“Beauty is only skin deep, but ugly goes clean to the bone.” — Dorothy Parker
Wow there are a lot of non-helpful definitions of interaction design out there on the web! (Wikipedia, for example) Too bad, because a lot of people don’t know what it is, and a surprising number of companies don’t practice it.
Interaction design is the design of the collaboration between a human and an artificial system. Done well, it minimizes the human’s effort while maximizing their enjoyment and offloading to the computer everything else.
Intelligent systems have the potential to be good at this. Intelligent interaction design realizes that the computer system can behave intelligently, and thus the surface design of the system can be completely simple.
Those of you who used the web extensively before Google came on the scene had to navigate a series of categories — much like using the old style physical library card catalogs.
Google’s famous one box search is the epitome of intelligent interaction design. The visual design itself is hardly worth mentioning: a white screen, a text box. When we all first saw it we were stunned. But it was perfect: its goal was to offload almost every bit of work to the computer. As a user, I just type a word, or a few letters of a word, or whatever I can remember about a topic and the computer does almost everything else. Especially now that your experience is personalized and Google gives you straight up answers and not just links.
And many intelligent systems designers tried to improve on it. Famously Cuil tried to help users by disambiguating their search queries. But nothing works better than asking a user to do almost nothing. Cuil demonstrated one wrong way to use an intelligent system.
A good way to get started with intelligent interaction design is to understand the user’s relevant job (what they are trying to do when they use your product) and to minimize the effort needed to accomplish it successfully. Check out work-centered design or use-centered too. (Note, that’s use-centered, not user-centered.) Rather than think only of the user, or the design of the product, you explicitly focus on what it is the user is trying to do. You can model the work itself and build that model into the system. You can quantify the user’s success or failure when they use each possible design. You can explicitly create designs that measurably reduce the amount of effort, pain, or time it takes to accomplish a task.
After all, helping users do stuff is what intelligence is for.