A look inside the Trump administration approach to artificial intelligence

Tom Temin The Biden executive orders on AI were fairly prescriptive in how both business, federal contractors, and also the government could use it. And I don’t think there’s anything philosophically all that different from what the current administration may want. It’s got to be fair, it’s got to be accountable, and all of these things. But what do you take from what’s going on?

Neil Chilson Yeah. So, you know, the Biden executive order was sort of unprecedentedly long. And there were chunks of it that were large chunks of it that were essentially directives to agencies to nerd harder, right? Like, do what you’re doing, but do it. Do more of it and do it with a lens towards AI. One interesting thing about this rescinding is that a lot of that report, those reports and guidance are still sitting out there, right. Like the agency’s completed a bunch of these tasks. And so the question will be, as the Trump administration, as new leadership comes into these agencies, how thoroughly are they going to go through what was what was decided and how much of that are they going to repeal or replace or change? And so, you know, the ones that I’m paying the most attention to, our guidance on procurement, as you as you pointed out, there might be tweaks that that the Trump administration will want to make to, you know, the frameworks by which federal agencies need to consider when they choose these technologies. You know, the executive order, although there was some it seemed like there was some shifting in the Biden administration towards the end of the executive order, was very focused on risks of AI and very focused on how do we prevent this from causing the types of harms that maybe Democrats prioritized more than Republicans. And yeah, I think that trickled into the procurement guidelines that that the agencies had set. And so I would expect some of that to go away and for a lot more of it to be about, what are the opportunities that we have here to make government more efficient and more effective. And so you could wipe out the executive order with, you know, just in one day. But the effects, I think, are longer term and will need to. And the Trump administration has some work ahead of it to make sure that it gets its vision for AI embedded in the federal agencies.

Tom Temin And you had a technological job in the first Trump administration, and it’s fair to say at least agree or disagree, they do have an eye on technology and technology deployment, especially in this AI area, because they had an AI executive order in Trump 45, correct?

Neil Chilson That’s right. Some of the earliest, maybe, I’m not 100% sure on this, but I think it might have been the first executive order directly addressing artificial intelligence technology. And this was, you know, obviously well before the Chat-GPT era that we’ve seen now. So I expect that the Trump administration will, while it’s cleaning up the leftovers of the Biden executive order, I think they will come out with their own executive orders on these issues. And we’ve already seen, as you mentioned, in the lead-in, some activity and some an emphasis on the need for the U.S. to maintain its global leadership in artificial intelligence. And I think that will include maintaining our leadership in how our government uses the technology.

Tom Temin But in the meantime, agencies doing AI projects should maintain the same values and best practices they had two weeks ago, with respect to training, with the proper data and not the wrong data, making sure that the data itself doesn’t introduce hallucinations and biases and all these things. I mean, these don’t change. These are kind of eternal values, right, for AI.

Neil Chilson I mean, good good governance is good governance. You want the tools that work, that help you achieve your job, you know? And so I think you’re right, like wanting tools that actually work well, that provide the right answers, not the wrong answers. A big chunk of that is knowing what what are the strengths and the limits of the technology. And and those strengths and limits continue to change as the technology itself continues to improve. And so I think people who are using this in the federal government, they have you know, it’s a tough challenge, but it’s also a really exciting one. These are powerful tools, and using them properly is going to be really important.

Tom Temin We’re speaking with Neil Chilson. He’s the head of AI policy at the Abundance Institute. And what do you make of this $500 billion from SoftBank and Oracle and one other company? A lot of ceremony at the White House to introduce this. What’s the meaning of this, do you think?

Neil Chilson Yeah, $500 billion is a lot of money. You know, the initial pledge is that $100 billion of this is going to be deployed directly and that the rest of it will be invested over the next several years, I think four years or so. I think there’s a lot of questions about how much of this is a prospectus, an investment strategy, more than like a commitment. But it is really interesting that Trump was the one up there announcing this as a great need. And I think he’s right. Infrastructure. This is the data centers, the electricity generation, the cooling mechanisms, the land that you need to build these types of data centers is a really key part of the AI industry. It’s not something we think of. I think we have to think of A.I. as primarily a bunch of the software geeks who are out there coding somewhere, but there’s real hardware involved. We’re talking about big warehouses full of computers that will bring, you know, know blue collar jobs to the places where this investment goes. And so we’re talking I think some of the initial investments are in Texas. And so I think that that should be exciting to the, you know, the regions that are going to get this investment. Like I said, I don’t think this the checks have been written. But but this is this does demonstrate from some of the, you know, the most powerful and capable innovators in this space that they believe that investment in infrastructure is really important. And and having the you know, the certificate of approval from the Trump administration that, you know, the types of permitting that maybe they need to build new energy sources in Texas or wherever, that’s going to help a lot with this process.

Tom Temin What we’re seeing, I think, is an emerging confluence of artificial intelligence and energy policy and environmental policy, because permitting it takes I mean, there are projects that take 25 years to get permitted pipelines and so forth. And then there is the power needed. And we by Energy Department’s own estimates AI could use up double digits of the nation’s electricity supply in the next few years. And then you’ve got AI infrastructure itself. And he didn’t say it directly, but the power needed by these data centers are going to be located at the data centers and there are going to be nuclear plants.

Neil Chilson Yeah, definitely one of the one of the great ways to get, you know, both clean energy that is, you know, very reliable that doesn’t now fade with the weather and that is, you know, sustainable over time. And so and it’s tough to build those things in the U.S. right now, in part because of regulation.

Tom Temin And so but that’s also really a new technology. They’re not talking about building another Three Mile Island, but there’s these totally new technology, very small footprint nuclear generators that don’t look anything like those of yesteryear that you can put on a campus. They’re the size of a, you know, a shipping trailer. Yeah, I think is what they’re talking about now.

Neil Chilson These small modular generation technologies are really impressive. They’re so much safer than, you know, what we think of as the and smaller than what we think of as the, you know, traditional nuclear developments. And so hopefully we can streamline the regulatory system to, you know, be calibrated properly to this new technology in a way that helps us, you know, generate not just the electricity. I mean, energy abundance would be great whether or not AI is, you know, also awesome, right? Like and so cheap energy. Who doesn’t love that? And and so I think that getting these technologies online and getting them streamlined and getting the regulation calibrated to the to the risk and the benefits is really important.