The focus will be on why HCI has failed to transfer insights to AI. What does HCI ‘know’ that AI seems to need to find out again, or reinvent? Is it too late or does HCI still have insights to offer to AI?What are they?

With that premise, this leaves interesting meta-questions:

1. Does HCI not have good enough models and what sorts of models would be relevant?
2. Does HCI say “context matters” too much without being able to deliver on what context is?
3. Are AI researchers like physicists, who need to invent everything on their own?
4. Do we not have the right insights from HCI and can we offer them to AI?

Tentative schedule

Minutes 0-30: A couple of short State-of-the-Art talks

Minutes 30-60: The panel kick-off addressing the above questions – Panel members will be recruited in the community (and ourselves).

Minutes 60-80: Group talks over coffee

Minutes 80-90: The panel takes a few questions and summarizes

Target audience

People from research and industry interested in human use of AI (and the challenges that follow).

Organizers

Susanne Bødker, Aarhus University (bodker@cs.au.dk) with Kasper Hornbæk, Pernille Bjørn and Serge Belongie, University of Copenhagen.