While the emerging field of artificial intelligence (AI) tools holds great promise, there are serious questions about its capabilities and reliability. This study would identify and explore potential applications of AI tools to air quality for transportation, testing their capabilities and validating the results against known data or information. For this purpose, the study would first identify candidate AI tools and their pros and cons. It would then work with the study panel to identify potential applications of interest to state DOTs, conduct pilot testing of those potential applications and, to the extent feasible, fact check or otherwise confirm the accuracy of the results. The intent is to start with relatively simple tasks and then move to more challenging ones, and in the process identify any limits in the capabilities of the AI tools. A key deliverable of this study is the identification of deficiencies or limitations of the AI tools that would need to be resolved before the tools could be reliably applied for air quality applications for transportation. Finally, the study would develop conclusions and recommendations on the realistic potential for effective implementation of AI tools for air quality. Potential applications may include (but are not limited to):

MODELING
• Model hypothetical projects, including compilation of input data, running the models and generating a air quality report. Compare the results from the AI tools to those generated by air quality modeling specialists for the same projects.
• Review available state DOT guidance documents for project-level air quality analyses and develop a best practices generic template that any state DOT could customize for their respective jurisdictions and use.
• Prepare documentation for hypothetical atypical event and exceptional event demonstrations following EPA guidance.
• Check EPA regulatory emission and dispersion models for possible coding errors and recommend fixes as well as other general improvements. As part of this review, check the coding for MOVES for possible errors that may be causing or contributing to the highly anomalous behavior observed for emission factors versus speed for higher road grades. Also, check the coding for AERMOD for any errors related to RLINEXT applications for noise walls, particularly for receptors near the walls.
• Create a GIS-enabled user-friendly interface for regulatory emission and dispersion models (MOVES, AERMOD and CAL3QHC) that can read in CAD files (AUTOCAD and MICROSTATION)
• Identify and assess options for a next generation regulatory dispersion model for transportation, including consideration of standard gaussian models as well as more advanced numerical models. Develop the code for the preferred option and compile and test it against the field data and regulatory criteria presented in or with NCHRP 25-55.

REGULATORY REVIEW AND POLICY
• Draft comments on proposed regulations. Check proposed regulations for overreach from Clean Air Act (CAA) requirements as well as their potential to delay or limit transportation projects being cleared for air quality for NEPA and conformity.
• Check EPA guidance for project-level analyses for consistency with the CAA and associated regulations and with the Office of Management and Budget (OMB) bulletin on “Agency Good Guidance Practices”.
• Develop and assess technical options that state DOTs can use to check whether AI tools were used for project level analyses prepared for them.

RESEARCH
• Conduct a literature review for research topics of interest to be specified by the NCHRP panel. For example, review the literature for health impacts of fine particulate matter and assess its accuracy and reliability for use in risk assessments for air quality in transportation projects.
• Recommend and prioritize new research topics, covering the range of AASHTO/NCHRP funding programs (full NCHRP, synthesis etc.) and including recommendations for research for each funding level and timeframe specified for the AASHTO TERI database.