Sounding the Alarm on AI: A Conversation Across the Political Divide | Civil Discourse

[FULL TRANSCRIPT BELOW] As President John F. Kennedy once said, “What unites us is greater than what divides us.”

We live in a time when viewpoints are highly polarized, even to the point where finding common ground seems impossible. We believe this is a recipe for societal disaster.

But is meaningful discourse even possible in this day and age? As a starting point, we’re bringing together decision-makers from across the political spectrum to explore issues where enough common ground can be found to have meaningful discussions.

We are inspired by George Washington, Abraham Lincoln, and John F. Kennedy—people who united the country during times of great division and unrest, and forged a path forward.

Join us as we chart a new course with this special series—a joint production of The Epoch Times and NTD.

In the first episode, Epoch Times senior editor Jan Jekielek and NTD host Steve Lance sit down with Rep. Ted Lieu (D-Calif.) and Rep. Ken Buck (R-Colo.) to discuss the dangers of artificial intelligence (AI).

What role should Congress play in regulating AI development? Should America have an agency focused on regulating AI? If so, how do we prevent it from being captured by powerful corporate interests?

To what extent should AI be used in the military? Should AI be able to launch missiles or weapons? How can the United States maintain its technological advantage over the Chinese Communist Party, a regime that does not have the same moral or ethical scruples as the United States when it comes to technological development?

What ethical responsibilities do tech companies have in deploying AI? What if a child goes on an AI platform and asks about the best way to commit suicide, or a nefarious character asks how to create a lethal virus?

We discuss all this and more in this episode of Civil Discourse.

FULL TRANSCRIPT

Jan Jekielek:
Today we welcome Republican Congressman Ken Buck and Democratic Congressman Ted Lieu, and our issue is AI artificial intelligence. This is Civil Discourse.
Congressman Lieu, it’s great to have you here. Congressman Buck, great to have you.

Rep. Ted Lieu:
Thank you.

Rep. Ken Buck:
Thank you.

Mr. Jekielek:
There seems to be a near consensus view that AI needs to be regulated in some way, and there are very different viewpoints on how that might happen. You have introduced legislation to deal with that. One bill has to do with not having autonomous AI-launched nuclear missiles. The other bill is the National AI Commission Act. I want to understand your views on how this regulation is supposed to happen. Let’s start with Congressman Lieu.

Rep. Lieu:
Let me first say it’s an honor to be here with my friend and colleague, Ken Buck. We work together meaningfully on a number of different issues. On artificial intelligence we both believe the same thing, which is that AI has helped society and will continue to move society forward. We want America to be in the lead on artificial intelligence. At the same time we want to mitigate any significant risks that could harm people. That’s why we have introduced a national AI commission bill to make recommendations to Congress as to how we look at what kinds of AI to regulate, and how we should do so moving forward.

Mr. Jekielek:
Congressman Buck?

Rep. Buck:
I appreciate working with Ted. We’ve worked on antitrust legislation together, and now artificial intelligence, along with many other issues. Ted was the president of the freshman class for the Democrats the same time I was president of the freshman class for the Republicans. We’ve known each other and worked together a lot, and this is an honor.

Jan, I would like to move in a slightly different direction. Our job isn’t to regulate, our job is to pass laws and then have the administration regulate. Those regulations would be much more precise. What we’re trying to do is create a broad landscape for the regulators to work within. Ted and I agree that the idea is to have a light touch from Congress, so that we are identifying harms and helping the administration, whichever party is in power at the time, to be able to regulate with specificity. We’re trying to make sure that we are encouraging AI, encouraging innovation, and not putting a wet blanket on something that this country is really at the forefront of right now. We need it to stay that way.

Steve Lance:
Congressman Lieu, I’d like to ask you, not too long ago there was a big AI summit on Capitol Hill that was drawing a lot of attention. Elon Musk was there, and he called for a federal department of AI. You’re one of three members of Congress with a background in computer science. What do you think of such a department?

