Who Should Regulate AI? – with Kristian Stout and Bruce Schneier

Are you looking for another podcast?

Debating America's constitutional crisis? Then listen to the oath in the office. The politics law and democracy podcast hosted by constitutional scholar, co-redirection writer, and serious XM hosts John Pugelsign.

Each week, the oath in the office breaks down the biggest political stories through a constitutional lens in claiming English for broad audience. The show consistently ranks an Apple's top five in government and features guests

like Senator Sheldon White House and Supreme Court journalist, Dalia Lithwig. Smart, accessible, and focused on how power actually works. Listen to the oath in the office wherever you get your podcasts and on YouTube with full video episodes each week.

“There is a market opportunity there, and I think that if”

Chatchee PT and Claude and Gemini actually became

not useful overall, the market would respond with another alternative. I want to see states doing their part in regulating this technology. I think regulation is the way we are going to incent innovation here, regulation is how we steer companies

towards the sorts of outcomes that we like. Welcome to You Might Be Right, a place for civil conversations about tough topics, hosted by former Tennessee governors, Phil Bretison, and Bill Haslam. I'm Marian Wanamaker, dean of the Baker School

at the University of Tennessee, where members of our producer circle make this show possible. To learn more about our work, visit YouMightBeRight.org. Welcome back to another episode of You Might Be Right. I'm Phil Bretison, and it's good to be back with my co-host

Bill Haslam. I'm sorry, Governor Haslam. It's been so long ago, I think I can just be Bill now. This is a topic that affects all of us. By the way, thank you, Phil.

It's good to be back with you, but whether we're aware of it or not, the impact of artificial intelligence, and then how we're going to regulate that. Should Texas do it different in California? Should we have a unified federal approach?

You might think, well, that didn't really affect my life,

“but I think it might, and hopefully we'll”

be able to bring that up in this conversation. Yeah, well, you know, it affects everything. I mean, employers are using it for hiring. It's becoming widespread in healthcare. A lot of states are starting to write their own rules.

In, listen, it's former governors, we believe in this idea of states as the laboratories of democracy that let each state figure out how to at best. But with AI, the stakes are very high, and there's no natural borders when you're artificial intelligence doesn't recognize a wall

between Tennessee and Kentucky or Tennessee and Georgia. And so this is one of those things that I think we have to think about it differently. I think the real question is, does one size fit all, or do you need to capture unique needs?

I mean, what people even Silicon Valley might be different than in the industrial Midwest, for example. But at the same time, you don't want to have this patchwork that makes it difficult to create an innovative space. The other thing that we can't ignore in this is,

we're not the only US is not the only folks that are introducing artificial intelligence in the world. We have international competitors, and one of the main arguments with not having just letting every state make their own laws is we're competing with China and a lot of other folks.

If we have 50 different fragmented approaches to this, we probably won't be internationally competitive. And as people talk about this, the question is, how quick does this need to be? Does it need to play out a little more?

And let people understand a little more

“what some of the issues are, or is it time to really step in?”

I mean, how quick is too quick here? Our founders set up this system of federalism that there are certain things, there are states of responsibility, certain things that the federal government's. This might be one of our bigger tests yet

in this digital age of how is AI going to be regulated? We have a couple of guests. I think we'll help us answer that question. Let's dive right in.

Well, Phil, our first guest today is Christian Stout.

He's the Director of Innovation Policy at the International Center for Law and Economics. He's really focused on issues around law, technology, and the regulation of it, and why that matters. He has great experience, we're thrilled.

Christian, welcome to you. You might be right. Thank you for having me, Governor. Christian, let me start out. This whole subject of AI is a big broad area.

And the first question is, what pieces of that? Where do you think people should be most concerned at this point, or where should the discussion around regulations and so on, really center?

And it has a slightly complicated answer

in that the first thing to keep in mind is that there

is no such thing as AI, OK? And we talk about it a lot. But AI is, in fact, a category of technologies. And many of those technologies, we have had in our society since the 1970s in different forms.

People got used to things like predictive text inside of Google search. That's a form of what we would call AI. Different kinds of self-regulating systems and computers. You could call different kinds of AI.

