Washington, D.C. – Senate Majority Leader Chuck Schumer (D-NY) today was interviewed at the Washington Post’s AI Summit about his AI Insight Forums, the future of AI, and AI regulation. Below is the interview:
Leigh Ann Caldwell: I am Leigh Ann Caldwell. I am the anchor of Washington Post Live, also coauthor of The Early 202 newsletter. Senator Schumer, thank you so much for being such a great sport—
Schumer: It's great. It's fun.
Caldwell: — and sitting through that, but I want to get your reaction to that.
Caldwell: What are you thinking?
Schumer: Well, we've realized this. I mean, the ability of AI to interfere with our elections, to spread misinformation that's extremely believable is one of the things that's preoccupying us as we go through our actions at our AI group and within the Congress. Lots of people in the Congress are examining this. Excuse me. Sorry.
Caldwell: Oh, it's the flip phone.
Schumer: Mr. Tech.
Caldwell: So you have created a bipartisan working group on this. You have been a leader on this issue, trying to prioritize it. There was just a new forum on Tuesday.
Caldwell: What came out of it, and what happens next?
Schumer: Well, the first thing we want— look, just as Geoff's little example showed, he wants to have a pledge of candidates, and maybe most candidates will make that pledge. But the ones that won't will drive us to a lower common denominator, and that's true throughout AI. In other words, if we— there's so much good that could come out of AI, but there's also so much harm. And if we don't have government-imposed guardrails, the lowest common denominator will prevail. And in our first forum, which had leaders of all different viewpoints: there were tech leaders, and there were people who were sort of anti-AI. Let's curb it.
Caldwell: And to remind, this was the forum with Elon Musk, Mark Zuckerberg—
Caldwell: — and a whole host of—
Schumer: But there was one consensus, that government had to be involved, and that was sort of our goal. We have to keep this bipartisan, and that's why we have a group of four of us sort of leading this, myself, Senator Rounds— he's been a great partner— Senator Young, who I worked on the CHIPS and Science bill with, and Senator Heinrich, who is head of our little AI group in the in the Senate. And we had to keep it bipartisan, but we also had to sort of send a sort of lightning bolt through Congress and the country that we have to do something. The second forum, which we had yesterday, said not only should government be involved, but it needs significant resources, and a broad spectrum of people on the panel agreed that that should be, including our Republican friends. The minimum number that was talked about was $32 billion, and that we—
Caldwell: To invest in for what?
Schumer: Invest—invest in many different things, and that's part of the— part of the goal of the forum to figure out where to invest and where not to, how to encourage the private sector in places to go forward, because with these huge databases, if you want to have more than just a few companies have access, it may be the federal government that has to be involved in either creating the database or at least bringing data together so that people have access. It's so— you need— you need real involvement, federal involvement. And the other thing that came out, Leigh Ann, is you had to move quickly. China is moving very fast here. They're investing lots of money, and they're doing it in a different way. They don't have the same kind of values that we have. You know, they're putting a lot of focus on facial recognition, on following people around. If we don't move forward, it will lead to not only a different dominant system of AI with different values, but we'll fall behind economically rather dramatically. So we've really made great progress in terms of now getting a bipartisan consensus. A, the federal government should be involved. B, we need significant resources. And now we're beginning to figure out the best places to use those resources.
Caldwell: So just to recap, so federal investment to ensure that the United States is a leader in AI, but also at the same time, there needs to be some guardrails around it. So have you all figured out that balance yet?
Schumer: Well, that is one of the key questions. Innovation should be our North Star. We should encourage innovation, but innovation should be on both sides of the ledger; in other words, in encouraging the new cures for cancer, ending world hunger, or dealing with our national security. But it also should be involved in creating the guardrails, and there, government probably has more of a role. The private sector will be pretty good at finding out the good stuff, but in preventing the bad stuff, you really need guardrails. And you need some kind of federal investment in how to figure out those guardrails. You know, one word that's used but it's very important and hasn't— there's not really been the breakthrough on that yet— is "explainability." In other words, when you use AI, when you ask a system to spit back something, they should explain where it came from. Where did that picture of Chuck Schumer come from? Why did he say this? And where did that come from? That can break— that can— if we can really discover, figure out how explainability can be accessible to the average user, you can have major success in creating guardrails and giving the consumer great information as to what they need. I mean, for instance, you know, if there's— if you go online and you find this diet that says eat a lot of hot dogs, oh, and it's very convincing because it's looked at your— what you think and why you like hot dogs and encourage you to eat hot dogs. But then if you learn, oh, all the information came from Oscar Mayer Inc., you sort of know what's going on. That's what explainability is, simply put.
Caldwell: So transparency. Is the United States too far behind? Did they miss an opportunity?