Rep. Lieu:
I am a recovering computer science major, that is correct. One of the reasons that Ken Buck and I are doing this National AI Commission Bill is because it’s going to be fully transparent. We’re going to know what experts they consulted, what these people said, and how they got to their conclusions. It’s great that the Senate is doing these closed door hearings, but we don’t actually know what’s being said behind closed doors. It’s not really helpful to the American public if 27 AI experts talk to Ken Buck and me, because no one knows what they told us. We also think this commission bill is helpful because it’s going to be fully transparent.

In terms of how we go forward, one of the reasons we’re doing this bill is there are people far smarter than us who can think about this at a very deep level for a number of months, get the data, then write a report and give us some recommendations that we can accept or reject, or we can at least have a transparent baseline from which to start.

Mr. Jekielek:
Transparency is obviously incredibly important in this piece. Congressman Buck, you’ve been very vocal with your concerns about antitrust issues with some of our Big Tech companies which are at the forefront of doing all the AI development. A number of these Big Tech companies also have development in communist China, and there is a whole technology transfer element that is involved. How do you respond to this reality that the core AI development that is being done in some of these big companies has exposure to the Chinese regime?

Rep. Buck:
Given what Communist China is doing with cyber crimes and hacking and other activities targeting the United States, it’s really important to me that we stay far ahead of communist China with our AI products and development. At some point in time it’s going to become obvious that countries that are doing business in communist China and helping communist China develop their artificial intelligence are going to have to choose which side they’re on. I’m hoping that we stay far enough ahead so that when they develop new ways to hack into our businesses and personal information, we have an AI system in place that can counter that and keep Americans safe.

Mr. Lance:
Congressman Buck, just to follow up, you mentioned China in passing. The State Department Global Engagement Center recently released a report saying how the People’s Republic of China seeks to reshape the global information environment. When it comes to dealing with our biggest competitor or adversary, depending on how you look at it, with China in the AI realm, how do you see that playing out in the future?

Rep. Buck:
We’re certainly going to have to be aware of what China’s doing and careful about how they interact with our country and our systems. One of the things that Russia has done for the last 30 or 40 years is to try to undermine democracies by putting enough money into an election to create doubt in people’s minds, especially the people who’ve lost the election, whether the election was legitimate.

China is sending fentanyl to this country to try to create social strife in this country. I have no doubt that they will use AI in the way that Russia is engaged in elections and the way that China’s engaged in fentanyl to try to tear this country apart and its fabric, and we just need to be stronger than that. But they will definitely use their technology, not just in a military sense, but in a way that creates strife and tension in our country.

Mr. Lance:
Congressman Lieu?

Rep. Lieu:
I’m going to answer your question with a story. Last year I talked to a CEO of a company in my district who was making computer chips for AI that were faster, and he also happens to be an immigrant. We were talking, and then at some point I said, “How much faster are these chips?” He goes, “50,000 times.” I was like, “Whoa.”

Then I got to thinking about if he would have gone to Moscow to try and do this. No, because they’re going to shove you out a 12-story window if you say something bad about Vladimir Putin. Would he have gone to Beijing? No, because they’re going to reeducate you and kidnap you if you say something bad about Xi Jinping.

He came to the United States because he knows that we have a free, open, and democratic system. We have the rule of law, we have talent, and we will let him execute his idea and try to change the world. I believe we’re still going to be ahead of our peer competitors because of the constitutional republic that we have. The best way to remain ahead is to keep it a constitutional republican, to lead by example, and to keep a free, open, and democratic society.

Mr. Lance:
Congressman Lieu, I’d like to get your thoughts on that as well, especially as AI evolves into deep fakes and with our election system.

Rep. Lieu:
Sure. The reason that Congress passed a bipartisan CHIPS and Science Act last term was to bring back chip manufacturing to the United States. That is important to AI development as well, and we want America to continue to be the leader in this kind of technology. I also support export controls against the CCP. It’s important that America continues to be in the lead against China, as well as other countries, when it comes to this important technology.

In terms of deep fakes, one of the best things I want people to get out of this is to just be skeptical of what you see on the internet. Do a double check, because the video you see could be totally fake, and the audio you hear could be completely fake. With AI now, if you give it half-an-hour’s worth of your voice it’s going to sound pretty much like you. Give it two hours worth of your voice, and your family members will believe it’s you. We just want people to have some dose of skepticism when they see videos or audios that might seem a little bit off.

