Will Artificial Intelligence increase inequity problems in transportation?
- Jul 14, 2021
- Focus Area
- Transportation Systems Management & Operations
- Submitting On Behalf Of
- TRB Committee on RTSMO
- Critical - High Priority
- $250,000 - 499,000
- 1 - 2 years
- Type of Research
- Full Research Project
- Date Posted
- Jul 14, 2021
- Not Funded
Extensive work is underway implementing machine learning/artificial intelligence capabilities in traffic management systems (TMS). AI promises to improve the speed and accuracy with which TMS respond to data and increase the volume of data TMS are able to process - ultimately contributing to safer more effective mobility. Some of the actions controlled by TMS which impact safety and mobility include: traffic signal timing, ramp metering, dynamic sign messaging, variable speed limits, lane control systems, tolling, high occupancy lane systems, traveler information, diversion routing, data interfaces to other systems, and others. In addition to real-time operations, the data from TMS are important source information for planning.
Bias is inherent in Artificial Intelligence and must be carefully accounted for in real-world AI implementations. If the data under observation by an AI solution are flawed, AI systems will learn, propagate, and amplify these flawed behaviors. Notable examples include a 2018 MIT report that found gender recognition algorithms were 99% accurate for white men and 65% accurate for dark-skinned women and Google Translate's highly publicized gender bias in translating from gender-neutral languages (he is a doctor, she is a nurse, he is hardworking, she is lazy). The research question here is, "As transportation agencies seek to improve safety and mobility through the inclusion of AI capabilities in Transportation Management Systems, what are the risks that existing equity problems will be ignored or amplified? What steps can be taken to ensure advancements in AI benefits to all road users?"
Additional Supporting Information
The original author of this idea is:
Mike Haas IBI Group 802 2nd Ave Ste 1400 Seattle, WA 98104
Mr. Haas is the co-Subcommittee Chair for the Data for Operations Subcommittee of the Regional Transportation Systems Management and Operations Committee of TRB (ACP10)
- Submitted By
- Thomas Jacobs
- Center for Advanced Transportation Technology
This idea has no attachments.
Join the Discussion
You can make your thoughts known by commenting, and by rating the idea and other comments. All you need to do is sign in, or register for an account with us!