← POLITICO Tech
POLITICO Tech
What OpenAI wants from Trump
00:21:43
Two years on from the release of ChatGPT, OpenAI has been beefing up its efforts to shape the way governments manage artificial intelligence. And earlier this year, the company tapped Democratic political veteran Chris Lehane to oversee global affairs. Now, Lehane is laying out proposals for how the U.S. government should invest in data, energy and chips to stay competitive with China. On POLITICO Tech, Lehane joins host Steven Overly to explain the company’s agenda and expectations for the incoming Trump administration.
Loading summary
Transcript31 lines
- [00:00]
Snapchat Announcement
Presented by Snapchat. Stay connected with new location sharing features coming soon to Snapchat's family center. Snapchat's in app Parental tools that help families navigate the digital world together.
- [00:15]
Stephen Overley
Hey, welcome to Politico tech. Today's Friday, November 22nd. I'm Stephen Overle. Next week, we'll mark two years since ChatGPT exploded into public view and thrust the company that made it OpenAI into the headlines. Since then, Washington has become increasingly fascinated with and concerned about generative artificial intelligence. And OpenAI has beefed up its presence to shape exactly how the government goes about regulating it. Earlier this year, the company hired Democratic political veteran Chris Lehane to oversee global affairs. And lately, Chris is making the case that keeping the US Competitive with China will require Washington to invest in AI infrastructure, a combination of energy chips and data. On the show today, Chris tells me what OpenAI thinks those investments should look like and how he plans to navigate a Washington that will soon be controlled by Republicans, including an incoming president whose most outspoken advisor, Elon Musk, was an OpenAI co founder who has since become a big critic. Here's our conversation. Chris, welcome to Politico Tech.
- [01:40]
Chris Lehane
Thank you, Steven. Thanks for having me. I feel like I'm, you know, I've really arrived now to be on this podcast.
- [01:45]
Stephen Overley
Right. It's the hottest place to be in tech, that's for sure. You know, Washington right now, we're obviously anticipating a major change with a new administration coming in. In particularly, I think we're expecting a lot of change in how the government is looking at AI. And I think there's an expectation maybe of less regulation and less focus on safety, perhaps more of a focus on things like national security and competitiveness. What are your expectations?
- [02:13]
Chris Lehane
Yeah, I think, I think the issues you were touching on really tease it up. I think we saw this over the course of the presidential campaign. Both candidates, frankly, were talking about AI to the extent that they did much more in the context of national security and the imperative of ensuring that US Led AI would continue to be the case and that being a real priority, I think for President Elect Trump, this is something where I think you can potentially see it playing out at a couple of different levels. I think first and foremost, doing what the administration that will be coming in can do to make sure that the US does continue to lead just on the basic innovation and the reason for that. It's really clear there's a binary choice out there. The world is either going to. There's two countries in the world that can build AI at scale, one is the US the other is the People's Republic of China. So I think first and foremost is just really that mindset that we do need to continue to lead on the innovation. I think secondly, this is an administration or certainly people around the President elect that has talked about this idea of sort of Manhattan style infrastructure projects. And I think what that really means and we can get into this deeper and more, but at a high level, I think what that really means is can we do any number of big AI data center clusters that will help generate the compute that we will need to continue to maintain our lead. And then I think the third piece of this countries succeed when they're able to really marshal their existing resources, resources needed to really for democratic AI to prevail. For that US led AI to prevail are going to encompass some version of chips, data and energy. All three of those things ultimately comprises that infrastructure. In particular, I think we're at a moment now where you can see an incoming administration and frankly the politics in D.C. and particularly up on the Hill, coalescing around the idea of actually really thinking about the big infrastructure here, but particularly the energy piece.
- [04:35]
Stephen Overley
That's where we want to pick up on the energy component. Because when we did hear President elect Trump speak on the campaign trail, he often brought up kind of the energy consumption of AI and sort of made this point that we'll need more energy to power data centers and to stay competitive for OpenAI. What's your pitch to this administration on that front?