Mr. Jekielek:
Congressman Lieu, I was reading some of your statements around AI. Your specific interest is in issues around systemic discrimination that might arise from AI being deployed. You’ve said that you don’t want to see it used against any group of people, whether it’s facial recognition or loan approval. We have seen examples of, and this has been documented in a number of court cases now and in the Twitter files, a kind of algorithmic discrimination against certain viewpoints. How do we mitigate against that?

Rep. Lieu:
That is a great question, and in terms of AI it gets particularly complex. Because at a very basic level, we don’t actually know what it’s doing. Essentially you can feed an algorithm a million pages of dogs, and it figures out what’s a dog and what’s not a dog. But we don’t know how it goes from point A to its conclusion that this picture is a dog, or it’s not a dog, and we can’t really trace it.

I’ll give you a famous example. There were these researchers who were trying to figure out, “Can we diagnose tuberculosis from x-ray scans?” When they developed the technology it seemed to go really well. When they deployed it, it totally failed. They realized afterwards that the AI was detecting if this x-ray scan was from a newer machine or an older machine. Because it turned out that people in western countries with newer machines didn’t get tuberculosis.

But they didn’t know that, because they couldn’t figure out what the AI was actually pattern recognizing on. That’s why it is a big problem trying to figure out how we can try to regulate these kinds of biases if we don’t even know how the algorithms get to the conclusion. Again, that’s why we think a national AI commission of blue ribbon, bipartisan experts would be helpful.

Mr. Lance:
Congressman Buck, I want to get your thoughts on the Federal Department of AI proposed by Elon Musk. I’m sure big government is not something that you’re fond of. In the process of figuring out AI and how to legislate, how do you make sure that you don’t get into a situation that a lot of people think we’re in right now with something like Section 230?

Rep. Buck:
That’s a great point. One of the things that Ted and I have been working on is to make sure that we don’t just take a snapshot like we did with the internet, pass some laws, and then Congress just disappears. One of the purposes of this commission is to continue to look at issues like Section 230 and make sure that whatever laws we pass and whatever regulations are passed are effective and up to date. We don’t know what AI is going to look like in 10 years. We didn’t think 10 years ago AI would look like what it does now, so the commission gives the flexibility and continuity for the oversight that’s really necessary here.

But you raise a great point. The federal agency will undoubtedly be captured by the businesses that care the most about it, and that’s why they don’t want the antitrust division and the Federal Trade Commission in charge of AI. Microsoft has not had good luck with the antitrust division in the last 30 years, and Amazon, Facebook, and Google have not had good luck with the FTC or the antitrust division, so they want to create something new that they can capture. The reason they capture it is they have a lot more interest in who gets appointed to that than consumers do. Consumers don’t unite and target a new agency the way a business whose life depends on that agency will target it, and so I am not in favor of an agency that is just focused on AI.

Mr. Lance:
Congressman Lieu, I will give you a chance to respond.

Rep. Lieu:
I am agnostic as to exactly how these regulations happen, whether it’s one federal agency or we empower different federal agencies in their fields they already have. What I do know is there’s a lot of AI in different sectors, some of which could actually kill us. If your laptop malfunctions, it’s not going 50 miles per hour. But there’s a lot of AI in moving objects; planes, trains, automobiles. We just need more regulators who are more attuned to the unique aspects of AI, so they can figure out how to prevent the most significant harm. I’m agnostic as to how we get that to happen. That’s one reason we have this National AI Commission to give us recommendations.

I do want to follow up on something Ken Buck said, which is that the commission just doesn’t issue one report and leave. No, we would require it to issue an initial report within six months after it becomes law, and then another report a year later, and then a third report two years later, particularly because of how fast AI is moving and changing.

Mr. Jekielek:
There is often legislation passed by Congress that becomes outdated, but still has the force of law and isn’t updated. You mentioned that this technology is developing incredibly quickly. Congressman Lieu, would you support sunset clauses in legislation that is passed, so that it has to be revisited?

Rep. Lieu:
That’s a great question. There’s a point where it will become too late to regulate AI, just like how social media developed quickly and Congress got in there too late. There’s a point where it could be too early as well. If you were to say, “Next week we want you to regulate AI,” I don’t know how we would define it, because it’s changing so quickly.

