YouTube as a Source of Information in Understanding Autonomous Vehicle Consumers: Natural Language Processing Study
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This study investigates public attitudes and concerns regarding autonomous vehicles (AVs) by analyzing user-generated content on YouTube, addressing a gap in systematic content analysis of video-based public opinion. While traditional surveys have measured AV acceptance, they often fail to capture real-time, unstructured feedback. The researchers aimed to determine how consumers view AVs and identify polarities regarding video content and automation levels. The motivation stems from the need to understand end-user perceptions before mass deployment, leveraging YouTube’s large user base as a source of instantaneous public sentiment. The methodology involved collecting data from the 15 most-viewed YouTube videos related to autonomous vehicles, which accumulated approximately 60.9 million views and 25,629 comments. The videos were manually clustered by content type (marketing, proof of concept, comparison/test, and violation) and automation level (Levels 2 through 5, based on SAE International standards). The researchers employed Natural Language Processing (NLP) techniques, specifically text mining and sentiment analysis, using R software packages. They utilized Term Frequency-Inverse Document Frequency (TF-IDF) to identify significant trends and rare keywords within the corpus of seven million words, allowing for the discovery of key themes and sentiment variations across different video categories and automation levels. The findings reveal that key issues in comment threads centered on efficiency, performance, trust, comfort, and safety. Sentiment analysis indicated mixed but predominantly positive sentiments toward AV technology. However, the perception of safety and risk increased significantly in textual content when videos presented full automation (Level 5). TF-IDF analysis highlighted distinct linguistic patterns: Level 2 videos were associated with terms like "autopilot" and "Tesla," while Level 5 videos featured words such as "blind," "Google," and "broken." Videos categorized as "comparison/test" received the highest like-to-dislike ratios, whereas the single "violation" video (depicting an Uber car running a red light) had the lowest ratio and contained strong negative language. Generally, lower automation levels were better received than higher levels, suggesting greater public comfort with partial automation. The significance of this study lies in demonstrating that social media mining offers a viable, low-cost alternative to conventional surveys for capturing real-time public opinion on emerging technologies. The results suggest that while interest in AVs is high, concerns about safety and loss of control persist, particularly regarding fully autonomous systems. The disparity in reception between partial and full automation levels implies that public acceptance may be gradual, tied to the degree of human involvement. These insights are crucial for automotive manufacturers and policymakers aiming to facilitate the transition to autonomous vehicles by addressing specific consumer fears and managing expectations through targeted communication strategies.
Key finding
Sentiment analysis of YouTube comments indicates that while public opinion on autonomous vehicles is mixed with positive sentiments slightly exceeding negative ones, concerns regarding safety and risk are more pronounced in discussions about fully automated vehicles.
Methodology
dataset
Sample size: 25629
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. Discovered via author_sweep_intake on 2026-05-27.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | author_sweep | — | — | 2 | 2026-05-27 |
| archive | success | canonical_url | — | — | 7 | 2026-06-06 |
| extract | success | cached | — | — | 3 | 2026-06-10 |
| clean | success | clean | — | — | 1 | 2026-06-04 |
| chunk | success | chunk | — | — | 1 | 2026-06-04 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-04 |
| enrich | skipped | — | — | — | 3 | 2026-06-04 |
| promote | success | — | — | — | 1 | 2026-06-04 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 2 | 2026-06-10 |
| tag | success | vector_similarity | — | — | 15 | 2026-06-11 |
| verify | success | — | — | — | 2 | 2026-06-10 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-10; 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