Researchers found twelve recurring interaction patterns across AI tools — can mixing and matching them actually invent new ones?
Can twelve atomic paradigms compose into novel generative AI interaction designs?
This explores whether a catalog of twelve basic, reusable interaction patterns for generative AI tools can work like building blocks, mixed and matched to invent new interfaces, and what the rest of the collection says about where that kind of recombination helps and where it falls short.
This explores whether twelve recurring interaction patterns, pulled from existing generative AI tools, can be recombined like building blocks to produce interface designs nobody has built yet. The short answer from the collection is that the patterns are designed to combine, but no study here shows a recombination producing a new design that was then tested. The source is the IAI model How do interaction signals reshape what generative AI actually receives?. Its key move is to say that the AI never just receives your prompt. It receives an 'augmented instruction': your words plus everything the interface adds, such as sliders, selected regions, templates and constraints. After analyzing 66 tools this way, the authors found twelve patterns that keep recurring and can be combined. That framing matters because it lets you compare two very different-looking tools by asking what each one actually sends to the model.
Other notes give a better sense of what would be combined. Friedman's five design levers (capturing intent, shaping output, refining, handing off agentic tasks, and embedding AI in existing workflows) amount to a parallel menu of building blocks Can task-oriented UI patterns reduce the friction of AI chat interfaces?. Medeiros offers a rule for choosing among them: chat is often the wrong default for structured work. Two questions should decide between talking and clicking: how fast the task needs to be, and where its value comes from Is chat the wrong default interface for AI products?. Read together, these suggest that composing patterns is less about stacking features and more about matching each piece of a task to the right way of interacting. Generative AI raises the stakes because users now describe *what* they want rather than *how* to do it. That makes outputs unpredictable, and older consistency rules from interface design stop applying cleanly How should users control systems with unpredictable outputs?.
Composition also has a practical limit that the collection shows clearly. A design that combines well on paper doesn't guarantee a tool that behaves that way. A benchmark of generative UI tools found that about a quarter of stated design rationales were never implemented, rising to a third for functional requirements. The tools also recognized only half of the UX principles written into their prompts Do generative UI tools actually implement their stated design rationales?. A related point is that the context an AI works from (prompt, history, retrieved data, hidden state) keeps shifting in ways a fixed interface diagram doesn't capture How does AI context differ from conventional software context?. So any combination of patterns sits on top of a moving foundation.
The most interesting implication may be that the set of building blocks isn't closed. Some design dimensions sit outside an input-focused catalog entirely. One is how human-like the AI should seem, which one paper treats as an adjustable setting rather than a fixed default Can interface design reduce harmful anthropomorphism of language models?. Another is how AI should fit into team rather than individual work How should AI design connect individual work to team knowledge?. The user isn't always human either: interfaces can be tuned for computer-use agents without hurting human usability Can interface design help agents complete computer tasks better?. And a single model given branching prompts can reproduce what multi-agent systems do Can branching prompts replicate what multi-agent systems do?. That suggests some 'new' designs come from restructuring the prompt rather than adding interface widgets. The twelve patterns give a useful shared vocabulary for comparing tools and spotting gaps. Whether a new combination actually works is a question the collection leaves open.
Sources 10 notes
The IAI model formalizes human-GenAI interfaces as a six-entity graph where Augmented Instruction—machine-readable encoding of prompts plus interaction constraints—is what GenAI actually receives. Applied to 66 tools, it distills twelve recurring, composable interaction paradigms and enables design comparison where prior taxonomies could not.
Friedman identifies five design levers—intent capture, output shaping, refinement, agentic tasks, and workflow embedding—that sidestep slow typed prompts. Examples include node-based intent visualization, style lenses for output, and canvas manipulation, shifting AI interfaces away from conversational chat toward direct interaction.
Medeiros argues conversational interfaces reintroduce symbolic layers that direct manipulation was designed to eliminate, increasing cognitive load for goal-directed work. Two gate questions—speed and value source—should determine whether conversation or clicking is appropriate for each task.
Generative AI shifts interaction to intent specification rather than method specification, creating unpredictable outputs that violate traditional consistency heuristics. Six design principles—including co-creation, imperfection tolerance, and mental model support—address this novel paradigm.
A benchmark of 24 tasks across five tools found roughly 25% of design rationales go unimplemented, rising to 34% for functional requirements. Tools recognized only half the UX principles embedded in prompts.
Show all 10 sources
AI interactions operate on a substrate of constantly shifting context—prompt, history, retrieved data, hidden state—that users cannot internalize like traditional UIs. This structural mutability demands a new design discipline centered on context engineering rather than interface design.
The paper argues anthropomorphism should shift from a fixed interface default to an adjustable design variable. Using materiality-based metaphors across a spectrum can counteract uncritical user trust and expose LLMs' dependence on human labor, infrastructure, and data.
Atlassian's published design guidelines for its Rovo AI product specify that AI should connect individual work to team knowledge through context-awareness, flow preservation, transparent system state, and fluid human-AI multiplayer collaboration, operationalizing abstract collective productivity principles as concrete UI commitments.
UI-Verse benchmark testing found that augmented heuristics targeting agent perception, grounding, and control consistently improved agent task completion and efficiency. Human studies confirmed these designs preserved usability for human users with no observable regressions.
Research shows single LLMs using dynamic persona simulation achieve multi-agent cognitive synergy without multiple model instances. Solo Performance Prompting validates that structured prompting techniques map directly to multi-agent debate architectures, enabling equivalent outcomes through structural equivalence.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- What does Generative UI mean for HCI Practice?
- Interaction-Augmented Instruction: Modeling the Synergy of Prompts and Interactions in Human-GenAI Collaboration
- Design Theater: Evaluating the Gap Between User-Facing Design Reasoning and Implementation in Generative UI Tools
- Probing How Users Interact with Turn-Level Design Frictions for AI Chatbots
- Augmenting Interface Usability Heuristics for Reliable Computer-Use Agents
- The Conversation Trap: why defaulting to chat might be the biggest interaction design mistake of the AI era
- Design Principles for Generative AI Applications
- Large Language Models for User Interest Journeys