Again, that’s why we should have an AI commission. Not only is it transparent, it gives some passage of time to see how AI actually affects society. Do all the harms that we think it’s going to visit on society actually happen? Maybe they don’t happen, or maybe it’s some random thing we never thought of. We’re still in too early of a stage, because I don’t even know how we would define it right now.

Mr. Lance:
Congressman Buck, I want to get your thoughts on this. There’s the question of job disruption and job displacement for millions of Americans. There have been studies done that estimate up to 18 million jobs could be replaced. What do you say to that, in this process of developing artificial intelligence?

Rep. Buck:
Yes, that is one of the key questions that we have to answer. The reality of the harms that Ted and I are talking about is that now a kid can get on the internet and ask, “How do you commit suicide? The AI gives them different ways of doing it. This is a terrible result. The companies are now working on ethical delivery of AI to make sure that those kinds of harm aren’t out there.

Then there’s the kind of harms that we see that are nation-state harms. How does a department of defense and a military use AI in attacking? There are also the unforeseen areas, and that’s really the challenge that this commission has to deal with. All of us have to really be aware that we don’t know what will be out there in five or six years with AI in terms of the harm that can happen.

Mr. Lance:
Congressman Lieu, one example is robotics. We’ve seen that in the BMW plant in Spartanburg, and this was over 10 years ago, where it basically replaced the factory line. Can we foresee this with other positions and jobs? How big of a concern is this to you?

Rep. Lieu:
When it comes to AI I want people to understand, you don’t have to understand how to program AI or even how the algorithms work to use it, because it’s going to come to you in the form of products. For example, I don’t know how to write a word processor software, but I use a word processor like most people use word processors. You’re going to be given these products that you can use. I want people not to be scared of AI or developing technology, because those are going to be packaged for you in a way that you can use it quite easily.

You just have to learn to use it, and I encourage people to try out different products and applications. That’s the way it’s going to be delivered to you, just the same way you have this amazing computer in your pocket called a cell phone. It’s amazing technology, and you don’t have to really know how it works to use a lot of the different applications on it. That’s how AI is going to come to most people, and I encourage people to not be fearful of it.

Rep. Buck:
If I can add something to that, one of the dangers that we are anticipating, at least I’m anticipating, is that our country is really evolving into more of a have and have-not country. The folks that have the education to be able to write code to be able to understand how AI can be used are really going to accelerate in terms of their wealth and power and control. The folks that don’t are really going to fall behind.

I’m afraid that our education system may not keep up and make sure that we have the strong middle class that we do now that really sets the United States apart from many other countries in the world. We’ve had a strong middle class for 150 years in this country and it is one of the things that keeps us together. If AI does displace a lot of workers, and if AI really does enhance that difference in our country, it’s something that we’ve got to be very careful about.

Mr. Jekielek:
You mentioned ethics earlier, and you touch on this in your no autonomous nuclear launch bill that you introduced. During the Cold War, there was an escalation around nuclear weapons, with basically a mutually assured destruction doctrine. We have communist China where there really are no ethical boundaries by which the folks over there are going to be constraining themselves with regards to AI. Here, we hopefully do have significant ethical boundaries by which we will be constraining ourselves. Yet, there may be an accelerating arms race. I would like to get both of your perspectives, and let’s start with Congressman Lieu. How do we deal with that reality?

Rep. Lieu:
That’s an interesting question. In terms of AI, you could actually use AI to counter other kinds of AI. Right now there are companies working on watermarking technology that can tell you if a picture you see is the original picture or if it’s an altered picture. There are potentially ways for AI to be used in a way to help you figure out if something is fake or if it’s not true. But that’s a difficult question, because you’re right, there are going to be bad actors, there are going to be bad countries, there are going to be criminals, and there are going to be hackers using it.

Again, being somewhat skeptical of AI is healthy. People also need to understand that in this new world you’re going to have a lot of folks trying to hack into your data. You need to be practicing better cybersecurity practices and to be more aware that there are bad actors out there who are going to try to steal information from you. You need to take certain steps to protect yourself.

