“Wildly irresponsible”: DOT’s Use of AI to Draft Safety Rules Sparks Concerns
The US Department of Transportation (DOT) has ignited controversy by utilizing artificial intelligence (AI) to draft regulations that impact the safety of various transportation systems, including airplanes, cars, and pipelines. A recent investigation by ProPublica reveals significant concerns regarding this decision.
Potential Risks of AI in Rulemaking
Experts warn that if the DOT becomes the first agency to employ AI for drafting rules, it could lead to dangerous outcomes. AI is notorious for its tendency to produce inaccuracies and hallucinate fabricated information. Staff members worry that failure to identify these AI-generated errors might result in flawed regulations, potentially leading to lawsuits, injuries, or even deaths across the transportation network.
DOT’s Perspective on AI Implementation
Despite these concerns, DOT’s top lawyer, Gregory Zerzan, remains unconcerned about the potential for AI errors. According to notes from a December meeting, Zerzan argues that the primary objective of using AI is not perfection but rather to expedite the rulemaking process. He stated that rules that traditionally take weeks or months to draft could be completed in as little as 30 days using AI tools like Google Gemini.
Zerzan emphasized, “We don’t need the perfect rule on XYZ. We don’t even need a very good rule on XYZ. We want good enough.”
Staff Skepticism and Concerns
ProPublica’s investigation included interviews with six DOT staff members, who expressed their deep skepticism regarding the efficacy of Google Gemini for this intricate task. They highlighted that DOT rulemaking requires extensive expertise and familiarity with existing statutes, regulations, and case law—knowledge that AI might lack.
During a demonstration, it was noted that Gemini produced a draft missing key components, which a staff member would then need to manually complete. This incident raised additional red flags, as AI-generated errors could have serious implications—affecting everything from keeping airplanes safely in the sky to preventing gas pipelines from exploding.
Political Context and Future Implications
Former President Donald Trump has been a strong advocate for rapid AI adoption across federal agencies, and he appears to be particularly enthusiastic about the DOT initiative. Zerzan indicated that Trump envisions the DOT as a leader in this transformative approach, potentially influencing other agencies to follow suit.
The DOT claims that Gemini could manage 80 to 90 percent of the regulatory writing workload, which raises questions about the future roles of federal workers who may transition to mere oversight positions, monitoring interactions between AI systems.
Google’s Role in AI Regulation
As discussions surrounding AI’s role in government regulation unfold, Google has remained relatively silent regarding the implications of Gemini. However, the tech giant has actively promoted Gemini for governmental use, promising assistance with “creative problem-solving” in critical areas of federal work.
In a December blog post, Google boasted about the DOT being the first cabinet-level agency to transition its operations to Google Workspace with Gemini, suggesting that this shift would enhance the safety and efficiency of the US transportation system. Google is eager to encourage other federal leaders to adopt similar strategies.
Conclusion
The DOT’s decision to implement AI in drafting safety rules has sparked significant concerns among experts and staff alike. As AI technology continues to evolve, the implications of its use in critical regulatory frameworks must be approached with caution and thorough oversight to ensure the safety of the public.