Schumer: We're still ahead. We're still significantly ahead. And, you know, the great thing about America in a certain sense is we're entrepreneurial. We encourage new thinking, and we have many different universities and other places that invest, companies. That gives us a big advantage over China. But if they're going to put tens of billions of dollars into figuring this out and we're not going to do anything— and, you know, the days are over, given competitiveness among companies, when they have— you know, we used to have a Bell Labs. They had a monopoly, and they put a lot of money into pure research to figure out questions that don't immediately affect their profitability. We don't have that anymore. If the government doesn't do it, no one will.
Schumer: So, you know—
Caldwell: Well, we talked about this after the AI forum. I sat with you, and we— anyway, we had a previous discussion a few weeks ago. One of the things that you said then is as far as regulation is concerned, you almost don't want to move too fast. Do you still believe that? Has the EU moved too fast?
Schumer: Look, this is one of the hardest things, maybe the hardest thing that Congress could ever approach. AI is complicated. At one of our forums, Eric Schmidt said, "I don't understand these algorithms." Well, if he doesn't, how are we going to do it?” Second, it affects everything. It's not just one segment of society that's affected. Third, it's rapidly evolving; and fourth, Congress doesn't have the history. You know, you look at defense or health. There's a long history of what's going on and familiarity with the granularity of it, little less so with AI, a lot less so. That's one of the reasons we're having these forums to help provide some of that granularity. So that makes it hard, very hard to do, but we have to do it. If you move too quickly in this, you may screw it up. The EU went very fast, and now they're backing off because they realized they made a lot of mistakes. So that's another— there are so many needles to thread, threading the needle between encouraging the good innovation but having guardrails that prevent bad innovation and not having the one unbalance the other. But another is, do you move too fast? Do you move too slowly? What I have basically said— and I've stuck with this, I think, when I talked to you before, Leigh Ann— it's not going to be days or weeks before we put out proposals. But it's not going to be years either. We have to do it in months-type time, and I would say this: we may not— we have to approach this with a degree of humility because it is so hard. And, you know, I was advised by lots of people, "Don't get involved. It's going to be next to impossible," but, you know, as Theodore Roosevelt said, "We're in the arena. If we don't do it, who is going to do it?" And so we have to approach it in a way that we don't get left behind. But at the same time, we don't move so fast that we screw it up. And I think we're on a good track, and that's why we're going to have two more forums. This is a little news I'll make. On November 1st, we're going to have two different forums in the morning on workforce. One of the great problems people bring up with AI, average folks, "We're going to lose millions of jobs." How do we deal with that issue? We made a big mistake with globalization. Globalization, it may have increased international wealth, but millions of people were hurt by it, and we didn't pay any attention to it at that point. Can't do that. And so workforce is important, and that's positive and negative. How do we train people? Because there's going to be millions of new jobs that are created by AI. But also, how do we deal with people who might lose their jobs because of AI? And the second one is high impact. There are areas right now where AI is having an effect: finance, healthcare, that kind of stuff. And they're having an effect on people. Are they doing it in the right way, or is there bias? Because there's built-in bias in the databases because of decades at least or centuries of discrimination and all of that, and so we're going to have to look at those things. And we will have a forum rather soon on election reform, and that's the one we may have to move the most quickly on because the elections are coming up.
Caldwell: So are you willing to move separately on any sort of election-related legislation overseeing AI, and could it be done before the 2024 election and implemented?
Schumer: That is a great question. In terms of the second, I think we have no choice. We have to do— and again, humility. We can't let the perfect be the enemy good. If we can't solve every problem— you know, watermarking is an easy way to do it.
Schumer: Like Mr. Fowler did up there, he had a big watermark at the top of each of my little astronaut and kittens and all that stuff and—
Schumer: But explainability is harder. Guardrails that prevent real deception are harder. We're going to try to move quickly, whether we move it separately or not remains to be determined. It depends how quickly we can move on other pieces. And I have said, if we can't solve every problem in AI, but we can solve some of them or attempt to solve some of them— again, humility, who knows if we will? Let's hope so. We shouldn't hold back. So we'll have to see. But, you know, by— as I said, by months, we got to have legislation, and the election— need to deal with election reform, and making sure our elections are not so totally jaundiced or perverted that no one believes anything anymore, is a top, quick priority.
Caldwell: Senator Klobuchar has election-related AI legislation. Is that the way to go, or is there other ideas out there?
