AI in Libraries: How Libraries can Live Up to Their Responsibilities

we were talking to Dr Argie Kasprzik (ZBW)

The use of artificial intelligence (AI), especially large language models (LLMs), has great potential but also a range of severe potential drawbacks. Among the undesirable consequences can be, for example, considerable costs for the environment and climate, for society, information quality and cognitive skills, as well as possible negative economic and psychological effects. Conscious use of AI is therefore essential, especially for libraries, given their social role and exemplary function. Dr Argie Kasprzik has dealt with this issue in depth in their new article “Die KI(rche) im Dorf lassen” (German, phrase meaning: Keep a sense of proportion with regard to AI). A brief insight:

What are the dangers of artificial intelligence for Open Science and for the quality of information?

Currently, deep learning models and services provided by a handful of large tech companies dominate the market. These are generally not Open Source in the strict sense. In this respect, they contradict the reproducibility principle of Open Science and lead to a dependency on a few providers, with predictably unfavourable consequences. Furthermore, these solutions are being pushed out into the market with highspeed leading to all kinds of immature services and unmarked AI-generated content of dubious quality on the internet (“AI slop”) – more than could ever have been produced by humans. The result is an increasingly messy information landscape, which in turn will also impact future models that are being trained on it.

What is the most important thing libraries should do in terms of the responsible use of artificial intelligence?

Standing up for the development and use of genuine Open Source models in the information infrastructure sector, for the common good and free from the influence of profit-driven interests. To achieve this, appropriate conditions must be created on a permanent basis, both at the legal and political level and in terms of the necessary infrastructure and data sets. Responsible use of AI can only succeed if institutions create wide-ranging coalitions in order to jointly create solutions that are superior to commercial models in terms of transparency and openness.

Further background information and recommendations from Dr Kasprzik can be found in the article in the journal “o-bib – Das offene Bibliotheksjournal” (The Open Library Journal):

This might also interest you:

• Die KI(rche) im Dorf lassen (German, phrase meaning: Keep a sense of proportion with regard to AI)
• Hypertrophic AI – can libraries lead the pushback?
• AI in Academic Libraries, Part 1: Areas of Activity, Big Players and the Automation of Indexing
• Human-Centred AI: First Steps for the Enrichment of Library Work
• AI Inspirations for Libraries: From Study Bots to Conversations with Books