Mind the Temporal Gap: Evaluating Video Models for Temporal Understanding in Driving
Modern video-language models are increasingly treated as general-purpose video reasoners, yet it remains unclear whether they robustly capture temporal dynamics in safety-critical settings such as driving.
ProofPaper ↗
Key points
- We study this question through controlled ego-motion tasks that require identifying actions from driving videos.
- Across video-language models, we find that standard prompting, visual prompting, few-shot prompting, and chain-of-thought prompting remain unreliable and highly task- and model-dependent, with persistent class biases, although newer models show substantial task-specific few-shot gains.
- In contrast, lightweight LoRA finetuning and simple probes trained on intermediate visual features recover strong motion-discriminative performance, indicating that motion-relevant cues are often present inside the models but are not reliably used by the final video-to-language pathway.
- These results suggest that, in driving, current video-language models remain unreliable temporal reasoners and often behave more like frame-level, bias-driven recognizers than dependable temporal world models.
Sources (1)
- [1]Mind the Temporal Gap: Evaluating Video Models for Temporal Understanding in DrivingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 06:02 PM
Modern video-language models are increasingly treated as general-purpose video reasoners, yet it remains unclear whether they robustly capture temporal dynamics in safety-critical settings such as driving.
We study this question through controlled ego-motion tasks that require identifying actions from driving videos.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 8, 2026Cited but Not Consulted: A Counterfactual Audit of Legal Chain-of-Thought Faithfulness
- Oct 8, 2026Where to Adapt Matters: Layer-Selective Fine-Tuning for Capability Retention
- Oct 7, 2026SLDR: Defending Against Malicious Fine-tuning via Selective Layers Recovery and Dynamic Routing
- Oct 7, 2026Beyond Anonymous Captions: Grounding Character Identity in Video Captioning and Question Answering
- Oct 7, 2026Juno: Taming Predictive Latents for Vision-Language-Action Models
- Oct 7, 2026When Routing Reveals Membership: Privacy Leakage from MoE Router Telemetry