Don’t fail me! The Level 5 Autonomous Driving Information Dilemma regarding Transparency and User Experience
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper investigates the "information dilemma" in Level 5 autonomous driving, where vehicles operate without human control and may exhibit unexpected behaviors due to machine learning imperfections. The authors address the challenge of maintaining positive user experience (UX) and acceptance when systems fail in non-critical situations, such as misclassifying traffic signs or pedestrians. While transparency is generally recommended for AI systems, the study explores whether providing explanatory information during these failures mitigates negative user reactions or exacerbates them by highlighting system limitations. To examine this, the researchers conducted a mixed-method study comprising a quantitative online survey (N=113) and a qualitative Wizard of Oz on-site experiment (N=8). Participants interacted with a simulated Level 5 autonomous vehicle via a conversational user interface that provided real-time feedback on driving situations. The simulation included three normal driving scenarios and three failure scenarios where the vehicle incorrectly interpreted the environment (e.g., stopping for a poster resembling a stop sign). In the online study, participants were divided into a control group (no option for additional info) and an experimental group (could request detailed explanations). The on-site study allowed all participants to request additional information while verbalizing their thoughts. Data were collected using standardized questionnaires measuring UX, trust, transparency, and the subjective feeling of control. The results reveal a significant information dilemma. While increased transparency generally improved overall user experience and acceptance, providing additional explanatory information during failure scenarios did not mitigate negative effects. Instead, users who received detailed explanations during failures reported a lower subjective feeling of control compared to those who did not. Furthermore, users significantly more often requested additional information and expressed a desire to take over the driving task during failure situations, despite the impossibility of doing so in a Level 5 context. The control group, lacking the option to request more information, showed half the desire to take over compared to the experimental group. The study concludes that neither a fully opaque design nor a highly transparent design with additional explanations optimally handles non-critical failures in Level 5 autonomous driving. Transparency improves general acceptance but can undermine the user's sense of control when failures occur. The authors suggest that a "human-in-the-loop" approach or careful design of explanatory information is necessary to balance transparency with user comfort, as current methods may inadvertently increase user anxiety and reduce perceived control during system imperfections.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-18 |
| archive | success | openalex | — | — | 5 | 2026-06-25 |
| extract | success | cached | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-20 |
| chunk | success | chunk | — | — | 1 | 2026-06-20 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-20 |
| enrich | success | openalex | — | — | 1 | 2026-06-20 |
| promote | success | — | — | — | 1 | 2026-06-18 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-20 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Empirical Findings: self report data