Anyone can interact with ChatGPT via an internet browser. You type in questions or commands and ChatGPT responds (to almost anything). Within five days of its release, 1 million people had signed up to use it. The claim is that ChatGPT can explain, program, and argue with humanlike efficiency.

Wessels, who is based at Kiel University of Applied Sciences, is amazed by the technology. “It’s like entering another world,” she said.

Mike Sharples, an emeritus professor at the Open University in the United Kingdom, has seen a few such “major breakthroughs” during his 40–year career in artificial intelligence—including a precursor to ChatGPT called GPT-3.

Sharples warns that “GPT democratises plagiarism”. Some students have been open about using the technology to help them write essays in perfect academic language. It is like free ghostwriting for everyone, but there are examples that show ChatGPT’s responses can be factually incorrect.

ChatGPT can be used to write research papers. Sharples had the AI generate a scientific article that he said “could pass a first academic review”.

This worries Wessels, who said universities were in danger of being left behind. On the one hand, there’s the software industry, which is developing ever-more-powerful AI systems. And, on the other hand, there are the students, who are learning how to use AI in education faster than their teachers can keep up with the developments.

Students often learn about new AI technology rapidly in real time via social media and are keen to try new methods, while some academic staff and professors may be slower on the uptake, or set in their ways.

Wessels sees a “possible horror scenario” in which unsuspecting professors might think that they have done an excellent job of teaching if all their students submit error-free assignments—when, in fact, it is all down to ChatGPT or a similar system.