DOGE’s AI: A Shortcut to Smarter Government?

0:00Doge, the Department of Government Efficiency inside of the United States has announced they've created an AI tool to cut down on regulation. Today on the podcast, I'm going to be breaking down examples of basically where they've used this in the past, where they see this going, because they say they think they can cut down on 50 % of all regulation in the United States. Is this going to be dangerous? Is this the end of us having to wear seatbelts? Where does this go in the future? I'll be diving into all of that. And I think this is probably not a topic that's going to be super political, as I think it's a fairly bipartisan concept that the government should be spending less money on conservative causes or liberal causes.

0:38I think a lot of people just want the government to balance the budget to a greater extent and slow down the insane spending. I hope that's bipartisan. If not, if that triggers you, you don't have to listen to it. But I think this is really interesting how AI is hopefully going to make the government not go bankrupt as fast. So we're going to be talking all about that today. The one thing I wanted to mention before we do is if you want to try any of the AI models that I'm talking about on this podcast all the time, I'd love for you to try out AIbox.ai. Every single, you know, basically the top 40 AI models are all hosted on here.

1:10It's one price. It's like$20 a month. So instead of having to pay$20 for every single platform, you get access to Anthropic, Cohere, DeepSeat, Google, Meta, Microsoft, Mistral, NVIDIA, OpenAI, Quen, X, a bunch of image models, a bunch of speech-to-text, a bunch of text-to-speech models, tons of stuff. And you can chat with them all in the same thread. So you can ask Cloud a question, it answers. And you can ask OpenAI a question, it answers. And if you're like, man, I have this question, like what's the best AI model to ask it to? You can regenerate the response with Grok, OpenAI, Claude, and just see who answers it best.

1:41No more, you know, having to listen to the benchmarks to decide, you know, what's the best. You can test it out for yourself. So go check it out. It's AIbox.ai. There is a link in the description. All right, let's get into what's going on with the Department of Government Efficiency. So a lot of this is coming out of an article by the Washington Post, which basically is titled Doge Builds AI Tool to Cut 50 % of Federal Regulations. This is pretty interesting. Basically, what's going to be happening is there's this PowerPoint presentation that I don't know if it got leaked or shared or it got out basically.

2:16I have to say, I think it is absolutely hilarious to have a PowerPoint presentation from one of these agencies explaining like why they need to use this tool. I'm assuming that the Department of Government Efficiency had to like make this to pitch it and be like, come on guys, please let us use this to cut your regulation. Probably pitching it to Congress who's like loves spending money and hates it, but whatever, that's funny. The title of their pitch deck is Doge Deregulatory Opportunity, July 1st, so not too long ago. And this is, I love it. It's like a business pitch deck. They have the problem and the opportunity.

2:51So the problem they say is$3.1 trillion compliance costs a year, which is about 12 % of GDP. So that's a lot of money that we're spending on compliance costs. They say$1.2 trillion of reduced investment regulation estimated to lower external investments by$650 billion and lower corporate reinvestments by$550 billion. I mean, they're kind of doing the runoff thing where it's like, if you spend a dollar, then that person takes the dollar and they keep 30 % and they spend, reinvest 70%. And then that company takes it and reinvest 70%. So I mean, they're kind of like doing the trickle down effect there too.

3:26But I mean, it's not a lie. And it's probably a sound economic principle. So interesting what the opportunity is, they say. 55 or 50 % of all federal regulations can be eliminated, yielding a potential$3.3 trillion per annual value opportunity. All right. Is it possible or is that crazy? That's what we're going to be debating on the show today. But first, I want to say, what are they using AI for in all of this? And what is the challenge that they're trying to solve? Basically, the challenge they say is that there's four steps you have to take when you want to remove a regulation. Number one, you have to research what statutory is required, and then you have to basically select what to keep.

4:05You have to then draft the regulatory submission documents. You have to analyze and respond to about 100 ,000 comments. And then you have to draft the final rule, including the comment responses. So the average hours required per regulatory section to do all of that is about 36 hours. And, you know, there's 100 ,000 sections, it would be 3.6 million man hours to deregulate 50 % or 100 ,000 sections, a lot of time, like basically, some people say it's impossible. So their solution is this new Doge AI program, which they say saves about 93 % of man hours. So how it works is it creates a regulatory slash statute database.