Mr. Jekielek:
Autonomous weapons systems are a very robust area where AI development is happening right now. You can imagine an escalation of an arms race where AI would respond very quickly. Shouldn’t we really have a human decision-maker in there in this process? Is that reasonable? Maybe we should lower our guard over here to be able to deal with that threat.

Rep. Lieu:
I see. Both Representative Buck and I have come to the conclusion that we don’t think AI should ever launch a nuclear weapon by itself, and we would have that view whether or not China or Russia or other countries do that. We think it would be insane if they do that, but we’re still of the view that the United States should never get to that place where we just let an AI launch a weapon by itself.

Rep. Buck:
Yes, that’s right. I would even go a few steps below a nuclear weapon, because if we have an autonomous system that sends a conventional missile into a foreign capital, a foreign major city, it starts something that humans might want or might not want. There’s no way of anticipating enough of the factors with AI to really trust AI in initiating a conflict.

Responding to a conflict that exists is a different story, but when you initiate a conflict, or respond within moments to some other country initiating a conflict, there are issues. Our bill deals specifically with nuclear weapons. I would even go a step further and make sure that our military is prepared to deal with non-nuclear or conventional weapons.

Rep. Lieu:
To add to that, fundamental and inherent in this technology is the way it’s designed, because we really don’t know what it’s doing and we can’t trace it from point A to its conclusion. This AI algorithm could be correct 99.999 percent of the time, but that is not good enough if it’s controlling nuclear weapons. Because there could be this one random scenario where it goes totally bonkers and we would not know. That’s why it’s so important to have a human in the loop in this area. Now that Rep Buck mentioned what he just said, we’re going to maybe look at other bills to introduce as well.

Mr. Lance:
To bring it an even further step down, not with just nuclear and other missile systems, but even with drone systems or autonomous robots that can do door-to-door urban combat, is there a concern that this type of technology can actually level the playing field with foreign countries that are adversaries to the United States, like Iran?

Rep. Buck:
It can do more than level the playing field, it can level buildings and all kinds of things. We’ve got to be careful how we use it and how we empower AI. At some point we hope there is a switch that can be turned on and turned off. If we don’t have a switch that you can turn on and turn off, then you get into the sort of sci-fi movies that we’ve all been seeing in the last few decades.

But it’s really important in the concept of the bill that Congressman Lieu and I introduced to make sure that there is a human in the chain of command for nuclear weapons. What you’re talking about, we absolutely also need it for non-nuclear weapons. We could send a drone someplace as long as we have control over the target of that drone. Should we trust the drone so that we don’t have to maintain some sort of communication with the drone? That’s something the military is going to figure out. It really goes beyond my capability. That’s why Ted and I feel so strongly about having a commission.

Mr. Jekielek:
When it comes to AI, let’s take it a little further with an autonomous car that’s driving. Let’s say it causes an accident and kills some people. That is definitely going to happen, and we already know they make mistakes. Who is responsible in those situations?

Rep. Lieu:
That is a fascinating ethical question. In the United States, approximately 40,000 people die every year in car accidents, and most of that is human error. Autonomous vehicles are going to reduce that number, so it’s going to save lives. Now, what if the autonomous vehicle manufacturer knows though that if they spend another six months on their technology, it can better identify certain risks and not result in accidents that would harm passengers or their driver. Do they wait six months to deploy it or do they deploy it right now because it’s already better than a human who’s driving? I don’t really know the answer to that. It’s an ethical question that people smarter than me probably should think about.

But there are these questions that are going to be raised where you ask, “Do you deploy the technology immediately once it’s better than a human driver, or do you try to make it as safe as possible, or something in between? I’m just raising those issues, and I don’t really know the exact answer to those questions.

Mr. Jekielek:
Congressman Buck, have you thought about this?

Rep. Buck:
The answer is no, but as Ted is talking I am thinking about it. Undoubtedly, what we’re going to have to rely on is some sort of insurance process where if you have an autonomous vehicle, you have that vehicle insured. If that vehicle creates an accident, causes an accident and damage, you’re going to have something to cover it with. But to have uninsured motorists with an autonomous vehicle is a headache that none of us really want to start thinking about.

