Generative Interfaces

Software design has long been dominated by fixed interfaces trying to please every type of user, often resulting in overwhelming and cluttered experiences. Microsoft Office, with its countless buttons, menus, panels, and tabs, is a prime example of this issue. Despite designers' best efforts to streamline these interfaces, achieving the perfect layout seems almost impossible. However, with the rise of generative AI, we may witness a revolution in software design centered around the user's mental model.

Traditional software interfaces are designed by people with technical backgrounds, who sometimes neglect the diverse needs and expectations of their users. This can lead to a one-size-fits-all approach that ignores the unique requirements of each user. For instance, while some users may appreciate many features and options, others may find them rather confusing and intimidating. The outcome is an interface design that forces users to conform to its limitations, rather than the other way around—a phenomenon explored in my article on Mental Models.

Microsoft's attempt to address this issue with its virtual assistant Clippy ended in disaster. Clippy's annoying presence and lack of usefulness made it the subject of many jokes, and soon became synonymous with bad software design. However, the idea behind Clippy, a support feature that essentially acknowledges the user's struggles with software navigation, was not entirely misguided. After all, removing Clippy didn’t exactly make things better either.

Imagine starting a new Office document and having a blank slate in front of you, but with zero controls. Instead of being overwhelmed by a sea of buttons and menus, you could simply ask your AI assistant for what you want to do. Insert a decorated image, apply a certain type of font, or build a table for specific data – your AI assistant would take care of it. This approach would allow the user to focus on their current needs, rather than struggling with the interface.

It could even help you construct convenient buttons, based on your most frequently performed tasks. To take things a step further, you could ask it for entire working environments; From the occasional blogger to professional journalists, graphic designers, students, and the elderly, they might all want their own environment—ranging from intricate interfaces to simplified ones with larger buttons.

With this kind of generative AI, users could create their own set of "formulas" that resonate with their specific needs. These formulas could either be shared with others, creating a community of users who can build upon and learn from each other's presets. This ecosystem would encourage collaboration, create a sense of community, and would have the potential to elevate the entire creative process.

Some might feel "leaving software design over to ordinary people would be a bad idea", potentially leading to even more cluttered and disorganized workspaces. However, AI could simultaneously assist us in organizing our environments in an intelligent layout, based on the user's preferences and frequently used features. This would ensure their favorite tools are always within reach, while keeping the workspace tidy and efficient.

In conclusion, generative interfaces may just hold the potential to revolutionize the way we interact with software. By focussing on the user's mental model, rather than the implementation model, we might witness an evolution in software that is truly user-friendly and efficient. With the power of personalization and AI-assisted organization, we’d be able to break free from the constraints of traditional interfaces in favor of highly fluent designs tailored to the unique needs of each user.

We may finally have a version of Clippy that is actually considered useful. 📎