The so -called government department Elon Musk (Doge) works on the basis of the basic assumption: the United States should be conducted like a startup. Until now, this meant mainly confused fire and the desire to produce steam. But no deck is complete without an overdose of artificial intelligence, and Doge is no different.
The artificial intelligence itself does not deserve reflex forks. It has real applications and can cause real performance. The introduction of AI in the flow of work is not unexpected by nature, especially if you are aware and you can manage its limitations. It is not clear, however, that Doge accepted any of this nuance. If you have a hammer, everything looks like a nail; If you have the greatest access to the most confidential data in the country, everything looks like input.
Wherever Doge went, and she was towing. Given the coverage of the organization, a lot remains unknown about how exactly it is used and where. But two revelations this week show how extensive – and potentially erroneous – AI aspirations.
At the Department of Mieszkanie and Cities Development, he is subject to studies in AI studies in order to find where HUD provisions may go beyond the closest interpretation of the basic provisions. (The agencies traditionally had a broad interpretative authority when the regulations are unclear, although the Supreme Court recently transferred this power to the court branch.) It is a task that makes sense for artificial intelligence that can synthesize information from gigantic documents much faster than a human. There is a certain risk of hallucinations – especially the model spitting quotes that actually do not exist – but man must approve these recommendations regardless of this. At one level is what is quite good at the moment, in this matter: performing tedious work in a systematic way.
However, there is something fatal, asking the AI model for aid in dismantling the administrative state. (Apart from the fact; your course will vary there depending on whether you think that low -income apartments are social good or are not in any kind of yard.) And he doesn’t really “know” about the regulations, or they meet with the most lovely possible reading of laws, something that even very experienced lawyers will not disagree. It should be conveyed quickly describing what to look for, which means that you can not only work with Refs, but write rules for them. He is also extremely willing to be satisfied, to such an extent that he will certainly come up with things and not refuse to answer.
If nothing more, this is the shortest path to the maximalist looking of the authority of the main agency, with a chance to throw in dispersed nonsense.
At least this is an understandable case of exploit. The same cannot be said about the next AI effort related to the dog. As the Wided reported on Friday, the early Doge recruiter is once again looking for engineers, this time “design reference tests and implement AI agents in live flows in federal agencies.” His goal is to eliminate tens of thousands of government positions, replacing them with agent artificial intelligence and “releasing” employees for seemingly “higher influence”.