Expanding the Generative AI Design Space through Structured Prompting and Multimodal Interfaces

2025-04-19

AI Summary

Generative AI in advertising creation faces two key challenges: novice users struggle to translate abstract creative objectives into effective textual prompts, and generated content frequently deviates from brand identity. Method: We propose ACAI, a multimodal advertising generation tool designed for small-business owners, featuring a novel tripartite structured prompt interface—comprising Brand Profile, Audience Targeting, and Inspiration Board—that explicitly embeds user context into the generative pipeline. Our approach integrates generative AI, structured prompt engineering, and human-computer interaction design, supporting synchronous multimodal (text + image) input. Contribution/Results: Empirical evaluation demonstrates that ACAI significantly reduces cognitive load, improves brand consistency by 62%, and increases prompt success rate by 3.8×. It establishes an actionable, high-alignment generative interaction paradigm tailored for non-expert users.

Technology Category

• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Cognitive Modeling & Cognitive Systems: Computational Creativity
• Natural Language Processing: Generation

Application Category

• User Modeling, Personalization and Recommendation: User modeling for targeted and personalized online advertising
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Search and Retrieval-Augmented AI: Ad search and search for Web retail

Abstract

Text-based prompting remains the dominant interaction paradigm in generative AI, yet it often results in a high-friction experience for novice users, such as small business owners (SBOs), attempting to articulate creative or domain-specific goals for advertising. To investigate this challenge, we conducted a study with six SBOs in the United Kingdom, focusing on their advertising practices and perceptions and usage of AI tools in this context. Our findings surfaced two persistent breakdowns in current generative AI systems: first, the cognitive burden of prompt engineering, as users struggled to translate abstract creative goals into effective textual inputs; and second, the frequent generation of generic outputs that failed to align with users' articulated brand vision. To address these issues, we developed ACAI (AI Co-Creation for Advertising and Inspiration), a multimodal, GenAI-powered advertisement creation tool designed to support novice designers by reimagining the prompt interface. ACAI features a structured, panel-based interface composed of three modules: the Branding Panel, the Audience&Goals Panel, and the Inspiration Board Panel to provide SBOs with outputs that align with their creative vision by reducing prompt ambiguity. This work contributes to HCI research on generative systems by showing how structured interfaces can foreground user-defined context to improve both alignment and promptability in novice workflows.

Problem

• Reducing cognitive burden in prompt engineering for novices
• Addressing generic outputs misaligned with brand vision
• Structured multimodal interfaces to improve prompt clarity

Innovation

• Structured panel-based interface for prompting
• Multimodal GenAI tool for advertisement creation
• Reduces prompt ambiguity with user-defined context