What we've seen of Generative AI integration in digital products so far has mainly consisted of chatbots and text prompts. But what would it look like if we could use AI integration within our design principles and design systems to enhance large language models (LLMs)? And what are the opportunities and risks that come with that?

‍

In regards to design systems, Big Medium explains how they've begun doing just that. In their article AI and Design Systems, Brad Frost discusses the ways they are applying AI integration in design system build and maintenance, and producing promising outcomes. With one of their first areas of application being generating component code, Frost states, "we guestimate this approach could help developers create components 40–90% faster than manually writing things from scratch."

Other areas where Big Medium has begun applying AI include component translation, testing and QA, as well as documentation. Some of their goals are to alleviate their clients from the more laborious tasks of a developer, while also tailoring design system language to the individual, taking into account their discipline, their skill level, their context, and even their preferred spoken/written language.

‍

However, implementation of AI in these systems requires human-centric guidance in order to maintain effectiveness. As our Founder, Eve Eden states "Feeding design principles and a design system into an Al engine involves leveraging these principles and systems to guide the training, decision-making, and output generation processes of the AI." The quality of the AI’s output depends on the quality of the training data. Poor or outdated design systems can lead to ineffective recommendations.

‍

When discussing the emerging landscape of AI tool integration into our design systems and principles, it's important to take a few key areas of impact into consideration.

Respect: it's important we remain respectful of human talent and work effort, leveraging AI to allow us to take on more fulfilling work, not replace our craft or humanity.

Security & Privacy: in order to have AI integrated successfully into an organization's workflow, security & privacy are critical. Only on-premise AI tools should be utilized to keep information secure.

Human Ownership: Organizations need to have human team members controlling input and output, modifications, and training at all times.

Predictability & Reliability: until AI tools are producing predictable results, they shouldn't have any role in your front-end infrastructure. Refining your AI systems to behave reliably is key before implementation.

‍