Testing Intelligence: The Difficulty of Evaluating AI Systems - open expert meeting by SwissNLP x data innovation alliance

SwissNLP, a non-profit association, regularly hosts expert meetings for its members - organised in partnership with the data innovation alliance. For Swiss AI Weeks, we're opening the doors to everyone for a special session!

We will dive into the topic of evaluating AI systems. As AI rapidly transforms industries, businesses are under pressure to adopt systems that are not only powerful but also reliable, fair, and aligned with real-world needs. Evaluation is the key to ensuring AI systems deliver measurable value - whether it's improving customer experience, driving efficiency, or supporting critical decisions.

But evaluating AI is not straightforward. Unlike traditional software, AI systems learn from data, which makes their behavior unpredictable and context-dependent. Performance can vary across user groups, environments, and use cases - and standard metrics often fail to capture what matters to businesses.

At this event, we’ll hear from three speakers why evaluation is both a strategic necessity and a technical challenge - and how it can be tackled.

Stick around afterward for an apéro and informal networking!