Schumer: Just a general point. So our forums are not to replace the committees but to augment the committees, to give them the information and the ask— and have them ask the right questions when they do the legislation, but we intend the actual legislation to be in the different committees. And I've had great discussions with Senator Klobuchar. She's a very capable legislator about what could— she has the Rules Committee, and the Rules Committee governs elections. She's the chair, and Fischer is— Fischer from Nebraska is the ranking Republican. And they are talking about what can be done. And, you know, the good news here, Leader McConnell has encouraged his Republican members to participate and get involved. So at the moment, we are not in any kind of partisan loggerheads over this. And I was relieved that yesterday, when the panel seemed to indicate we needed a significant amount of dollars, because there's no way to do this without significant amount of dollars, there was broad bipartisan consensus that we needed it.
Caldwell: The job issue, the workforce issue, there are all sorts of different points of view on how problematic that could be. Where do you stand? Do you have an— do you even have an opinion about it or what is—
Schumer: Sure. You know, sort of like the whole conundrum of AI, tremendous opportunity and tremendous potential problems. You got to deal with both. And, you know, there was some talk, a lot of talk there at our— both the first forum and the second about how to train people for the various jobs. And this is not just PhDs in, you know, computer science, but the— you know, people even with a high school degree or a community college degree who will be needed as AI expands in certain areas, many areas.
Caldwell: What keeps you up at night regarding AI, if anything?
Schumer: You know, everyone talks about the doomsday scenarios, and there are some people who worry about that, and we're going to look at that for sure.
But I guess it's— there is such revolutionary change that is happening so much more quickly than, say, the electricity revolution or the steam revolution, decades— or a century before that, even quicker than the sort of internet connectivity revolution, that the pace of change is so fast, how is humanity going to keep up? That's what worries me, broadly speaking, and you can look at it in specific areas or you can look at it more generally.
Caldwell: Is that something that Congress can do something about?
Schumer: Well, guardrails can help direct us in the right way, but that's why I said humility. This is difficult but so important. Certainly, if we do nothing, the outcomes will be worse than if we do something that's decent.
Caldwell: President— my colleagues reported yesterday that President Biden is going to introduce an Executive Order on— regarding AI and workforce issues, immigration issues on Monday. Is the White House under— and the president— do they realize how important this issue is, and are they doing enough?
Schumer: Okay. First question is the answer is yes, they do. We're in constant touch with them, and they've been very encouraging of what I have been doing and our little group of four has been doing. We're going to meet with them pretty soon, with the president. As for the second question, there's probably a limit to what you can do by Executive Order.
Schumer: We need to fund certain— there are proposals that actually have passed into law to help us deal with AI and— but it hasn't been funded. They can't just come up with the funds. So on the second, they're concerned and they're doing a lot regulatorily, but everyone admits the only real answer is legislative.
Caldwell: There is a Republican House of Representatives within—
Schumer: So I've heard.
Caldwell: They have a brand-new Speaker of the House. I'm not sure if you've spoken with him yet.
Schumer: I did last night.
Caldwell: You did? How was that conversation?
Schumer: Nice. Look, I told him what I've been saying to everybody: "I want to work with you. Let's work together." But the only way we're going to solve any of these problems— AI, we're on a great bipartisan track here in the Senate, but everything, whether it's, you know, funding the government or dealing with Ukraine or Israel or all these other issues in terms, you know, of a big broad bill, you can only do it bipartisan. You can only do it bipartisan. And Boehner and Ryan and McCarthy learned that lesson the hard way, and so I hope that this new Speaker understands that. He cannot get things done if he tries to just be partisan.
Caldwell: Did he seem to understand that?
Schumer: Well, we said we'd talk, so let's hope. It's a little early. I just— you know, it was a short conversation, but I look forward to sitting down with him and trying to work with him.
Caldwell: On AI specifically, former Speaker Kevin McCarthy, he was interested in this issue.
Schumer: He was, yes.
Caldwell: Did you get any indication that this new Speaker has any interest in doing something?
Schumer: Do not. You know, he's a bit of a tabula— not a tabula rasa, but something we don't— I don't know that— I don't know him. I never met him before, and so I can't give you that answer, but it's something we'll explore.
Caldwell: Okay. On as far as funding AI is concerned, another challenge with House Republicans is they are trying to dramatically decrease government funding. Do you think that in appropriations bills this year or an extra bill this year, there could be funding for AI?
Schumer: Yes. There's a broad mainstream of Republicans, and a lot of them realize— you know who tells us we need to go forward on AI, among others? The military and the defense establishment and national security. So Senator Rounds, who's been great on this across the board, but what got his first interest— he's a national security expert— was that issue. So there's— look, I don't think you could look at the— on AI— and hopefully not anything else. I hope not. That's been the problem. You can't look at the whole Republican Caucus as these 30 Freedom Caucus people who just want to cut spending on everything. They're not a majority even of the Republican Party, even though sort of the tail wagged the cat. Or the two tails or whatever it was. I know if it was a leg or a tail that was extra. But in any case, so I think there's going to be mainstream support for this, and particularly, a lot of our more conservative Republicans are sort of similar to me, and in a certain sense, we've been China hawks. And this— if we don't do anything, China's going to get ahead of us. Andreessen was at our thing yesterday, our forum, and he said that. He said that's the one place he thinks we need government involvement because we can't let China get ahead of us.
