Machine Learning (ML) has recently become one of the most popular buzzwords, but artificial intelligence (AI) has been used across industries for years. For example, the tailored ads we see on our Facebook, Instagram, and Twitter – this is all machine learning. While many businesses may assume ML is impossible or inaccessible, it’s a defining factor to digitally transforming your firm.
Combining ML with User Experience (UX) is a new method designers are being introduced to. ML is a tool or set of tools that designers can use to enhance experiences for users. Because ML is so new, the UX field is just beginning to explore the capabilities of artificial intelligence. Moving toward investing in artificial intelligence for UX is a big step forward for any firm. But it’s evident that ML is not just a trend, it’s here to stay.
Machine Learning + UX
At the core of UX, a designer’s goal is to provide users with a positive experience when they interact with a product or brand. Good UX should excel in visual design, information architecture, content strategy, interactive design, usability, and user research. While UX teams incorporate all these disciplines, ML takes the experience hundreds of steps forward.
ML allows us to interact with users more intelligently, and as designers are becoming more experienced in AI, more progress will be made towards incorporating AI within UX. ML is a developed application that uses AI to learn from experiences through data collection and from that data it will predict outcomes, without explicitly being programmed.
What this means for designers is ML could transform the user experience to give each user a unique and personalized experience when interacting with a product or service. With ML, the artificial intelligence will collect data surrounding the user; this can include data about the user’s likes/dislikes, interests, and their interactions. The information collected by the artificial intelligence is used to learn about the user on a deeper level. With this data, the machine learns to understand each specific user and provide them with a customized interaction.
A simple example of UX + ML is Google’s Personalized Search feature for Search engine results pages. The ML collects data on which links you click on from your results page. In return, ML provides you with results more correlated to the links you have clicked in the past. Let’s say you are planning a trip to Brazil and one of your destinations is the Amazon rainforest. When you search “Amazon” you will be more likely to see results relating to travel and trips for the Amazon rainforest, whereas if a different person searched “Amazon” their search results will direct them to the eCommerce website. In this instance, the ML has learned about you as a user and it predicts what results you want to see. Overall, Google’s Personalized Search feature improves your search experience and gives customized results, saving you time.
Taking a look at the competitive landscape, every company is striving to be the Uber or AirBnB of their industry. To do so, it’s crucial to differentiate as a disruptive business – this is where ML comes in. Businesses that are customer-obsessed (read more about digital strengths here) can take their services to the next level with ML. Through the collection of data to understand users, you can deliver a customized experience to each person that will keep them coming back.
Case Study: Netflix
Most people are familiar with the online streaming service, Netflix; however, what some may not know is that Netflix is one of the strongest examples of UX and ML on the market today.
Netflix built an interface that provides user’s with a personalized experience. The ML learns the user’s behavior patterns and in return, the platform recommends movies and predicts what the user would like. The user design of this product delivers an excellent experience to their subscribers because they do not need to sift through pages of movies to find one they are interested in. This is also the reason why we can spend hours watching Netflix – the system’s algorithm can figure out what we will enjoy watching. If a user comes across a recommended movie that they dislike, once they give the movie a poor rating, the information will be taken into consideration for future recommendations.
Netflix is disrupting the industry by providing their users with customized experiences on their platform. As a company worth over $70 billion and growing, it’s clear that investing in ML is contributing to their growing success.
Considerations to Think About Before Beginning the Process
If your company is interested in integrating ML, there are a few considerations to think about before beginning the process. Below we have outlined questions you should be asking before implementing ML in your strategy.
Why do you want ML in your UX?
While ML is exciting, it is not meant to identify your problems – the purpose is to take an existing problem that you know you have and help to provide a solution. If there isn’t a true need, you will be building ML for a non-existent problem.
Is AI the best solution?
When you identify the problem you need to have solved, analyze whether ML is the proper solution. Some projects benefit from AI and others may have a better solution. Also, consider the cost vs benefit – will creating an ML solution be worth the cost?
How will you create a prototype?
To test a machine learning system, a prototype must be created. To truly test your ML, you must use unique data sets. It is also important to test the UX during this time as well to ensure it’s up to par. There are two methods to prototyping:
- Personal Examples: Bring in participants with data; for example, their music recommendations, photos, movie preferences – any information/data about them. These participants are then used to test the program.
- Wizard of Oz Studies: This method is usually used for ML that has not been developed yet. Participants are used to test the ML program, but a human is controlling the system and its output (the participant does not know a human is manipulating the system). This helps navigate design thinking by observing how the participant interacts.
What mistakes will your AI make?
Machines will make mistakes, but it’s important to consider what kind of errors your end user will run into if they use your ML. What needs to be considered is if it’s more important for your system to provide more answers with the chance of having incorrect outcomes OR providing less answers that are more accurate at the risk of leaving out possibly correct answers. For example, if Google images of “Disney World” your search will provide images of the theme park, as well as images outside of that scope, like the Disney characters. In this scenario, the ML is giving the end user more answers that are less precise.
How will you label your AI?
For ML to properly work in UX, the algorithm needs correct labels. If you are training your system to identify flowers, you must go through a set of photos with flowers and without flowers in them, and label the photo as “with flowers” or “without flowers.” After that, your AI can be trained to identify if a photo has a flower in it or not.
How will you improve your ML?
Successful ML should be continually improving. When ML begins to be used by humans, the system is learning about users, therefore the model should be adjusting to provide the best outcomes. This is especially vital for improving the UX design.
As consumers are exposed to more intelligent systems, they will demand better capabilities in products, specifically features only possible with ML. All in all, companies cannot reach the full potential of their digital transformation journey without implementing machine learning in their UX strategy. Ultimately, ML will optimize user interactions and develop a unique and individualized experience that people have come to expect.
To get started, consider Google Cloud’s AI, which provides modern machine learning services, with pre-trained models and a service to generate your own tailored models. This neural net-based ML service has better training performance and increased accuracy compared to other large scale deep learning systems. The services are fast, scalable and easy to use. Major Google applications use Cloud machine learning, including Photos (image search), the Google app (voice search), Translate, and Inbox (Smart Reply).
Contact us to learn more about implementing Google’s AI services into your UX strategy.
Get the latest industry news and insights delivered straight to your inbox.