X+Slides: Benchmarking Audience-Conditioned Slide Generation

2026-06-17

AI Summary

Existing evaluation benchmarks for slide generation overlook audience heterogeneity, making it difficult to assess how well content is tailored for distinct listener groups—such as domain experts versus decision-makers. To address this gap, this work introduces X+Slides, a new evaluation benchmark encompassing 113 topics and 7 presentation scenarios, along with 8,133 deduplicated, source-aligned probes. It further proposes the first audience-conditioned evaluation framework, which defines four complementary metrics: audience coverage, domain coverage, efficiency, and correctness. This framework integrates a dynamic utility-weighting mechanism with multidimensional automated assessment to enable fine-grained quality evaluation. Empirical results at τ_A = 0.7 reveal that leading systems—DeepPresenter, SlideTailor, and NotebookLM—achieve audience coverage scores of 0.714, 0.594, and 0.853, respectively, highlighting significant shortcomings in current approaches regarding effective transmission of critical information to target audiences.

Abstract

Automatically generating slide decks from source documents is an important application of large language models (LLMs). Existing benchmarks primarily assess slide completeness and technical depth, while overlooking the target audience as a critical real-world factor. For instance, specialists demand rigorous proofs, whereas decision-makers prioritize actionable conclusions. To bridge this gap, we introduce X+Slides, a benchmark specifically designed for audience-conditioned slide generation. Built on a diverse corpus spanning 113 topics and seven presentation scenes, X+Slides employs a dynamic evaluation framework constructed from 8,133 deduplicated, source-grounded probes. By assigning audience-specific utility weights to the same source-grounded probes, X+Slides reports four complementary metrics: Audience Coverage measures how much audience-essential information is conveyed, Domain-wise Coverage shows which information types are covered, Efficiency measures delivered utility per unit of attention cost, and Correctness verifies whether slide claims are supported by the source. Experiments on DeepPresenter, SlideTailor, and NotebookLM show that current systems can recover a substantial but still incomplete part of audience-essential information: at $τ_A=0.7$, DeepPresenter reaches a best Audience Coverage of 0.714, SlideTailor reaches 0.594, and the NotebookLM ablation reaches 0.853 while showing clear grounding differences. These results indicate that visual quality and broad topic coverage should not be treated as evidence support without source-grounded evaluation.

Problem

• slide generation
• audience conditioning
• benchmarking
• source-grounded evaluation
• large language models

Innovation

• audience-conditioned generation
• slide generation benchmark
• source-grounded evaluation
• utility-weighted probing
• LLM-based presentation