Caldwell: So that goes back to a question I asked earlier, and that's exactly what I was thinking, if the United States is behind, because did— the Defense Department is usually leaps and bounds ahead as technologically. Is the Defense Department, is the US—
Schumer: We're not— if you talk to our leading experts, whether it's the DoD or NSA or CIA or any of those places, we're ahead, but the gap is narrowing, and we cannot sit, just relax.
Caldwell: I guess I want to go in three different directions, but let me ask you again about—
Schumer: Just go in your computer and go in three different directions.
Caldwell: AI, can I have some help on which direction to go next. There are— is there anything that the EU or any other country has done, specifically that the United States is looking at that can be replicated here?
Schumer: We are looking at what the EU has done. We're looking separately, and I think they have been a little more successful, of what Great Britain has done. And we're trying to learn from their— you know, in a sense, that some of them went first is good for us. We learn what seems to be successful and what their mistakes are. So we are trying to learn from them. But as I said, EU has now had to pull back because they went too fast, and most of the practical companies looking at this said they couldn't abide by these regulations. It would just shut down AI in many areas. And another area that's really important that's difficult, we don't want just a few, you know, tech— a few big companies dominated. We don't want that to happen with AI, and one way to deal with that is open source. That's— and a lot of people are very positive on open source. But how, if you have open source, do you prevent the bad guys from using the same database that will be accessible to good people? I guess can't say "guys." I don't know. And how do you thread that needle? Now, there are some open source advocates— we had some of them at our forum late yesterday— who say you can do that, but that's not so easy, easier said than done.
Caldwell: We just have a few minutes left, and so I want to turn to some news of the day. You mentioned the supplemental, the $106 billion—
Caldwell: — for Israel and Ukraine, Taiwan, some border security.
Caldwell: Do you— are you confident that that will be able to stay intact and move forward?
Schumer: Well, the good thing is Leader McConnell and I are on the same page as the president. In the old days, he would have said, "Okay, that's, that settles it." But, of course, now with the centrifugal forces, particularly in the Republican Party, you're never exactly sure, and so I'm hopeful. It is my goal to keep it intact, and it is McConnell's goal. And at the moment, I think we're in decent shape, but, you know, it's a little too early to make any conclusive judgment.
Caldwell: Okay. And on the war in Israel and Gaza, between Israel and Gaza, there's increasing amount of violence in the West Bank as well. Should Israel stop settlement expansions in the West Bank at this point?
Schumer: Yeah, well, I'm not going to get into the granularity of that. I mean, our United States policy, led by the President, which I support— and I believe Israel supports— says there are three goals right now. One is to eliminate the threat from Hamas. The second is to try to release, get the hostages released. I was— when I went to Israel last week, I sat with the families, and we had a bipartisan group. Senator Romney and Senator Cassidy as well as myself, Senator Rosen and Senator Ossoff. We all cried for an hour. I mean, it was just so horrible. So second is release the hostages. But third is to minimize civilian casualties, and that's a serious problem. I think the humanitarian aid for Gaza that the president has proposed should— some of the right wing is saying take that out of the bill. I think it should definitely stay in the bill, and we'll see where it goes after that. It's a little, you know, down the road to talk about what else could happen. One thing I will say, what showed great potential and might deal with the issue you talked about— it's unclear yet— is the tripartite agreement between Israel, Saudi Arabia, and the United States. And some people speculate that either Hamas or Hamas importuned by Iran, one way or the other, did all this viciousness, this horribleness to take that off the rails. We got to keep that on the rails, and that might have a potential as it goes forward for a broader solution as well.
Caldwell: Is it on the rail still? Is it—
Schumer: Well, we had a bipartisan delegation led by Cardin and Graham go to Saudi Arabia, and they seemed to feel it— you know, that there is still a desire to keep it going.
Caldwell: And one— we just have a few seconds left, but there was a horrible shooting in Maine last night.
Schumer: Oh, God.
Caldwell: Have you spoken to Senator King or Senator Collins yet about that?
Schumer: I spoke to Senator King. He's on his way up there right now. It's just awful. Just awful. And it's premature to talk about it. They don't— haven't even— he's still at large as of an hour ago, and— but it's— it just breaks your heart, this— and we see this all the time.
Caldwell: Senator Schumer, thank you so much for your time today.