Ezra Klein: prediction

Transcript context

…It is so hard to get rid of people. Don’t you need to start with the chainsaws? Maybe the next Democratic administration rehires what’s needed over — I don’t think you start with the chainsaw this way because one, you’re breaking things right now that AI’s not about to fix. I think you have to have a very low opinion of the federal government and what the people in it do. Maybe we have different opinions of them, probably, to think this is all going to be . . . you’re not going to break things that were needed. I worry that a lot of things I’m seeing them break, like people who I know who worked at the Consumer Financial Protection Bureau on financial scams, I don’t think knocking them out of government is going to create a macro problem in the US economy, but I do know these people were getting money back for consumers who had gotten scammed. Frankly, I think the capacity to scam people on AI is about to go much higher. I think that a good thing that was happening is going to happen less well right now. My gut is that AI, as it emerges into the kind of view and flexibility where we could integrate it into core systems — which it’s not at yet in most cases — is going to create places where you need to expand and places where you need to contract. That’s why I was trying to ground this in an actual example, like drug development. If it is throwing off way, way, way more good candidates for drug development, yes, there are parts of the system that can contract, parts of it they’re going to need to expand. Energy, again, if you read any of these projects or any of these slightly more utopic visions of AI, which I certainly hope we’re going to live in, the idea is that it’s going to throw off a bunch of realizations and discoveries about how we can make much more energy-dense materials, and so lay down everything from solar panels to get better at doing advanced geothermal. That would mean some parts of DOE maybe become less important. It would also mean our ability to permit and do other things becomes more important. I am not sure, particularly at the beginning, what you need is less so much as the ability to reallocate. Now look, your view that you do mass firings and then you hire back what you need — I guess I understand where you’re coming from on that, but I want to see the thing we are doing the firings for working first, not fire people and hope that Sam Altman’s view of billion-dollar companies with three people is actually going to work at the Social Security Administration in the year of our Lord 2027. You know full well Steven Teles’s work on kludgeocracy. We have at least 50 years of the kludge that is just accumulating. Everyone in government knows about this. It’s not some secret. Various administrations have tried to address it. Al Gore did a bit — that was fine, but it didn’t really stop it. I don’t see that there’s any other recipe besides quite a bit of disruption. Again, there’ll be future administrations to sort it out. Like the New Deal agencies — they weren’t so great to begin with. They didn’t have experience or data or staffing, but over time, pieces fell into place. Maybe the options are just more and more Steven Teles’s kludgeocracy or we take some chances today and do some things that actually hurt. Then over an 8- to 12-year period sort it out with AI and most of all, with future administrations, I don’t really see what the alternative is.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.