- [04:57]
Chris Lehane
Yeah, well, first of all, I think it's really simple. Infrastructure is destiny. Like if you want the US to win and lead on AI, that's foundationally an infrastructure project. And look, you know, we've had these conversations. I mean, frankly, we were meeting with both campaigns, you know, over the course of the summer and the fall. Now with the election results in right this as a company, we continue to engage with the President Elect's team and folks in and around that team. So this is a conversation that we've already been having. Yeah, we actually released something last week called the infrastructure blueprint. And I'll just do it really, really, really fast because there are five basic components, some I've actually sort of alluded to. One was can we create economic hubs, states, expedite permitting process, they get some compute back what we just talked about. Secondly, do we think about in the same way that Eisenhower did the national highway and defense bill to create an interstate highway system in the 50s, can we do that with our transmission lines like we do need to optimize and prioritize a transmission system that's actually built for moving AI quality volumes of data around the country, both for national security purposes, but also for commerce, so everyone can participate. Third, how do we think about government buying compute? Government is going to need to buy compute because they're going to need to be building up their own AI systems. Again, national security is going to be an imperative for this government right now. Beginning to commit to buying compute, particularly if it's targeted to some of these key parts of the country, effectively creates a magnet to attract the roughly $175 billion in dry capital, dry powder that is looking to invest in AI infrastructure from around the world. Fourth idea was creating something that we called the North American AI Compact that was like minded countries and nations that support the US vision for AI coming together to work together, particularly on supply lines. You could also think about energy as part of that. And then the fifth piece was this idea of we do need to begin to revitalize or at least reconsider how we think about nuclear power. We obviously have traditional fission, but you also have incredibly interesting innovations coming on fusion right now. And the example that we used is right now, based on public reports, there's roughly 100 US nuclear powered submarines in the water all around the world. Those are what are called SMR, small modular reactors. US Navy has been doing this for 75 years. If we can put these things on submarines that are traveling often in fairly turbulent water conditions and maybe not always the most safe areas like we should be able to figure out how to do that here in this country. And you know, I know the US Navy has an awful lot on its plate, but you know, is there a way to tap into their expertise? So those are sort of the five things that we listed that, you know, if you sort of took those and put them together, could really be a blueprint to accomplish this. AI infrastructure build out.
- [08:00]
Stephen Overley
It sounds expensive I guess is one thing that comes to mind. Or it sounds like it would require some investment, at least from the federal government. And when I think about the incoming administration and Republican controlled Congress, you're going to have a lot of budget conscious lawmakers who are concerned about federal spending.
- [08:20]
Chris Lehane
So can I jump in on this? Let me push you on that.
- [08:22]
Stephen Overley
If I could not, I'm curious how you convince them.
- [08:24]
Chris Lehane
Yeah, yeah, yeah, yeah. So first of all, I think if you actually look at the economics around this, like I'm pretty confident that this will score out in the way where you generate a lot more revenue than you are spending. I mean, just think of these economic hubs. Like what you're basically doing is asking folks to expedite a permitting process to attract the private sector capital that then pays for the bill. And then you take that second part, which is this idea of doing the US potentially agreeing and committing to buying compute. The US is going to have to do that anyhow. It does need compute, right? I mean, national labs, DoD, various three letter agencies, they all need access to the AI compute to support their various needs. This is not a question of if. This is only a question of we have to do this and so can we do it in a way that actually helps catalyze both infrastructure and bringing broader growth. So people do benefit from this in a more shared way. So I didn't mean to interrupt, but I actually think that as you think about this, it actually is going to generate economics and not cost economics.
- [09:30]
Stephen Overley
Got it. So it may require some investment upfront in some ways, but you see the payoff.
- [09:36]
Chris Lehane
That's exactly. We use this phrase that like look, this is a country that has done really big things, right? We've built out electricity, right? We built out the arsenal of democracy. You know, we built out the Internet. Look at specific projects. The TVA in Tennessee, the Michigan plan which was public private building out Detroit, you know, at the beginning of the 20th century, like the Research Triangle in North Carolina which involves public sector universities, the federal government state. Like we have built really big things. And I do think that this country is really well served by thinking big, acting big and ultimately building big in this area.
