Lately AI has been used more and more frequently. Just take a look at this graphic showing relative data over time for people searching for “Artificial Intelligence” from Google Trends.
Of course it has been a topic of interest in science fiction for some time (for example, Hal 9000 in the 1968 film 2001: A Space Odyssey), but only more recently there has been strong customer facing examples of businesses making a lot of money with AI. And it should come as no surprise that science fiction often inspires reality.
Examples of AI
So it is not a new term but it is one that is having a real impact on the world. There are numerous examples, but here are just a few that all improve over time:
- Alexa’s voice assistive technology – what information or action does a user want to take?
- Tesla’s self driving cars – what is the optimal and safest way to get a user to a destination?
- Netflix’s predictive algorithms – what show might a user like?
- Nest’s adaptive temperature controls – what temperature is best considering user preferences and the outside weather?
- Google’s predictive text – what is a user wanting to search for before finishing typing their question?
All of these technologies have one thing in common (two actually). They’re all examples of artificial intelligence, of course, and they all include computer interactions with users — also known as Human-Computer Interaction (HCI).
HCI is prime real estate for user experience to help improve how AI interacts with customers. This means that any good UX practitioner should be learning about it just as they should be learning about other emerging technologies Virtual Reality (VR) and Augmented Reality (AR). And not only those professionals, but any company wanting to do AI seriously should be considering the UX aspects of it.
So What?
So what needs to be done? Well, user testing for one, which is a critical aspect of doing UX well. Amazon does this with Mechanical Turk. Interestingly enough, this tool is a good thing to look at if you want to know where the cutting edge of AI is. This is because if the questions asked of Mechanical Turk participants were easily answered by computers, Amazon would likely already be doing it with AI.
One way Amazon uses Mechanical Turk to learn more about their products is by paying users to provide examples of how they’d ask Alexa to do two things at once. She doesn’t do so great with that yet. For example, how might a user ask to check the weather and the news one after the other? Participants of Mechanical Turk are asked to give ways in which they’d ask her to do two actions. The goal of course is to take this testing and design requirements for the system that are more user centric and understand people better.
Another way in which user experience can help is by designing any interaction points with an AI to clearly convey what it is the AI is doing or thinking. For example, with the Nest you may want to know the temperature and also anticipate what temperature you can expect in the future (what the AI predicts you want). You know your in-laws are coming to visit in the evening and they like a cooler temperature, so you ask Alexa to tell you what Nest plans for the temperature in the evening. You learn it will be warmer than they’d like, so you’d like to lower it and change the duration it stays cooler for the remainder of the time your in-laws will be there. Designing the AI of both systems to respond well and understand a user in this situation would be good UX as well as good AI.
These are just two examples of what a great combination of AI and UX look like and here are a few more to ponder:
- AI could learn how different users respond to displays and modify the displays to pick amongst numerous designs that users respond well to and designers helped create.
- AI could automate a lot of the busy work that designers conduct. For example, when creating a new eCommerce site/app there are common elements that exist, so an AI could produce a predefined display (has a place for products, a shopping cart, hamburger menu, and logo) for the UX designer to begin modifying.
- AI could better detect emotional state of the crew on a trip to mars. The AI could show various media (music/calming sounds)/pictures of family/different lighting/indicate to the user that it detects they are feeling down so they are aware of changes they need to make. The design aspect of it would be how these different pieces of information are presented to users.
- AI could anticipate times in which users want to drive instead of having the computer do it. This could be based upon past behavior of the users. The user experience aspect of it would be a smooth transition — think a gentle sound and extending of the driving controls to the user.
- AI can assist a surgeon who is under a lot of stress during a surgery to process data points from the patient and ideal next steps for a surgery to go successfully. Displaying this information unobtrusively and with little confusion is the realm of good user experience.
- AI could anticipate future performance of a market so a company could capitalize on those coming trends. User experience would help show this information clearly to users.
My Company: Saxony Partners
At the company I work for, Saxony Partners, we take good artificial intelligence seriously and have been implementing it. And while powerful, AI isn’t always the ideal solution for your problem.
“Artificial Intelligence algorithms require extensive data. Only about 5-10% of the business problems really need an AI solution. There are a wide variety of algorithms that can help a business understand and predict things. As data scientist we have more and more tools available at our finger tips to create predictive, or prescriptive solutions for businesses. AI is just one of the tools available.” – Charity Wilson
Here are just a few examples of work we’ve done in this space:
- Predicting the appropriate rent rates to charge.
- Predicting what types of rental properties have the most potential for profit.
- Improving the healthcare space by better determining what claims are related to a medical event.
- Giving users information about how many wings to prep to meet demand
We’re always interested in hearing more about business challenges that you have whether it is AI related or requires a different solution. Feel free to reach to to me, Daniel Baskind, Charity Wilson, or Yasir Bashir to learn more!
Conclusion
The possibilities are endless for how AI and UX will work together in the future. Neither are going away and any company or AI/UX professional wanting to remain relevant should be examining these technologies and more to better adapt to the changing marketplace.