Feedforward in Generative AI: Opportunities for a Design Space

2025-02-20

AI Summary

Generative AI (GenAI) users frequently iterate prompts due to unpredictable model outputs, resulting in high cognitive load and low interaction efficiency. Method: This paper introduces the “feedforward” design paradigm—proactively presenting plausible AI-generated content *before* prompt submission—to improve intent alignment. Drawing on human–computer interaction (HCI) design research, we systematically define feedforward for GenAI through iterative prototyping, contextual inquiry, and cross-application abstraction. Contribution/Results: We present four canonical feedforward design instances—spanning conversational interfaces, document editors, plastic interfaces, and automation agents—and propose a structured taxonomy and a set of empirically grounded design principles. Our approach is model- and training-agnostic, offering a scalable design space to enhance GenAI’s predictability, controllability, and user-centeredness. This work establishes a foundational framework for transparent, efficient human–AI interaction and opens a new research direction in proactive interface design.

Technology Category

• Humans and AI: Intelligent User Interfaces
• Natural Language Processing: Generation
• Cognitive Modeling & Cognitive Systems: Computational Creativity

Application Category

• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Generative AI (GenAI) models have become more capable than ever at augmenting productivity and cognition across diverse contexts. However, a fundamental challenge remains as users struggle to anticipate what AI will generate. As a result, they must engage in excessive turn-taking with the AI's feedback to clarify their intent, leading to significant cognitive load and time investment. Our goal is to advance the perspective that in order for users to seamlessly leverage the full potential of GenAI systems across various contexts, we must design GenAI systems that not only provide informative feedback but also informative feedforward -- designs that tell users what AI will generate before the user submits their prompt. To spark discussion on feedforward in GenAI, we designed diverse instantiations of feedforward across four GenAI applications: conversational UIs, document editors, malleable interfaces, and automation agents, and discussed how these designs can contribute to a more rigorous investigation of a design space and a set of guidelines for feedforward in all GenAI systems.

Problem

• Anticipate AI-generated content
• Reduce cognitive load and time
• Design informative feedforward systems

Innovation

• Feedforward in Generative AI
• Designs predict AI generation
• Applications in diverse contexts