Do you like hopping on an airplane or driving your car? If so, are you fond of rigorous safety regulations that make those things relatively non-deadly? Well, the Trump administration dares to ask, what if—instead of having experts carefully develop those regulations—we just let Google Gemini do it?
Yes, over at the Department of Transportation, whatever employees who remain after the administration’s purge of federal workers are now being told to use Google’s glorified chatbot to write brand-new safety regulations.
Transportation Secretary Sean Duffy
And if you were wondering whether there was a plan to ensure that these new AI-fueled regulations would be good, the DOT’s top lawyer, Gregory Zerzan, is just operating on vibes.
“We don’t need the perfect rule. We want good enough,” he said.
While Zerzan might just want “good enough,” most people actually do want the perfect rule when it comes to transportation safety. Regrettably, as ProPublica described, Zerzan “appeared interested mainly in the quantity of regulations that AI could produce, not their quality.”
Zerzan explained his approach as, “We’re flooding the zone.”
Does it engender confidence to have the DOT lawyer invoking Steve Bannon’s strategy of overwhelming your political opponents?
Zerzan apparently told DOT employees that “it shouldn’t take you more than 20 minutes to get a draft rule out of Gemini.”
But DOT staffers were also told that they could draft rules in a matter of seconds or minutes, rather than having pesky experts take up a bunch of time yammering about safety.
This is, to be blunt, terrifying. It’s not just the blind faith that somehow Google’s glorified chatbot is more knowledgeable than an entire agency of specialized experts, but also that complex safety regulations can just be generated in seconds.
But that’s what you get when you put people like Sean Duffy in charge of a government agency.
Related | Where do people fit into an AI future?
This fetishization of AI isn’t limited to the DOT. After firing basically everyone, the Trump administration seems to think that AI will replace everything. That’s pretty convenient, as it also allows the administration to shovel millions of dollars into tech companies that, in turn, give millions of dollars to Trump.
Last year, the administration proposed using AI to find “fraud, waste, and abuse” at the Social Security Administration. That didn’t happen, but what did happen was some random so-called Department of Government Efficiency tween sharing your private Social Security data with a group that wants to overturn election results. Terrific.
Google CEO Sundar Pichai and Elon Musk are seen at President Donald Trump’s inauguration on Jan. 20, 2025.
DOGE also deployed AI tools to review Department of Education grant information, including the personal data of everyone who managed those grants.
Better still, it used a non-government AI tool that it accessed through Microsoft’s cloud service. So safe, so secure.
Similarly, after firing thousands of Internal Revenue Service employees, the Trump administration is having Salesforce’s AI do their jobs instead. That will definitely work out.
But perhaps nothing is worse than the military going all in on Grok, Elon Musk’s child sexual abuse material-generating machine, with Defense Secretary Pete Hegseth burbling on about how Grok will also be fed information from sensitive intelligence databases.
Surely nothing will go wrong there.
The Trump team’s AI push is dumb and shortsighted, but hey, at least it’s also incredibly dangerous and expensive!
