When it comes to regulating pollution from heavy-duty trucks, economic concerns may carry more weight than climate and public health benefits, according to a Northwestern University study.
In new research, Northwestern engineers used artificial intelligence (AI) and other computational tools to analyze nearly 5 million words from public comments submitted to the U.S. Environmental Protection Agency (EPA) during the development of emissions standards for heavy-duty vehicles.
Their analysis revealed a striking pattern. The EPA was more likely to acknowledge comments about costs and implementation timelines as well as comments from the trucking industry than concerns around public health and climate change.
These findings raise questions about whether the regulatory process itself could slow progress toward cleaner transportation. The study also provides a powerful example of how emerging AI tools can help researchers better understand the complex processes that shape government policy.
The study was published on Nov. 1 in the journal Sustainable Cities and Society.
"Heavy-duty trucks make up 5% of vehicles on the road, but they produce 28% of greenhouse gas emissions," said Northwestern's Amanda Stathopoulos , the study's senior author. "Because heavy-duty vehicles make a disproportionate impact, we've been trying to understand why the transition from diesel to electric isn't happening. We decided to look into how the EPA weighs opinions from many different stakeholders.
"We found the EPA gives more attention to stakeholders from the heavy-duty vehicle industry regardless of how substantive their public comments were. And, ultimately, the EPA finalized requirements that were softer than originally proposed."
Stathopoulos is the William Patterson Junior Professor and associate professor of civil and environmental engineering at Northwestern's McCormick School of Engineering, where she leads the Mobility and Behavior Lab . Deirdre Edward, a former Ph.D. student in Stathopoulos' research group, is the study's first author.
Competing comments
Before finalizing major environmental regulations, federal agencies solicit public comments from manufacturers, industry leaders, environmental organizations, public health advocates and other stakeholders.
"The barrier to entry to submit a comment is basically zero," Stathopoulos said. "The process is open to essentially anyone who wants to weigh in on proposed regulations. The idea is the EPA will look at those comments to see if they made an error or missed anything. Based on that feedback, they might revise their plans and make a final ruling that becomes official."
Although diverse opinions help the regulatory process, these groups often have competing priorities. Some emphasize the environmental and health benefits of stricter standards, while others raise concerns about costs, technological feasibility and whether manufacturers have enough time to comply. Understanding how agencies respond to thousands of pages of feedback is a massive challenge.
Analyzing millions of words
To understand the impact of these comments, the Northwestern team examined the EPA's Phase 3 greenhouse gas emissions standards for heavy-duty vehicles. Proposed in April 2023 and finalized in March 2024, the Biden-era regulations sought to establish increasingly stringent emissions limits for new trucks and other heavy-duty vehicles. Although the EPA repealed the standards in February 2026, under the Trump administration, the public comments and regulatory responses still provided a valuable case study for how competing interests shape environmental policy.
With thousands of pages of public feedback, the researchers had plenty of material to examine. Stathopoulos' team analyzed 1,171 unique public comments containing approximately 4.74 million words, along with the EPA's official written responses.
"The EPA acknowledges some of the comments with written responses," Stathopoulos said. "But responses are only legally required if the comment is substantive. Otherwise, they are not obligated to answer."
Rather than manually sorting through the enormous collection of comments and responses, the team developed a computational framework that combined AI with traditional text-analysis techniques.
"If you were to read all of these words, it would take you months," Stathopoulos said. "We really needed a computational approach to make sense of all the text."
Mapping different themes
First, the researchers used computational methods to identify recurring themes in the comments, such as environmental benefits, economic costs and implementation challenges. Then, they used an AI language model to analyze the sentiment expressed in the comments. They also used generative AI to evaluate the substantiveness of each comment, distinguishing between general statements and more detailed arguments.
The team then categorized commenters by stakeholder type and compared their submissions with the EPA's official responses. By examining both sides of the exchange — what stakeholders said and how regulators responded — the researchers were able to identify patterns.
"We could map out the breakdown of different themes," Stathopoulos said. "That allowed us to visually plot out the discourse and pinpoint agreement or disagreement."
Not all arguments receive equal attention
The Northwestern team's analysis revealed significant differences in the issues raised by different stakeholder groups as well as how the EPA addressed those issues. Most notably, the EPA acknowledged comments focused on costs and implementation timelines more frequently than those advocating for climate and public health.
"This pattern persisted even after we accounted for the substance of the comments," Stathopoulos said. "We could see really strong trends there. Basically, it didn't matter what heavy-duty vehicle stakeholders talked about. They always got a response."
Stathopoulos emphasized that this doesn't necessarily mean that the EPA agreed with those arguments or changed its policies in response. Instead, the findings reveal differences in regulatory attention, raising questions about how agencies balance competing priorities. And those differences could have implications for cities working to reduce freight-related pollution.
"One explanation is that regulators are particularly attentive to stakeholders with greater legal resources, who may be more likely to challenge a regulation if their concerns aren't addressed," Stathopoulos said. "Smaller nonprofits and individual citizens don't necessarily have the same expertise to frame legal arguments. Our approach could help agencies recognize meaningful concerns that otherwise might get overlooked simply because they weren't expressed with the right legal language."
The study, "What shapes heavy-duty regulations? A computational case study of stakeholder participation, agency responsiveness and urban freight decarbonization," was supported by RFQ 24-1 of the Health Effects Institute, an organization jointly funded by the EPA and certain motor vehicle and engine manufacturers.
