Trump Administration Official Says Quiet Part Out Loud on AI-in-Government Plans

The US Department of Transportation (DOT) is planning to use Google’s large language model, Gemini, to draft federal transportation regulations, with the aim to outsource up to 90% of the usual regulatory work done by legal and policy experts.
DOT intends to be a leader in using artificial intelligence (AI) for speeding up federal rulemaking, potentially adopting a “good enough” approach rather than focusing on perfect or well-crafted regulations.
Despite the potential benefits of AI in handling technical tasks and generating high volumes of text, concerns exist about AI’s inability to produce error-free, well-reasoned, and evidence-backed regulations due to “hallucination” and bias reinforcement.
Overreliance on AI-generated regulations could lead to legal issues, as the rules may be invalidated in court if found to be factual or logical errors, contradicting evidence, or implausible.
The DOT’s approach raises questions about the legal obligation to provide a thorough analysis and consider public input in the rulemaking process, questioning whether AI can fulfill these duties.
ProPublica’s report highlights the need for closer scrutiny of DOT’s future regulations, the potential for litigation against DOT for rubber- stamping AI-generated proposals, and the importance of transparency regarding the agency’s use of AI in rulemaking.