By Peter Witten, AAU Communication and Public Affairs 
Illustration:  Søren Mølgaard

The large U.S. AI company Anthropic used pirated books to train its chatbot Claude, prompting a group of authors to sue the company. Anthropic settled the case and will pay up to 1.5 billion dollars - nearly 9.5 billion Danish kroner- to the authors of the pirated books.

Tenure track Assistant Professor Jeanette Falk from the Department of Computer Science at Aalborg University researches generative AI and how it affects humans. She hopes to see more similar rulings.

Gross exploitation

"It’s good that there are significant consequences for the AI company’s gross exploitation of pirated books. I hope we’ll see more cases like this. But I also believe that stronger enforcement of international EU legislation is needed before we see any real change in these companies’ practices."

"The way these AI companies have argued for ‘fair use’ of these materials seems extremely unfair to the creators of the material," says Jeanette Falk.

Human creativity

She believes that training chatbots on other people’s texts and artworks could have negative consequences for human creativity.

"Authors and artists have often spent many years developing their talents and creating the works that these language models train on. I am concerned about what this means for the development of authentic, creative talent in the future. If a chatbot can instantly generate something average and relatively ‘good,’ why should people invest the enormous resources - time and energy - needed to become skilled and creative enough within a field to produce something truly original?"

Significant consequences

Jeanette Falk hopes that the increased attention on chatbots’ use of, for example, authors’ texts without permission will lead to more significant consequences, as in the case of Anthropic.

"I hope that legislation will eventually catch up with technological development and help create fair conditions," she says.