- [10:23]
Snapchat Announcement
Stay connected with new location sharing features Coming soon to Snapchat's Family Center. Snapchat's in app parental tools that help families stay connected. Request your teens location and share yours back to help keep everyone in the family up to date on each other's comings and goings. Review privacy settings together and have conversations about what choices work best for your family.
- [10:48]
Stephen Overley
Let me ask you, there's an expectation that this incoming administration is going to be a lot less focused on things like bias and discrimination and safety than what we've seen from the past four years. We've heard some Republicans, for instance, refer to that as woke AI. How will that affect the way that OpenAI approaches those concerns about things like bias or discrimination?
- [11:12]
Chris Lehane
Yeah, great question. I mean we've spent a lot of time as we've built out our models seeking to make sure that we're addressing the bias issue right, so that they're. You're not getting sort of Biased results. Now, this is always an ongoing project and ongoing work. Given the complexity of this, I also think it's really important to understand, recognize. I know you do, and probably a lot of people listening to this do. But this is like, AI is not social media. It's a different type of technology. Right. People are not using it in the same way they've used social media. So there's a lot of things that you can do with the tech to allow individuals to, in effect, personalize it. Right. In effect, to be able to have agency over how they engage with it, including even from ways that you can sort of personalize it. So it's serving things that are consistent with your particular worldwide perspective or philosophy or ideology. And I just think that's. I know a lot of time when these conversations take place, there is a natural tendency to sort of compare this to the last big technology thing. But this is not social media. And because it's not social media and because people use it individually, there's a lot of things that you can do to make sure that you effectively can personalize it to what you're looking for and how you want to use it.
- [12:32]
Stephen Overley
I want to talk a little bit about your background, and you talked there about kind of comparing to pathways of technology. And you worked in the Clinton administration. You were here working in Washington in the 90s during a time where there was a lot of focus on Internet policy. Right.
- [12:45]
Chris Lehane
The World Wide Web, baby.
- [12:47]
Stephen Overley
Right. The World Wide Web. Right. Which was transformational then. AI is transformational now. I mean, how do you compare the two, particularly in terms of the way Washington is kind of talking about tech and the regulation of it?
- [13:00]
Chris Lehane
Yeah, fascinating question. You know, I know this is a podcast so people can't see us, but you're right. Way back when, when I actually had a lot more hair, I did work at the Clinton White House. I had some minor involvement with the 1996 Telco act, which ultimately became the framework that really did help position the US to be the epicenter of what we then. I was joking, but we actually called it World Wide Web, which we now know as the digital economy. And then I was there for Web2. I came out to the Bay Area in the early 2000s, ended up being very involved in what was then called the sharing economy. Companies. And now, if you think of this as the next chapter, maybe Web three. And I would say the big lesson for me is I think we're all better off if at the US Government level, there is close engagement between the public sector and the private sector and in particular developing an overall strategy for how we want this technology to impact and shape our society, our economy. And each one of these has gotten bigger than the other one. I do think what was and look at maybe this is the old guys trying to think about the past, but I do think that I am a history major in college. I do think the past is important to learn from. In 1996, these and ours, Democrats and Republicans came together and really did work on that Telco Act. And as I referenced, they had a very strategic imperative and that was to guarantee that the US would lead into this new economy, into this new age. When the effort began, there was talks about, well, should we regulate the Internet by the FCC like it's a radio station or a television station? The decision was made. No, we actually need to regulate this in a way that is where we're optimizing for the US to become the center of this. I think as we sit here today, artificial intelligence is on the scale of electricity. It is going to change how we live, it's going to change how we work, it's going to change how we talk to each other. It's going to change even how we play. We need to start thinking about it at that level. And I do think that that ultimately means there does need to be an overarching strategic perspective on that. If one thing really galvanizes D.C. galvanizes the political system, you know, it is the imperative of making sure that the US Prevails over China and this issue intersects squarely with that bipartisan consensus.
- [15:46]
Stephen Overley
Is that how you plan to navigate it? I mean, you know, your background, like we said, you worked in the Clinton White House. You're well known in Democratic circles as political entity. It is a Republican controlled Washington come January. So how do you navigate that?