We had the era of big data for 10 years before AI was considered a thing. That is technically a form of artificial intelligence. So we've had these technologies in a lot of places. That's the why we should care

is because they are so diffused throughout our society that if we do not understand what we're talking about, when we're asking for there to be something done, we have a severe risk of both overregulating and underregulating.

So you'll underregulate because there will be pieces of the new things, the new pieces of AI, that we don't really appreciate because our categories are too large. And at the same time, we'll overregulate

in that because we don't understand what it is we're actually putting into a legal context. We're not going to operationalize it correctly. And we will risk chilling the development of the technologies that we want and that make our life better.

- If I could just follow up quickly though, within the whole scope of that, I certainly agree with you about the breadth of it and that it's multiple technologies. One of the ones that, what are the areas

that concern you most that you think are most important

for us to figure out some way to deal with them? - That's another really great question

“and I think about it a lot because a lot of the discourse”

tends to be dominated, I don't want to say most of it, but a lot of it because it's popular and appealing from a new perspective. A lot of it is dominated, but what I call the existentialists of AI where there's fears that all the water's going to disappear

or the AI is going to become sentient and take over things. And those are not realistic from my view and when we give those kinds of concerns attention, we actually miss the things that we do need to care about. So there are some real potential threats

that we need to think about. One, the AI systems, the generative AI systems and some of the related technologies, they have the capacity to make criminals much more effective at doing their job.

So we need to think about how do different AI technologies

alter the capabilities of bad actors. At the same time, we also need to be concerned at things like government surveillance, right? So when there was the Snowden revelations from about 10 years ago and there was a lot of disquiet about the fact that our intelligence

agencies were able to spy on American citizens using intelligence powers, AI severely amplifies that capability. It's the same thing even on a more local scale. You can use facial recognition and related AI to help police departments potentially create

harassing approaches to their policing.

“So you need to think, and that's not to say”

that any of these things are definitely going to happen or that these are all bad guys. But these are the kinds of concerns that we should keep in mind as realistic. So you kind of said, we have to be careful

to want to have a Goldilocks approach. We don't want to make sure we don't overregulate or underregulate it, who figures out how to get it just right? - Okay, that's a difficult question. So there's trade-offs in every society.

So we frequently think about our competition with China. One of their, I don't want to call strength 'cause I don't like it. But we'll say one of their capabilities right now is because they're so centralized

in the way they approach things. They can have sort of a top-down command and say, this is exactly how we're going to develop AI. This is how we're going to regulate it. Everybody get in line.

We are not that kind of society. I think we are a better kind of society for it. But what that means is that we're messier and that we need to work through diffuse institutions from local all the way up to the federal level.

We need to then have court overlays on these things. So what that means is that we have an evolutionary approach. If we come from the common law style of creating law, which would be inherent from our colonial days. And what that means is that we have to be somewhat patient

while our policy makers and our courts see problems arise and they figure out what the actual harms are and what the actual regulatory approaches need to be. So we have to exercise both patients and prudence in that we need to actually understand the dynamics

of what we're dealing with and allow our institutions to allow those things to settle inside of our civilization. And a lot of the discussion around this also comes back to the issue to what extent does there need to be some overarching federal approach

to it is opposed to using states as laboratories of democracy and trying different kinds of things. Do you have a feeling about where this should be placed to this regulation?

“I think about it a lot and I have many thoughts”

that I will try to condense into a short answer here.

I really like distribution of powers across

the laboratories of democracy.

“That said, you need to make sure that you don't have”

lab leaks from these laboratories of democracy at the same time. And with something like technology category like artificial intelligence, unfortunately the threat that regulation in one state

that is ill-advised could spill over and have negative effects for both the national economy and our international competitiveness. Can you give our listeners give me an example of how that, how a relation from one state might spill over

and have negative effects? Can you think of a good example? Yeah. Well, so there is frequently counted California effect inside of regulation where

because California is one of the largest economies

in the nation, they will set some standard

for a particular industry say like emissions on vehicles. One of the ones that I work on is privacy policy. So they have different privacy laws in California that then because their economy is so important

companies will treat that as the standard for which they have to adhere. And sometimes that's okay, sometimes that's not.

