More UI Innovation, Not Less

I find it difficult to imagine a future in which the graphical user interface becomes obsolete.

The current narrative suggests that conversational UI will dissolve the need for graphical user interfaces. The “no interface is the best interface” quote is spreading among agreeable YC disciples. Software is being reduced to chat boxes, with voice input as a prominent feature.

While conversational UI is useful, language does not offer a universal replacement for visual interfaces. A future in which software is predominantly abstracted to conversations is not necessarily ideal, or desirable.

We often underestimate how efficient, beautiful, and intuitive visual communication is. It can often say much more than words, and does not require a sophisticated translation algorithms to cross borders.

One button can encapsulate intent and execute in one click. Option menus can surface complex capabilities without prior knowledge. Layouts can encode process, workflows, and priorities. Images can convey meaning without linguistic fluency. Visuals can compress multiple prompts in accessible form. We don’t need to learn, explain, or converse - visual interface elements serve as shortcuts for understanding, navigation, and execution.

Despite LLM chats making up an increasing share of our computing interactions, my fundamental belief is that AI will drive more graphical UI innovation, not less.

Intelligence will be able to express itself in more fluid, adaptive, and contextual ways - potentially in a variety of shapes, materials, and behaviours. Interfaces will no longer be constrained to static representations of predefined logic. Software UI will be defined by its responsive properties, tailored to the user and informed by its environment.

This new type of undeterministic software will require a different manifestation. Our existing component libraries and design systems will need to evolve their form and function. The rendering of our data and workflows will be ours to reinvent.

This provides a unique foundation for novel visual interface primitives to be developed.

Emerging hardware categories will also introduce new surface areas for software artists to experiment with. Increased computational power will enable more immersive, interactive, and dimensional software UI.

We can extend our imagination of visual software surfaces beyond hardware. Screens may be integrated in our environment, or our environment may become an ambient display. AI may allow software to translate itself into visual form responsively, adapting to the material and context it is rendered in or projected onto.

All of this, in combination with more advancements we may not have contemplated yet, creates a vast playground for more visual interface innovation than ever before.

History suggests that visual styles, aesthetics, and expressions rarely converge towards uniformity. We can expect software visuals to follow the same pattern, potentially at an accelerated rate.

Interfaces are about more than functional utility and efficiency. They provide an experiential layer that engages and stimulates us. Creative UI will offer an “experiential premium” on top of commoditised software capabilities, making interface design an economic advantage for product companies.

If biology remains relatively constant, visuals will remain core to our experience of the world, including the software we interact with. Estimates suggest that 70-80% of our sensory information is visual, with 30-50% of our cerebral cortex allocated to visual processing. It would be short-sighted to conclude that graphical user interfaces will be displaced.

Humans have developed a variety of languages to express different things in different situations. This applies to software, where our choice of interaction modality depends on context, preference, and application. Code for development, visual manipulation for design, computer vision for context capture, keywords for retrieval, and now language for general-purpose instructions or conversation.

Users will be able to interact with software in the modality of their choice, with more options to choose from. This will enable software to serve a broader range of edge cases and appeal to a larger variety of preferences than ever before.

It a false dichotomy (and useless debate) to compare and predict the “future interface” for software. There is no interface singularity. There will be more UI innovation, not less. We are too creative, too expansive, and too ambitious to limit ourselves.