Mr. Lance:
There are obviously so many questions right now, as we can tell as we’re brainstorming, when it comes to figuring out in our own country how to deal with AI. On a global scale, what does collaboration look like with other nations?

Rep. Buck:
Yes, with our allies we are collaborating now on the military side, and we’re collaborating on the business side in commerce. That’s really something that the government should encourage but not get in the middle of. The business world is going to take off in ways that we can’t imagine at this point. The government, in trying to control things, often really stifles innovation, and we don’t want that to happen.

There is a good collaboration at this point. My fear is that countries in less developed areas; Africa, South America, maybe some countries in Asia, are going to fall behind in that regard. If the government has any role, it would be to make sure that China doesn’t swoop into those countries. We need to be doing our very best to assist those countries to maintain a level, so that they can also take advantage of the great successes that are going to come from AI.

Rep. Lieu:
Let me add that one of the things that Representative Buck and I and other lawmakers do when it comes to legislation is that we plagiarize. We actually will look at what happens in these other places that do go forward with AI regulations and laws. Europe is going to go forward faster than the U.S., and we want to see what happens. Maybe it turns out that it’s a total disaster, and maybe it’s something in between. We’ll have the benefit of at least seeing how it turned out in this jurisdiction. Then we can learn lessons from that as we go forward in the United States.

Mr. Jekielek:
Who really gets to control AI? Who gets to have their hands on the levers? Obviously, whoever that is will have a profound influence. We know that the different ideological perspectives of the programmers themselves get filtered into the system and influence how the AI is making decisions or what information is presented to the public, like we see with some of these ChatGPT type models.

Have you thought about how to deal with that? You mentioned there’s an industry-capture element that will inadvertently happen to some degree as well. How do we deal with this, because someone is going to have their hands on the levers?

Rep. Buck:
In the constitutional republic we live in we have a First Amendment. It really is fascinating right now how, not even just Republicans and Democrats, but really there are so many different views on what is misinformation and disinformation and what is protected speech. AI is just going to speed up that whole process of how we deal with that.

Then we have the whole other issue of trademark and patents, like when someone paints a painting and AI alters it slightly and puts it out on the marketplace. We have all sorts of interesting issues that are going to be associated with AI. Again, we go back to this commission to think through how that impacts us.

The beauty of this country is that we can agree to disagree. I don’t draw the line in the same place that all my colleagues draw the line in terms of what’s protected and how we need to protect Americans from the dangers of misinformation, so they don’t assume they don’t need to take a medicine, or they don’t need to act in a certain way. It’s a really interesting issue and one that will evolve as AI evolves.

Rep. Lieu:
Let me raise something for folks to think about based on your question about who controls the levers. There’s a very interesting debate happening between open source, large language models of artificial intelligence, and closed source. With closed source, the company has proprietary technology. They don’t share it, they put it out there, you use it, and they’re controlling everything. With open source, they put out their source code for people. Meta has in fact decided to do that. They put out Llama 2, a large language model AI that’s largely open source so people can play around with it.

I generally favor open source, because it is much better in terms of reducing who actually controls the levers. The people who want to play around with the code get to do that. But it does raise an issue that Representative Buck raised earlier, which is you can also remove all sorts of guardrails. You can ask questions like, “How do I build a lethal virus? How do I commit suicide?” These are questions that normally other closed source AI models won’t answer, so these are very tricky issues. Again, that’s why an AI commission of experts would be helpful.

Mr. Jekielek:
Congressman Buck, as we finish up, a final thought?

Rep. Buck:
My final thought is thank you. Thank you for helping us get the word out, and thank you for the opportunity to talk to Americans about how important AI is and to really embrace it. There are some fears out there, but if we allow the fears to control us we’re not going to take advantage of this great technology the way we should.

Rep. Lieu:
It’s an honor to work with Representative Ken Buck, and I want to thank all of you for doing this interview. I encourage the American public to read about artificial intelligence and learn about it, because it’s coming and you really can’t stop it.

Mr. Jekielek:
It’s such a pleasure to have both of you on the show.

Rep. Buck:
Thank you.

Rep. Lieu:
Thank you.

Mr. Lance:
Thank you both.

This interview has been edited for clarity and brevity.