DesignLab: Designing Slides Through Iterative Detection and Correction

2025-07-23

AI Summary

Non-experts struggle to autonomously generate high-quality presentation slides due to the need to coordinate multiple interdependent design decisions; existing automated tools produce slides in a single pass and lack iterative refinement capabilities. To address this, we propose a dual-role iterative framework wherein a large language model alternately assumes the roles of “Reviewer” (identifying issues in layout, color harmony, readability, etc.) and “Contributor” (revising slides based on feedback), with controlled perturbations applied to intermediate drafts to enhance training stability and diversity. This mechanism emulates professional design workflows, enabling end-to-end self-refinement. Experiments demonstrate that our method significantly outperforms state-of-the-art automated tools and commercial solutions across slide professionalism, visual consistency, and information clarity. To our knowledge, this is the first AI-driven presentation generation system to realize a closed-loop design process with explicit reflection and correction capabilities.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

Designing high-quality presentation slides can be challenging for non-experts due to the complexity involved in navigating various design choices. Numerous automated tools can suggest layouts and color schemes, yet often lack the ability to refine their own output, which is a key aspect in real-world workflows. We propose DesignLab, which separates the design process into two roles, the design reviewer, who identifies design-related issues, and the design contributor who corrects them. This decomposition enables an iterative loop where the reviewer continuously detects issues and the contributor corrects them, allowing a draft to be further polished with each iteration, reaching qualities that were unattainable. We fine-tune large language models for these roles and simulate intermediate drafts by introducing controlled perturbations, enabling the design reviewer learn design errors and the contributor learn how to fix them. Our experiments show that DesignLab outperforms existing design-generation methods, including a commercial tool, by embracing the iterative nature of designing which can result in polished, professional slides.

Problem

• Helps non-experts create high-quality slides iteratively
• Automates design error detection and correction separately
• Improves slide quality beyond existing automated tools

Innovation

• Iterative loop for continuous design refinement
• Role separation into reviewer and contributor
• Fine-tuned LLMs for error detection and correction