PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
June 19, 2026High-Confidence ComputingOpen Access

Assessing the effectiveness of compact language models in repairing code with motor accessibility issues

View Full Paper
Ask AI
Bookmark
Share

Authors

CFCiaron FitzpatrickYLYan LiuSTSon T.

Discussion

Loading...

Member takes

Overview

Randomized trial assesses compact language models' ability to fix motor accessibility issues, suggesting further development is needed.

Key Points

  • The research aims to evaluate if compact language models can detect and correct motor accessibility issues in code.
  • Compared eight compact language models' abilities to resolve motor-related accessibility errors in code.
  • Utilized four prompt variations for each code sample to assess the impact on outputs.
  • Employed three large language models to evaluate and score the outputs.
  • CLMs corrected code errors with at least one prompt achieving success 40% of the time across multiple metrics.
  • Success appeared inconsistent, with models rarely performing well on all prompts for a given metric.
  • Prompt variations had minimal overall impact on the quality of outputs.

Cite This Study

Fitzpatrick et al. (2026) studied this question.

synapsesocial.com/papers/6a34de4165a5b0777af2dbdahttps://doi.org/10.1016/j.hcc.2026.100414
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Turning manual web accessibility success criteria into automatic: an LLM-based approach2024 · 45 citations
  2. 2On the Effectiveness of LLM-as-a-Judge for Code Generation and Summarization2025 · 21 citations
  3. 3Accessible or Not? An Empirical Investigation of Android App Accessibility2021 · 74 citations
  4. 4Don’t Confuse! Redrawing GUI Navigation Flow in Mobile Apps for Visually Impaired Users2024 · 7 citations