Research map / Large language models / AI and society
Thesis Summary: Operationalizing User-Inclusive Transparency in Artificial Intelligence Systems
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Doctoral thesis summary proposing user-inclusive transparency for AI systems via transparent UI prototypes (selective vs. exhaustive modes) for an auto-text generator and a clinical diagnosis system, to be evaluated by user surveys.
Problem
AI systems are black boxes with uncertain predictions and biased behavior, so users lack trust; existing explanations do not always align users' mental models with how the system actually works.
Approach
Represent an AI system as an amalgamation of the model, its data, and a user interface with visual interpretations, and let users control how much explanation they receive via selective or exhaustive transparency modes grounded in usability principles.
Key results
- Preliminary findings report users rate an AI system more positively when it provides explanations, even if the explanations are placebo or simply persuasive.
- Planned: surveys of a controlled user group to assess whether faith in the system increases with the transparent UI versus non-transparent UI.
- Timeline spans Fall 2023 to Spring 2024 with prototypes for an AI text editor (Class A) and clinical diagnosis software (Class B).
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Provide explanations for the AI system's decisions, including placebo or persuasive explanations | AI systems · User evaluation | user ratings | Helped |
| Provide explanations that increase understanding and bring the user's mental model closer to the system's actual working model | AI systems · User evaluation | transparency | Helped |
Surprising findings
- Users rated the AI system more positively when it gave explanations even when the explanations were placebo.
Limitations
- Results are planned, not yet collected; evaluation is a proposed survey of a controlled group.
- No implemented system or quantitative results are reported in the summary.
Experimental setup
Metrics: user trust user satisfaction
Baselines: non-transparent version of the user interface
Builds on: Shneiderman usability guidelines progressive disclosure mental model theory
Closest papers by meaning
- Making Transparency Advocates: An Educational Approach Towards Better Algorithmic Transparency in Practice (AAAI 2025 · similarity 0.83)
- Position: Explainable AI Cannot Advance Without Better User Studies (ICML 2025 · similarity 0.81)
- Making AI Policies Transparent to Humans through Demonstrations (AAAI 2024 · similarity 0.81)
- AI Agents Push Humans Out of the Loop (NeurIPS 2026 · similarity 0.80)
- Agent S: An Open Agentic Framework that Uses Computers Like a Human (ICLR 2025 · similarity 0.80)
- Trustworthy AI Must Account for Interactions (NeurIPS 2026 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.