Attended "Impact of Design on Human Decision-Making in AI Systems with Sofie Nabseth".
Interesting thoughts from the talk:
Interesting thoughts from the talk:
- Consider having layers of transparency of your ML system (ie not all users will care about understanding how your ML system works).. Might be useful to point users to a blog post about the specifics behind the prediction algorithm...
- Persuasive Design as a framework for design was discussed... the talk didn't really discuss how Persuasive Design applied to MLUX... however one of the key ideas behind Persuasive Design is the principle of Liking. This seems like a valuable way of understanding why explaining the reason why a recommendation is given (ie "listen to this song because you liked Backstreet Boys"). We want to do something because we know we liked something else (which is similar to this).