The world of generative AI and large language models (LLMs) balloons at a pace so fast it’s making us dizzy. For UX designers, this means keeping up to speed on the technology while understanding the new paradigms of interactions. So, how do you lead designers into this new frontier that brims with ambiguity? As a design leader, I’ve experienced the dawn of many new technologies, including mobile apps, chatbots, voice-enabled devices, and more. Each time, I’ve centered myself on some core practices that help me explore what’s new without losing my grounding.
Gain a foundational understanding of the new tech
To design for a product you need to first understand the domain. For generative AI, designers need to understand the principles of machine learning (ML) and AI to design effective solutions. This includes learning about data prep, feature engineering, ML modeling, model evaluation, and optimization.
On top of gaining understanding of core ML concepts, we need to bring ourselves up to speed on LLMs, as well. Creating interfaces for prompt engineering, fine-tuning, etc., requires designers to first understand these concepts and how they work. I encourage collaborative experimentation and foster a culture of learning via trying out new ideas and taking risks.
UX design for generative AI involves creating interfaces and interactions that enable users to interact with and provide input to the generative AI system. The goal of UX design in this context is to create an intuitive and engaging user experience that enables users to generate new content or output easily and effectively.
While we want to design for something new, this also should tie into the golden UX principle – designing for user needs. Stepping away from the cool tech and grounding the team in user needs and exploring use cases is where UX designers can provide valuable insights to teams developing generative AI innovations.
If your design doesn’t solve a user need, then it’s technology for the sake of technology. And that doesn’t get adoption beyond the initial hype.
Experiment and iterate
It’s essential to experiment with new tools, methods, and approaches. Collaboration between team members is vital, particularly when working on complex projects. Designers should collaborate closely with data scientists and developers to ensure that the generative AI system is built on a solid technical foundation, and that the design is implemented effectively. Designers need to experiment with different generative AI techniques and interfaces, and iterate based on feedback from users and stakeholders.
Consider ethical considerations of new technology
UX designers need to consider the ethical implications of generative AI, such as amplifying biases present in the training data or producing content that is harmful or offensive. They should design systems that promote responsible use of generative AI.
Partner with stakeholders to find appropriate use cases
Think about who you’re trying to serve and what they need. Start by understanding users’ work challenges and build systems that improve the customer experience. As we grapple with what LLMs can do, we need to understand that they do struggle with accuracy and are prone to hallucinations. Evaluating risk is also important as we explore the right use cases for UX design and generative AI. Can the generated output be damaging or is the consequence of errors really low? Balancing risk with value is of utmost importance while brainstorming ideas with your stakeholders.
Communicate clearly and transparently
The ambiguity of technology, a designer’s role, and project scope can make team members feel uncertain or overwhelmed. It’s crucial during these times to answer questions, provide feedback, and offer resources to help them succeed. Especially because generative AI is transforming the industry so quickly, it requires a combination of technical expertise, strong leadership, and excellent communication skills. It’s important to communicate clearly and transparently.
“In general whenever my team is stressed about something that is unknown or out of their control, I try to encourage them to focus on the things they can control and trust that everything else will work itself out the way it’s supposed to,” said Becca Yukelson, Senior Manager Design Systems at Salesforce.
Leading UX designers into the ambiguity of a generative AI can be challenging, but it’s also an opportunity to be creative and innovative.