4:47and then it determines which regulations are required. It enables the agency to comment and modify the results in a final delete list. They are then automatically drafted. They automatically draft a submission document for the attorneys to edit. They automatically analyze 20 to 500 ,000 citizen comments for the final rule conclusion. And it's really, really, yeah, honestly, it's kind of crazy. So a lot of people ask, has this been used? Is this new? And they actually had a case study. So they said for the CFBP, 100 % of deregulations were written using this new tool. They said for the HUD, the policy team completed decisions on 1 ,000 regulatory sections in two weeks, which is crazy.

5:34And they said they vetted and endorsed. This has been vetted and endorsed by Doge lawyers. Whatever. They say the lawyers that endorsed it, but I don't think that's super relevant. Anyways, they want to roll this out to two more people. So I think this is really fascinating. It's a really interesting concept. Basically, they say that it's going to analyze about 200 ,000 federal regulations, and then it's going to identify ones that aren't required by law, right? Because I think what happens a lot, whether it's Biden or Trump signing executive orders, and of course, there's like lawsuits and federal judges and like all of this mayhem that happens about laws and rules and Congress weighing in on things, but things are changing all the time.

6:13But we have this massive logbook of regulations that aren't always directly correlated to some sort of act by Congress or some sort of law, but maybe like tangentially. And it's hard to know like when the regulation disappears or when it appears. So basically, this is going to go clean out all of the books. It's going to eliminate half of those is the goal. And their goal is to eliminate half of those in the next, I think, six months. So they said by the first anniversary of Trump coming into office. so January they're hoping to have all of that done which would be insane um it seems that they've already been doing this now right as I mentioned the Department of Housing and Urban Development wrote 100 % I used it to basically write 100 % of their deregulations at the Consumer Financial Protection Bureau they have been doing a lot with that the White House basically was asked about this because I don't think this is really supposed to be like a public thing it was kind of like an internal document and PowerPoint stuff.

7:10So the White House was asked about this. And their response was interesting. They said, quote, no single plan has been approved or greenlit. And then of course, they said, you know, the Doge team is, quote, the best and brightest in the business. So I think what they don't want is for everyone to be like, oh my gosh, AI is running the government. Like, it's completely run away with everything. But like, I guess the question is like, would it be so bad? Like, I don't know. It's so tricky, because I know everyone thinks like AI has biases or can hallucinate or has issues. And like, there are causes for that.

7:39You can mitigate a lot of those challenges. I kind of love this use case of using AI in the government. Like if you could try to make it less biased, that's fantastic because obviously every single congressman and congresswoman has their own special interests. They have people that support them. And I'm sure if you're a Democrat, you think it's just the Republicans. And if you're a Republican, you probably think it's just the Democrats. I don't really care. I think everyone has special interest groups that fund their campaigns and they're sort of beholden to them. That's just kind of the nature of politics.

8:07And so it's sucks when you're trying to deregulate something and then maybe some guys like well you know this oil and gas company gave me a ton of money for my campaign so i don't want to do anything that would negatively impact them or maybe this electric solar panel company gave me a lot of money for my campaign so you know i don't know i don't want to do anything that would hurt them whatever right you can imagine uh both sides of this argument from both sides of the aisle but at the end of the day maybe there's a regulation that really just needs to be removed regardless of the company it probably it potentially hurts it could unlock a lot more for the economy and the country as a whole.

8:36And so I think AI would be the perfect tool to go in and use this without having any sort of biases or special interests. I think this could be amazing, theoretically. And are there issues with this? Could this be hijacked? Are there flaws? 100%. But same thing with people. I don't think it's any different. This just could speed up the process, just make us more efficient. So I'm very interested there. I'm bullish on the concept. I'm sure there's going to be a lot of debate on whether this is good or bad. But overall, I think I'm bullish on the concept. And I would love to have AI rewrite all bills by Congress to basically be three pages long that everybody in America could read before they get passed instead of having them be 2 ,000 or 50 ,000 pages long.

9:16It's just outrageous. So I would love AI to be applied to a lot more in government and I think it would actually be quite useful. In any case, thank you so much for tuning into the podcast today. I hope that this was interesting and sparked some new ideas and concepts for wherever you work or whatever you do. And if it did, the number one thing I would appreciate is a review on the podcast. I read them all. If you are on Apple, drop some stars, leave a comment. And on Spotify, you got to hit the about tab, I think, and you can drop some stars on YouTube. Subscribe, like the video. I really appreciate all of you guys.

9:49Make sure to check out AIbox.ai and I will catch you in the next episode.