The use of artificial intelligence poses new challenges for knowledge work. The final report of the 18-month research project CAIL (Critical AI Literacy) deals, among other things, with the contradictions and similarities between knowledge work and AI, the added value that AI applications can actually bring, and the central challenges that need to be overcome in the process.

The “CAIL-framework” developed in the project is used to show which factors and basic skills – in particular human interpretative capacity and coping capacity – are crucial when dealing with AI systems in order to be able to meet the challenges constructively.

The benefits of AI ‘are overestimated’

Stefan Strauß, project manager and AI expert at the Institute of Technology Assessment (ITA) of the ÖAW, says: ‘Among other things, the analysis shows that AI can be particularly useful for repetitive, formalized tasks. However, many questions remain unanswered: The benefits of AI are sometimes overestimated and there is a lack of knowledge about the real benefits and limitations of the technologies. There is a gap between using AI as an expert system and using it simply as an add-on. The more abstract the application, the greater the possibility of practical problems.”

According to author Stefan Strauß, there is a great deal of potential in areas with a high level of expertise, such as diagnostic imaging in medicine. In all areas, specialist knowledge, clear purposes and an awareness of the problems associated with the special features of AI-based automation (such as dynamics, complexity and volatility) are crucial factors for constructive use. The assumption that AI will reduce the workload, on the other hand, is difficult to sustain.

Human control is still urgently needed

The high expectations of added value are offset by an underestimated additional workload associated with the use of AI – particularly in terms of quality assurance. Although AI systems can produce results quickly, additional measures are needed to ensure that these results are correct and reliable. ‘Interpretation and verification remain an essential human task, and will become even more important in the future. That is why Critical AI Literacy is becoming a part of modern knowledge work,’ emphasises Strauß.