Elon Musk’s so -called Department of Government Efficiency (Doge) works according to a core underlying assumption: The United States must be managed like a start. So far, it has mostly meant chaotic shooting and an eagerness for steam regulations. But no pitch -deck in 2025 was completed without an overdose of artificial intelligence, and Doge is no different.
Ai himself does not deserve reflection pitch. It has sincere uses and can create sincere efficiency. It is not inherently unpleasant to put AI into a workflow, especially if you are aware of its restrictions. However, it is not clear that the nuance has embraced any of the nuance. If you have a hammer, everything looks like a nail; If you have most access to the most sensitive data in the country, everything looks like an input.
Wherever the dog is gone, AI was on tow. Given the opacity of the organization, it remains very unknown about how exactly it is used and true. But two revelations this week show how extensive – and potentially wrong – Doge’s AI aspirations are.
At the Department of Housing and Urban Development, an undergraduate university had the task of using AI to determine where HUD regulations could exceed the strictest interpretation of underlying laws. (Agencies have traditionally had broad interpretative authority when legislation is vague, although the Supreme Court has recently shifted the power to the judicial branch.) It is a task that can actually make sense of AI, which can synthesize information from large documents that are much faster. There is some hallucination – more specifically that the model spits out quotes that do not exist in reality – but one must approve these recommendations, regardless. It is at one level that is generative AI at the moment quite good: to do boring work in a systematic way.
However, there is something harmful to asking an AI model to help dissolve the administrative state. (Apart from its fact; your mileage will differ, depending on whether you think that low-income housing is a social property, or that you are more of a not in a backyard.) Ai knows nothing about regulations, or whether they will not agree with the strictest possible reading of statutes, something that even many will experience. A quick outline must be given to set out what you have to look at, which means you can not only work the refs, but write the rule book for it. It is also extremely eager to please, to the point that it will confidently make things up rather than refuse to respond.
If nothing else, it is the shortest path to a maximalist authority of a great agency, with the chance of scattered bullshit thrown in for a good benchmark.
It is at least an understandable use case. The same cannot be said for another AI attempt associated with Doge. As Wired Friday reports, an early Doge recler is again in search of engineers, this time to design “standards and deploy AI agents about live workflow in federal agencies.” His goal is to eliminate tens of thousands of government positions, replace them with agentic AI and “free” workers for seemingly “higher impact” duties.