aiqbench analysis
This research demonstrates that Large Language Models and Visual-Language Models used in autonomous vehicles inherit significant demographic biases regarding pedestrian yielding, mirroring historical human driver prejudices. The study introduces novel testing methodologies to quantify these downstream effects across gender, ethnicity, and socio-economic factors, challenging the reliability of current "common sense" model paradigms for safety-critical applications. Addressing this bias is essential not only for technical accuracy but also for maintaining public trust in AI-driven transportation systems.
LLM-Driven Autonomous Vehicles Inherit Human Driver Biases in Pedestrian Yielding: Results and Implications From A New Benchmark
arXiv:2609.00192v1 Announce Type: new Abstract: Public trust in Autonomous Vehicles (AVs) may depend not only on technical success but also on the fairness of their decision making. While a recent trend in AV research involves using general purpose "common sense" models to guide AV decision making, the degree to which these inherit human biases in driving is still understudied. Given that psychology studies have shown human driver biases exist, such as lower pedestrian-yielding rates to Black pedestrians in the US, we argue that analyses of model bias should also be part of AV evaluation. Concretely, in this paper we propose two new bias testing methodologies for Large Language Models (LLMs) and Visual-Language Models (VLMs)-"All Else Being Equal" tests and "Self-Consistency" tests-in order to assess bias in pedestrian-yielding decisions. Our findings show that both LLMs and VLMs make yielding decisions which are influenced by pedestrian gender, ethnicity, religion, disability, age, skin tone and socio-economic status. While the type and degree of bias is different from model to model, we highlight common patterns-and raise questions about the "common sense" model paradigm, particularly the need to either revise the paradigm or address issues of downstream bias.
Abstract:Public trust in Autonomous Vehicles (AVs) may depend not only on technical success but also on the fairness of their decision making. While a recent trend in AV research involves using general purpose "common sense" models to guide AV decision making, the degree to which these inherit human biases in driving is still understudied. Given that psychology studies have shown human driver biases exist, such as lower pedestrian-yielding rates to Black pedestrians in the US, we argue that analyses of model bias should also be part of AV evaluation. Concretely, in this paper we propose two new bias testing methodologies for Large Language Models (LLMs) and Visual-Language Models (VLMs)-"All Else Being Equal" tests and "Self-Consistency" tests-in order to assess bias in pedestrian-yielding decisions. Our findings show that both LLMs and VLMs make yielding decisions which are influenced by pedestrian gender, ethnicity, religion, disability, age, skin tone and socio-economic status. While the type and degree of bias is different from model to model, we highlight common patterns-and raise questions about the "common sense" model paradigm, particularly the need to either revise the paradigm or address issues of downstream bias.
Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY)
arXiv:2609.00192 [cs.AI]
(or arXiv:2609.00192v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2609.00192 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 296 words
Related coverage
The Growing Proof That Autonomous Cars Save Lives
IEEE Spectrum · Sep 8, 2026
Rivian’s Gambit for Full Autonomy
IEEE Spectrum · Sep 8, 2026
Uncertainty-Aware Motion Planning for Autonomous Driving in Mixed Traffic Environment
arXiv · Jun 10, 2026
Waymo’s driverless cars continue to crash less often than people
The Verge · Sep 24, 2026