UI/UX design for AI products
- Apr 10, 2023
- By: Slick
- 8 min read
UX Research and Design are crucial when developing a new AI product or feature. Leveraging proven methods of UX Research and Usability heuristics can ensure your product is meeting a user’s needs and that it will be an enjoyable tool to use.
Artificial intelligence (AI) is increasingly prevalent in our lives, from virtual assistants like Siri and Alexa in our homes, to machine learning algorithms that assist radiologists in detecting early signs of cancer in medical images. With the increasing momentum and excitement around incorporating AI into products, it’s crucial to remember that the end user should always be at the forefront of our design decisions. It’s not just about incorporating AI into products but creating useful and delightful experiences for our users. While there are many articles that tell you how you can use AI to streamline UX/UI Design processes, there’s less focus on how best practices in UX/UI can help you design an effective and enjoyable AI experience for your customers. Let’s explore some best practices that can help guide UX Design for AI products, using an understanding of best practices for UX Strategy and proven heuristics in UI design.
A mistake that many startups make is taking an idea that they have for an innovative product, and immediately launching into building and marketing it before they’ve invested in user research to validate the value and innovation that they are truly providing for their ideal customers. There are many great ideas that have failed to turn into successful products, because they didn’t provide a strong value proposition.
User Research can validate or invalidate assumptions that you’ve made about the product’s value proposition and how it will fit in the competitive marketplace. If you have an amazing idea for a new product that uses AI innovation, it’s important to conduct user research to validate every assumption you’ve made. Learn who your customers are and what their pain points are, gather evidence to prove that they struggle with the problem you’re trying to solve, and understand their existing mental models. With AI tools specifically, it’s important to make sure that you understand and respect your customers mental models and design for those models, as opposed to requiring your users to learn an entirely new model to learn your product.
Here are some methods you can use to conduct generative research for an AI product
- Contextual inquiry- observing users in their natural environment to understand their current pain points and identify opportunities for AI intervention. If your AI product is intended to simplify a complex task, it’s important to first observe users’ current workflows and identify inefficiencies and strategize how your AI solution can solve those real problems.
- Ethnographic research- in the context of UX Research, ethnographic research involves observing and interacting with users as they go about their day to understand how they interact with technology. This can provide insights into how users currently interact with existing AI tools, their perception of AI, and their expectations for AI-enabled products.
- User interview-User interviews can be used to dig deeper and validate and prioritize your AI product concepts. In a user interview, the researcher can ask users to describe their ideal AI product, what benefits they would expect, and what concerns they might have using an AI tool to solve their problem. User interviews provide rich qualitative insights that can be turned into actionable design recommendations around feature prioritization.
Design for AI Products
When designing for AI Products, it’s important to not carried away into the excitement of your innovative product and remember to design for your users with the understanding you gained from user research. Neilson Norman’s 10 usability heuristics apply just as much to an AI powered tool in providing an experience for your users that aligns with their goals and expectations. Here are some approaches to consider as an AI Designer.
Match your product to the real world – Nielson Norman’s 2nd Usability Heuristic. This is particularly relevant in the context of AI UX Design for products that may be less familiar to your users. The AI product should use language and concepts that are familiar to users and aligns with the user’s mental model and understanding of the task and domain. For example, if you’re designing an AI Virtual Assistant tool, the language should match their expectations for everyday conversation with a real person.
An example of a product that does this well is Amazon’s Alexa – the AI is designed to use natural language that users would use in everyday conversation. This helps the user feel more comfortable using the product because they can interact with it in a way that feels familiar to them. If the AI product uses jargon or technical language that users are not familiar with, it can create confusion, frustration, or unease, leading to a poor user experience.
Design for flexibility and efficiency of use- Neilson Norman’s 7th Usability Heuristic. When designing an AI product, it’s important to consider the varying preferences and skill level of different users. Neilson Norman’s Flexibility and efficiency of use heuristic (link) states that a product should provide multiple options for users to complete their tasks. The product shouldn’t require a user to complete a task in a specific way, especially if it is going to ultimately cause just as much friction and learning curve as the problem the product is trying to solve. An example of how this heuristic is used is Adobe Sensei- an AI powered tool suite for Adobe Creative Cloud products that incorporates a range of features that make designing more efficient and flexible. Adobe Sensei includes machine learning tools and tools for automating tasks that can be customized to suit the specific needs of the user. This allows an advanced user to set their own workflows to streamline their work.
Slick and Product Design for AI
Slick worked with The Collaboratory on an AI powered search engine tool and applied best practices for AI UX Design. First, we collected and evaluated all the relevant information to understand the value proposition and the users that we’re designing for. This research allowed us to focus on prioritizing features that would have the most impact on their users and solve problems that were evident through quantitative and qualitative analysis.
This led us to focus on problems that existing users had- such as seeing search results that didn’t fit their needs and the effort required to locate necessary internal documents and related public documents. These were features that were underutilized, and we used AI UX Design best practices to make those features easier to use- such as providing a simple button that allows the user to provide sentiment feedback on their search results, matching their existing models of how they would normally interact if a person was presenting them with documents that didn’t match their needs.
Slick worked with HelloRep.ai, a platform
that provides AI-powered solutions for businesses to increase customer
engagement and support. We focused on designing an onboarding experience for the product that implemented conversational UX to guide the user into their initial experience with the product, and utilized common UI patterns that provide feedback to the user while also implementing an element of delight. One goal with this design was to increase the new user’s comfort with the product.