- [15:59]
Chris Lehane
Yeah, as I touched on earlier, I mean, personally for me, like, I do think this is an issue that really does transcend partisan politics because of the stakes are so big. You know, in Washington, Democrats and Republicans have worked together on issues that, you know, really implicate U.S. economic competitiveness and U.S. national security vis a vis China. And I think if you're thinking about AI and how the US makes sure that it maintains its lead and frankly wins on this. OpenAI is going to be in the middle of that conversation.
- [16:28]
Stephen Overley
Right?
- [16:28]
Chris Lehane
We're leading on the tech. This tech is going to continue to progress, I'm sure, like as a company, we're going to do a lot of things and we've already done these things that, you know, smart companies that are engaging in Washington, D.C. do we have people from both sides of the aisle that work with us? And we've been having these conversations, as I noted, you know, even before the election. So I do think, you know, this issue really does, does transcend politics, and I think that's how both sides are going to approach it.
- [16:53]
Stephen Overley
Chris, you know, you're, you're a political mind, and so I've got to ask. Elon Musk has proven to have a lot of sway with this incoming administration. And I don't have to tell you that he's been quite critical of OpenAI and of Sam Altman, your CEO. And there are lawsuits right now, which is not what I want to ask you about. But I wonder how much that relationship worries you considering Elon Musk's kind of ties to and influence with this incoming administration.
- [17:21]
Chris Lehane
Yeah, I mean, look, Elon is a unique figure, and we're also in that unique moment in terms of what he's doing and building out there. I think for us, we control what we can control. And what we can control is the fact that we're winning and leading on the innovation. And the US Is going to want to be able to benefit from that. You know, deploy it, leverage it. It's pretty simple for me, which is if the US Wants to win and prevail in this contest with China, then OpenAI is going to have to be in the middle of that conversation. And, you know, I fully expect that an administration that, you know, has really made the importance of the US Winning against China across a whole array of subjects, but including AI is going to be eager to work with the folks who are really leading on this or certainly amongst the leaders on it. I also think, and we've spent a lot of time talking about this, we're going to be in the middle of massive infrastructure in this country that's going to be producing jobs. And this is an administration and the President elect who is particularly interested in building things. And so I think we work what we're developing. Building here is relevant both for the national security, but also for economic activity and prosperity here at home.
- [18:42]
Stephen Overley
Right. I mean, I do hear that point. The devil's advocate in me has to say that didn't stop him from attacking Jeff Bezos or Mark Zuckerberg or some of these other tech titans who have also sort of been big economic drivers.
- [18:57]
Chris Lehane
Yeah, I think they'll that we. And again, there are other players out there, but we are so much at the center of this transformative technology. I want to be enormously respectful of some of the folks that you referenced. I do think there are qualitatively different things at stake here than maybe some of the examples that you just mentioned. I also think, and I won't get into the details, but others can go, you know, those folks, you know, had some of their own individual stuff that may have been going on. Right. We're in a situation here where, you know, we've made really clear and have been working with both sides of the aisle for some time. That. And we do that because, again, not to be repetitive, but it bears repeating, this issue does transcend partisan politics, you know, period, full stop. If you want to win, like, we're going to be there with you and we want the us to win, great.
- [19:50]
Stephen Overley
Well, listen, Chris, it was great to catch up and appreciate you being here on Politico Tech.
- [19:54]
Chris Lehane
Steven, it's awesome. I hope I get an invite back at some point. So thank you so much.
- [20:02]
Stephen Overley
That's all for today's Politico Tech. Politico Tech is off next week, but I'll be back in your ears on Monday, December 2nd. Until then, catch up on episodes by subscribing on Apple, Spotify or your preferred podcast player. And for more tech news, subscribe to our newsletters, Digital Future Daily and Morning Tech. Music in our show comes from the mysterious Breakmaster Cylinder. Our managing producer is Annie Reese. Our producer is Afraid Abdullah. And our editors are Steve Heuser, Daniela Cheslow and Louisa Savage. I'm Stephen Overlay. See you back here on December 2nd.