Designing AI Interfaces: Design Principles for Creative and Autonomous AI by Louise Macfadyen

Designing AI Interfaces: Design Principles for Creative and Autonomous AI by Louise Macfadyen

The central thesis of Designing AI Interfaces is that the future of artificial intelligence isn't just about training smarter backend models—it is about transforming how humans steer them. Macfadyen addresses a common trap in modern software deployment: companies frequently "crowbar" AI features directly on top of legacy apps, dropping a basic, blank text box into the layout and expecting users to figure it out. This creates a steep learning curve and causes user fatigue, as humans project unrealistic capabilities onto the software and end up frustrated by unexpected model errors.

Rather than approaching design as a purely aesthetic layout task, Macfadyen focuses heavily on closing the "conceptual distance" between what a language model actually does and what a user imagines it can do. The book outlines a highly tactical, three-part Input-Computation-Output framework. It provides real-world design patterns for managing user attention during high-latency deep reasoning tasks, creating clear "affordances" (visual cues) that show users how to modify prompts, establishing strict content moderation warnings, and implementing "interruptibility" controls that let users stop a running agentic task mid-stream before it writes to a database.