“I think when a state can exercise its regulatory”

preferences across every other state, that's where we start to get that problem of an unjust unwarranted regulation. One of the things that I have not seen so much of is you know, is the use of the courts today

to begin to address these issues. You know, I mean, the internet grew in part because I guess it was Section 230 of the Deacency Act that basically absolved companies from responsibility for things that they published on behalf of others.

But what chat PPT or anthropic or someone does is very different from that. I mean, are people holding those commercial institutions legally responsible for what is happening with AI? Yeah.

So yes, again, a complicated answer because this is a very new problem. And sort of to your point on the Section 230 settlement, there had been cases that started finding either direction on how that question should come out,

which is why Section 230 came out the way that Congress looked at the question and said, all right, we want the internet to exist. We think this is good.

Here's what we're going to do with lead liability

on public reliability. And I think you're right. You're interested in this right. That that did help develop the commercial internet. On the chat PPT and generative AI question,

we haven't seen enough cases come to resolution yet in order to know exactly how court's going to look at that. We've started to see some resolution on the training side, the copyright side. And we are starting to see some cases come out

where there's been misuse of the services. We don't have resolution on those yet. So we don't have any answers unfortunately. But courts are looking at these cases right now, and they're definitely on legislators' minds.

I mean, one of the things that one of the things that's concerned me is watching is just you get people who are vulnerable, some way teenagers who get talking to these chat things. I mean, there's been suicides that have been

encouraged by the chats. And that seems to me something we're clearly there's some responsibility on the part of these large companies to put some car grails and to put that from happening.

Is that happening in the courts? So not yet, but that's not to say that they won't. So to go back to the 230 point you brought up, we've actually started seeing over the last few years that section 230, which for a long time

was almost an impenetrable shield for liability on a number of fronts. The courts have been accepting theories of products liability for poor design of algorithms and software systems

in order to find liability for some of these platforms. I have a feeling that in some of these cases you're going to see similar products liability theories applied to things like

commercial consumer deployment of AI systems, generative AI.

“I think with regard to that section 230 stuff”

that also courts have begun to make a distinction between things that you are directly making money out of commercially and the simple publishing of other views. But in the case of AI and these chat robots, for example,

I mean those are clearly, those are commercial products that are designed for the purpose of making money and not simply publishing other views. Yeah, no, I think that's right. And it's even more complicated in that

because there are a lot of these systems. So chatshevty and Claude are two of the more in Gemini are three of the more well known systems. But there are many of these systems that are being deployed. There are many open source systems.

And a lot of companies are starting to apply these technologies inside of products that are not otherwise thought of as chatbot. So the legal liability question is they'd be much more complicated as we get far

where your combining multiple systems to go out

“and just do stuff for you on the internet.”

We know our U might be right listeners care about solving public problems. At the Baker School, our online master of public administration is built for working professionals doing just that. Flexible and mostly asynchronous.

The program lets you advance your career while you lead. You'll learn from faculty and policy analytics, emergency management, nonprofit leadership, and economic and community development. And while you may not be on rocky top every day,

you're very much a part of our community. With regular touchpoints, the Baker faculty and staff, we're enrolling now for Fall 226. Come be part of the solution.

Learn more at Baker.utk.edu. We all remember the song. It made it all seem so simple and turns out it's not who writes influences and kills bills.

It gets messy. I'm Scott Greenstone. And I'm Libby Denkman. On sound politics, we tell that story. The inside track on how policy gets made

in this Washington and the other one. And how it impacts you. With now on the KUW app or wherever you get your podcasts. So one of the issues we're in, we talked a little bit before we're trying to dig into

is so who should regulate this? And I could say, well, there are certain things that Tennessee might want to have different standards on than California does.

There's other things that when it comes to, you know, how we're going to use this information. Everybody thinks, oh, I love it. Being, or most people would say,

using it to help us identify bad guys. Other folks would go, yeah, but let's not use it just to invade the personal information of Americans.

