AI profits from authors' knowledge, leaving them unpaid

Artificial intelligence models are trained on vast amounts of copyrighted texts, including books, articles, and research. These works, created over decades by authors, researchers, and journalists, form the foundation of AI's writing, summarizing, and argumentation capabilities. However, the creators of this knowledge often receive no fair compensation or even clear consent for their work's use in developing AI systems that generate billions in economic value.

“

Artificial intelligence is not just a user of data, but creates economic value from it.

Dr. Jan Kolar rightly points out that AI does not diminish the importance of national libraries. Instead, high-quality data, organized collections, and verified sources become even more critical in the era of large language models. The long-term value of curated collections is increasing. Yet, the discussion often overlooks the human element – the creators whose knowledge fuels these AI advancements.

“

It is not enough to talk only about data accessibility and new research opportunities; the question of how to ensure fair treatment for those whose works were used to develop these systems is equally important.

The question is not whether conversational robots will replace libraries, but whether we can establish a fair system in the age of AI. This system must ensure that authors, researchers, and publishers are appropriately recognized and compensated for the use of their work. Libraries, historically guardians of authorship, should be key stakeholders in shaping the rules for AI development, ensuring that access to knowledge does not translate into creators' works becoming invisible raw material for commercial systems.

“

Access to knowledge is a public good, but this does not mean that creators' products can become mere invisible raw material for training commercial systems.

While the National and University Library organized an event on new written heritage infrastructure, the timing might suggest an underestimation of this issue. Without creators, there would be no libraries or quality data for AI development. The National and University Library must actively defend copyright, not just store works.

“

The question is not whether conversational robots will replace libraries. They probably won't. The question is whether we will know how to establish a fair system in the age of artificial intelligence, in which authors, researchers, and publishers will be appropriately recognized and paid for the use of their works.