4 thoughts on the intersection of UX and AI

photo by Greg Rakozy (Unsplash)

I recently had the opportunity to take part in a panel discussion on the intersection of UX and AI. It was really fun. I got to meet some really smart people (Alex Candelas, Jon Peterson, Tcheilly Nunes) and we had a really interesting and engaging discussion about how AI is impacting the world of product design, and by extension, the digital world we all inhabit. You can watch the panel discussion here.

Designers absolutely have a huge role to play in AI — how it will be developed, productized, and introduced into user experiences. From that panel, I wanted to share 4 things I’m thinking about.

#1: AI will change everything. But the designer’s job will not change.

Everyone is predicting that AI will change everything in the next few years. It’s already upended so much, finding applications in healthcare, finance, retail, and media. Companies everywhere are rushing to add AI capabilities (often before figuring out what/why they need AI for their business). Data scientists and machine learning engineers are in high demand as LLM are developed and vast amounts of data are being crunched to power these systems. It’s all engineering-intensive and technical and certainly opaque to the uninitiated. How can a designer make a meaningful contribution to all of this?

By doing their job. A designer’s job isn’t to make something look aesthetically pleasing, stylized or beautiful. A designer’s job is to understand the user, understand the problems they face and the things they hope to accomplish. A designer’s job is to define the problem (correctly) and design solutions that address that problem in an effective and desirable way. And so it is with AI. The designer must figure out how AI can be used to solve real user needs (or not when AI isn’t the right solution). Everyone’s integrating AI into everything with little to no innovative ideas of how to apply it. Designers are needed to craft solutions that matter to users. They’re not designing chatbots. They’re designing experiences.

#2: AI is a tool, not the solution.

AI is a very powerful technology that seems to just be getting started. Its power and capabilities seem to grow exponentially every year. Companies are racing to add AI to everything, stand up some product enhanced by AI (and then market the hell out of the fact that they “have AI”).

Users don’t care about this in any meaningful way. Maybe they’ll try your product out of curiosity, but this is not likely to lead towards regular usage. AI is not the point for users because AI isn’t the solution to the problem they have or task they need to accomplish. AI can possibly help in addressing that need, but as a means to an end.

It’s the designer’s responsibility to figure out how to integrate AI into a product or experience that is effective and desirable for the user.

#3: AI should be invisible, not human.

ChatGPT can be fun to interact with. Alexa can be genuinely helpful in certain situations. Siri, not so much. All this effort to create AI that is more human, that you can chat with feels to me to be a wrong path, possibly a dead end. AI as it currently exists cannot become “more human.”

AI is trained on large data sets. It identifies patterns in those data sets and then offers up what it predicts will be the next sequence in that pattern. It cannot learn to be empathetic or creative or curious. Any behaviors it exhibits are merely programmed responses that a developer created to make AI appear more human. But it’s not real. For me at least, every time an AI-powered product provides a “human-like response” it feels fake and insincere. The contours of the response immediately become rough and jaggest, the illusion dissipates.

AI seems to be most effective when it’s not noticed, when it can’t be seen but the results it powers are desired. When Spotify plays a string of songs you hadn’t heard before but really love. When Netflix recommends a movie that turns out, you really like. When Google answers the question you typed into the search bar at the top of the results page rather than have you click into a page and find the result.

When AI creates a seemingly auto-magical experience — the answer or outcome you desire without having to go through the steps to achieve it yourself — the technology seems to be most effective. These should be the types of experiences we’re trying to build with AI.

#4: Be curious.

This is probably the most important thought I have on this topic.

AI is a new technology and it will impact and transform so many industries and experiences. It will be used to create unique experiences that we can’t even imagine today. So start getting familiar with it.

Play around with chatbots and image generators. Use ChatGPT to generate a list of ideas for blog posts. Have it create a first draft of an important email you need to send but have been putting off. Use it to generate a list of user interview questions and then refine them. Experiment with different prompts to get a better understanding of what’s effective. Read interesting Medium and LinkedIn posts about AI.

As much as you can, immerse yourself in AI tools and the discussion around it. By becoming familiar with AI, you’ll gain a better understanding of its capabilities and limitations. The edges of what is possible and what is not (today). This will position you to be able to leverage AI in the future experiences you will design and build.

Discussion on the intersection of UX and AI

Last Thursday I got an opportunity to be part of a panel on the intersection of UX and AI. It was a really fun experience. We covered a lot. Great points made by lots of people. You can watch the video here:

Panel discussion on UX and AI

Better things than chatbots

In the beginning, we had to learn it’s language to interact with it.

Computers required users to adapt to them in order to make it do something. Eventually we developed the graphical user interface to hide the programming-speak and but this is still mostly an illusion as we’re still constrained to think like a computer and structure our interactions in a way that makes sense to a computer (tasks need to be done sequentially and in order, meaning and intent don’t factor into it, etc.). But we are fast approaching an inflection point where this dynamic is reversed and the computer will respond to human natural-language input. We’ll be able to speak naturally and be understood by technology.

This post is about AI.

As AI becomes ubiquitous, interfaces everywhere are being redesigned to center on chat. Conversation design/UI is being rolled out everywhere. Companies are asking themselves, “Can we use AI to make this a better experience for users?”* In my role as design leader I regularly have internal discussions about deploying chatbots as a new feature enhancement, at Drizly (RIP) and Liberty Mutual. 

* actually, the questions are more accurately: “let’s use AI to improve the experience…” rarely is it “will AI improve this experience?” but I digress.

AI is going to be transformative. And it is coming for every sector. Consider customer service:

“Gartner predicts that by 2025, 80% of customer service and support organizations will be applying generative AI technology in some form to improve agent productivity and customer experience (CX).”

— 2023 Gartner study

I think it’s fair to ask however: “Will users find a customer service experience powered by AI better?” More succinctly: “Do users want to talk to a chatbot?”

Generally, users don’t want to talk to chatbots. Today. Chatbots will get better over time and the friction and sour feelings they create today will minimize (though probably not disappear). Will chatbots in the future pass the Turing test? I have no idea (and I appreciate it when companies don’t try, such as Slack).

Whatever the answer, I think chatbots are the least interesting application of AI from a user experience perspective.

Better UX through AI

Yes, AI can make a chatbot interesting. Go play with ChatGPT  for a bit. It’s fun and with some practice (i.e. learning to think like the computer) you can get it to spit out some really creative and interesting content. But that’s the end of the user experience, isn’t it?

I think far more compelling application of AI will be when it truly adapts to humans, anticipating and solving user needs. In this dynamic, there’s no need for the user to tell the computer what they want (in whatever language) because the computer is already working to address the need.

• Imagine AI rescheduling a doctor’s appointment because of a conflict in your schedule that has come up?

• Imagine booking travel to a new city and AI building an itinerary of activities and restaurants based on you that consults your schedule without being asked?

• I recently shared this post on generative UI where AI transform’s an app’s interface to suit the specific needs of a user based on their behavior and interests. AI to transform every interface so it’s adapted to you sounds pretty compelling.

• At Drizly we discussed an AI feature that would take a results page of various wines that would select specific bottles and generate a brief description, all based on the user’s taste preference, finding relevancy through the noise (unfortunately we never got around to building it).

• Seven years ago I wrote about the Starbucks app anticipating my morning order and placing it without me having to tell it to (I didn’t mention AI, but this theme holds - technology working for us and not the other way around).

IN CONCLUSION

Each of these AI examples are all relatively possible today, and I think offer a far more compelling user experience than a chatbot. With AI, let’s focus on user outcomes and design interactions that will provide rewarding experiences for real people.