Imagining a Future of Designing with AI: Dynamic Grounding, Constructive Negotiation, and Sustainable Motivation

TL;DR

The paper investigates how natural-language-enabled foundation models can uniquely support design beyond traditional tools. It identifies three affordances—dynamic grounding, constructive negotiation, and sustainable motivation—grounded in activity theory and collaboration literature, and illustrates them through design fiction and a diegetic prototype (Squirrel Game/Game Jammer). It discusses practical implications, including localization in the design space, intent elicitation, negotiation strategies, planning and integration, tropes biases, and system requirements and privacy considerations. The contribution provides a bottom-up, narrative-driven framework to guide future HCI research and tool development for human-AI design collaboration.

Abstract

We ideate a future design workflow that involves AI technology. Drawing from activity and communication theory, we attempt to isolate the new value large AI models can provide design compared to past technologies. We arrive at three affordances -- dynamic grounding, constructive negotiation, and sustainable motivation -- that summarize latent qualities of natural language-enabled foundation models that, if explicitly designed for, can support the process of design. Through design fiction, we then imagine a future interface as a diegetic prototype, the story of Squirrel Game, that demonstrates each of our three affordances in a realistic usage scenario. Our design process, terminology, and diagrams aim to contribute to future discussions about the relative affordances of AI technology with regard to collaborating with human designers.

Figures (4)

• Figure 1: (a) The fractal design spiral (FDS). Design iteration moves from a high-level, abstract discussion of a project and its goals, to lower-level activities and finally actions. The AI needs to keep track of 'where' one is in the spiral and integrate outcomes made at lower levels back into upper levels (for instance, the AI should 'remember' choosing a sprite for the squirrel protagonist while prototyping the first level, and 'integrate' it at the project-level as well). The motives of users may emerge throughout this fractal design process (according to activity theory motives are rarely prespecified or even well-understood by users themselves kaptelinin2006acting). (b) Disambiguation with an AI, mapped onto the fractal design spiral. The user is at the blue decision point, choosing between options. Deciding on project-level concepts and goals "covers more distance" through the design space, while nested sub-tasks cover less and less ground (for instance, choosing to center a game around a squirrel's life or a shark's, versus where to place a collectible in level four). Thus, AI 'antagonism' and constructive negotiation is arguably more important at higher levels of abstraction (related to Fig. \ref{['fig:conflict-vs-level']}).
• Figure 2: Benefits of Conflict vs. Level of Abstraction. Design tasks and decisions at lower levels of abstraction are more routine tasks that do little sway the overall concept (e.g., high-fidelity choices like UI button color). Activities and decisions at a high level of abstraction are less routine, demanding more creative energy and benefiting more from conflict (e.g., brainstorming).
• Figure 3: Alice notices a squirrel while sitting on a picnic bench. An idea flashes her mind: "I want to make a squirrel game."
• Figure 4: Alice draws the first level of Squirrel Game. The AI reminds Alice of a fox character decided in an earlier design conversation. The graphical style of the AI's suggested fox is grounded in Alice's sketchy style. When Alice adds ZZZ's atop the fox to represent sleeping, the AI edits the fox sketch to be sleeping, again mimicking Alice's style.