocabulary, complexity and language mastery, among other features.

With a scoring system of 0 to 6 — wherein 0 is the lowest score and 6 is the highest — the researchers reported that the AI-generated text outperformed student written essays, with ChatGPT 4 scoring 5.25 and ChatGPT 3 scoring 5.03 while the secondary school students earned an average score of 3.9.

An article detailing the findings, “A large-scale comparison of human-written versus ChatGPT-generated essays,” appears in the journal Scientific Reports.

To contact the author of this article, email mdonlon@globalspec.com