Nielson Norman Group, Therese Fessenden, Jakob Nielson, Nielson Norman Group
Topics: People, Experience, AI, Autonomous Systems, Big Data
This summary is a breakdown of a question asked about the relationship between artificial intelligence and the user experience and how UX people can get more involved with AI. In many ways, first of all, we have seen that history is repeating itself many times in the past, we have seen new technologies come around, and then being pursued their own sake. That’s also a lot of what we’ve seen with AI, is that it’s now possible to do things that did not use to be possible, and therefore they are being done, which is a bad reason. It seems that there is often a little bit of a UX kind of chasing after the train a little bit, trying to catch up with these developments, and that’s unfortunate. If it’s driven by human needs, and we can solve these human needs better by having a smarter computer that can do certain things it didn’t use to do, well that’s good. For example, many of these speech interfaces, that are AI to these intelligent agents, often have a lot of visibility problems which we have certainly found when we have studied them. One of the things that actually will, we believe backfire there is if we give people exposure to substandard products and their experience is hyped a lot but it doesn’t help. Then they’re going to be turned off from those things and it’s going to be very hard to get them to resemble or retry, or do it again, do-over, when a new and better product later. So, designing for what it does do for people is super, super important, and somewhat neglected in many AI projects. What we do think is that there are a lot of things in which we can enhance UX by taking advantage of less precise interaction styles. That requires less of that engineering mindset that most people don’t have that traditional UX interfaces used to require; also, to the extent that we can be a little bit fussier in our specifications and the computer has more of an ability to try and interpret what we want. Doing what I want, not doing what I say is really what we would like the computers to do. So, if we can make that happen, we can enhance user experience; and it can be in smaller ways such as auto-aligning when making a slide in PowerPoint, or it can be in much more advanced ways. Even from very pragmatic things that people experience has done in user interviews then getting them transcribed by a computer rather than sitting and having to do it alone. Well, if it’s cheaper to do, easier to do, we know it will be done more. That’s one of the basic usability lessons. If something is easier to do, people will do it. If it’s harder to do, people will not do it. In many, many aspects of design as well the computer can come up with some drafts of the design and the designer can more, kind of adjust it rather than having to do everything other than the pixels from scratch. So, there are many, many ways in which AI can also help us, which means that we will become more productive which again means that we will be used more because our results will be better and have more business value and that’s one of how we’ll grow the field.