For a long time, machine learning has been an integral part of how users experience the platforms and applications they love and use daily, powering things like ads and content recommendations, interactions with smart speakers and chatbots, and face filters on social media.
Today, the news is all about Artificial Intelligence (AI)—systems built with machine learning (ML) that encompass various interrelated computational technologies, including natural language processing, image processing and recognition, as well as data science and analytics. It’s a rapidly growing area and something we’re hearing more and more questions about from our clients and the UX community as a whole.
Over the past few years, the tech industry has produced breakthroughs around generative AI in natural language processing (NLP) and computer vision that will fundamentally change the way users experience and interact with technology in their daily lives. Image generators like DALL-E and Midjourney and chat-based text generators like ChatGPT are just the beginning of a huge leap forward in AI capability, which has the power to revolutionize the way people work, communicate, access information, and go about their daily lives. Productization around new generative AI models has already taken off, most notably with Microsoft’s AI-powered updates to their Bing search. The integration of AI into search has the potential to revolutionize the search experience. Bing's capabilities include summarizing information from relevant links to performing complex actions like building meal plans, creating full travel itineraries, and providing support explanations for homework assignments.
Looking ahead, many existing apps and tools will integrate AI capabilities in new and exciting ways. In March, Instacart rolled out plans to use ChatGPT to help power a redesigned search engine feature on their app. “Ask Instacart” will help users plan meals, build recipes, and find ingredients in a dialogue-based format. That same month, Microsoft announced that it was planning to add GPT-powered features across its 365 suite in the form of a tool called Copilot. This new tool will help users write emails, create and edit documents, and generate compelling imagery, all with the goal of improved productivity across both personal and business workflows.
These new generative capabilities—interpreting and generating human-like speech and recognizable imagery—have the power to facilitate many new modes of user experience and interaction. Beyond transforming the search engine experience, generative AI may underwrite new forms of automation, deepen personalization, and enhance new forms of interaction like speech and gesture-based UI.
But even with these fast-moving advances, there is the ever present challenge of making products that actually meet user needs. Good UX design based on solid UX research will be a critical part of successful productization. UX research is crucial to ensure these products are not only usable, but also provide a delightful, and most importantly, safe and secure, user experience.
In a moment of great hype and excitement around AI’s emerging capabilities, it’s important to remember that a successful product will have to address users’ actual needs.
Tools built with generative AI are highly capable solutions looking for problems to solve. UX research with users from diverse backgrounds can help product developers understand a wide range of user goals and needs and identify how a product could or should fit within their daily lives. It can also help uncover their existing mental models around AI to facilitate successful user onboarding and adoption. Not to mention, research on safety, privacy, and security will help companies understand users needs and concerns, assess risks, and better prepare for current and potential regulations around AI technology.
Generative research with users is foundational for defining problem spaces and understanding what opportunities exist, which is particularly important when considering AI integrations. Research at this stage can provide a jumping off point that can make the product path clearer, while assessing how AI’s unique capabilities could be necessary or useful for addressing specific needs identified during research.
Generative AI has already become a force within the creative sector for writers, musicians, artists, game developers, and content creators. User testing with key creative groups has surfaced insights around how AI-powered tools change the way they approach their jobs and how companies can build tools to better empower creators. On the other side of the product equation, customers' voices help companies understand how AI-driven products are perceived, so that the tone and personality of virtual assistants or the images in AI-generated ads remain relatable.
Beyond developing strong value propositions, UX research early in the product lifecycle can also help product teams assess and establish basic usability for an AI-enabled product or feature. Even before a demo version of a tool or feature has been fully built, UX researchers can perform “Wizard of Oz” style usability testing. In this method, researchers present participants with a human-powered simulation of the AI-enabled functionality. This type of usability testing can help product teams assess the initial user experience and identify areas for improvement before investing time and resources into developing a fully functional minimum viable product. Product teams can also benefit from user feedback on concepts, enabling quick design decisions without fully built-out functionality. Especially with AI-based experiences, developing a prototype can be a particularly heavy lift. Early research can save critical time, money, and resources.
Mental models are the existing beliefs of users, based on both current knowledge and previous experiences, about how a system, tool, or process works. For AI-enabled applications, understanding how users think AI works and what they think it can or can’t do is essential for designing human-centered tools. Understanding AI mental models can be complicated by the fact that many emerging AI-enabled interactions may feel totally novel to users, offer functionality they’ve never seen before, and work in non-transparent ways. This inherent lack of transparency and the potential for a mismatch between user mental models and AI capabilities can lead to lower rates of adoption. This can also damage trust with users when something goes wrong or an interaction feels “off” or creepy. UX research centered on user intent can help product teams test the contextual tone and tenor of AI-enabled experiences and help developers better understand what users expect from their interactions with AI.
…
As AI continues to revolutionize the tech industry, UX research plays a vital role in ensuring the successful integration of AI into products and services. While the capabilities of generative AI are expanding rapidly, it’s important to remember that the focus should always be on meeting user needs and providing a safe and secure user experience.
By conducting UX research, companies can create AI products that not only leverage the power of the technology, but also enhance the user experience and drive adoption.
Discover how we helped a client build a better AI-powered translation experience.