So one of those feels like, yeah, states by states make sense. The other is, no, that this is a federal issue. We ought to have a one blanket policy for all. How do we figure out what it's okay

for states to regulate? Because I think almost every state has some, there's past some legislation or has some before it.

Yeah, that's a billion dollar question.

My approach, the way I frame this problem, is to say, you need to think about these AI systems in two buckets when you're thinking about regulation. The first bucket is the creation of them in the first place.

The second bucket is the use of them in society. In that broad framing,

“I think states are well within their rights”

to apply their traditional police powers on the use of AI in society. How it's deployed, how kids interact with it, how, whether it's using housing decisions or employment decisions, things like that.

States have a definite interest in it. And it's okay for states to have different policies on those issues. I think that works out okay, because if you think about it broadly,

so obviously there's going to be cases where maybe it's not. But as a general approach, I think that is better because the use of the system is within the state. So how you deploy it within the state,

how you use it in particular products that affect the citizens. You're getting a lot closer to the traditional police powers. When you're talking about how do you train these systems? How do you teach them to recognize text?

How do you think about bias or something in the core technology itself? That's where you're starting to get outside of what is the competency of the states in my opinion. There's so much to talk about on this topic.

And every time you bring up something, it raises another question a little while earlier you talked about having bias built into a system. You know, I'm the Republican on this duo.

And our side is always felt like,

"Oh, the media is biased against us." Well, this feels like you could, you could see the water so to speak to benefit your side.

“Is that something we just let the market take care of it?”

And people will figure out where there's bias and go past that? Or should somebody, is bias something that should be regulated in any way by state or federal government? I mean, my, as sort of rate-leaning libertarian, I'm sympathetic to your concerns there.

And I will deliver the news that I think the market is probably capable of handling this problem. So the first thing that I was trying to keep in mind when I look at this problem is that the internet is the basic training data for a lot of these systems.

And the internet itself is full of people who have perhaps more of a left-leaning, left-leaning bias. So it's not because of evil intentions, it's just that young people tend to be more progressive when they're younger and they also tend to be more active online

and in discussion forums and things like that. So yes, there is that in there. However, the systems when they're developed, they do, from the research I've looked at, they do try to debias them and make them as useful

as they can be to ever have us to be using them. So it can come down to how the users using them. Now that said, there are other startups

the big companies that are relying on the existing training data. Like I'm personally aware of one that tries to be as neutral as possible. And there's techniques for even taking open source models that are trained on internet data that maybe we think is biased and you can distill it into different forms

that have different kinds of biases or less of biases depending on your perspective. So there is a market opportunity there.

“And I think that if chatGPT and Claude and Gemini”

actually became not useful overall, the market would respond with another alternative. Let me ask you this, maybe final question. And that is, if I were the governor today, we're both ex and so on.

And I wanted to just begin to do something to show that we were aware of the problem and maybe even make some models for other states to use or the federal government to look at,

what would you advise me to tackle first?

That's a fantastic question. Yeah, so if I was advising your governor, what I would say is we have a couple of priorities that we want to put into maybe some kind of executive statement. One, we want our existing agencies to identify

where their remit is hampered by the presence of different kinds of AI and where they might need more resources. So you're fraud and criminal enforcement wins. They might genuinely need more resources

to help combat this stuff. I would want to identify those things. I would also test the agencies with discovering where just generally speaking, it's harder to do their job. Maybe it's not a resources issue,

and it's a new problem. And then from there, I would then take a look

at how to take those first and second prongs

and one recommend legislation to the legislature based on what I discovered in our executive branch. And then two, start working with the federal government, NTIA and IST. There's congressional offices that are very active on this

and say, hey look, we don't want to be that state that is trying to regulate the idea of existence, but we want to be a partner with the federal government. What we're seeing, and here's how we want to work with you,

“to make sure that you're regulating what you need to regulate,”

and you know the areas we need to regulate. And within, say, like, a year to have a good plan in place that you're going to start executing legislation that likely survives on courts and won't be preempted by the federal governor.

Great question. This is so helpful. Phil said, maybe next last question. I'm going to ask one more than a lot of him as the final one. He talked to us about the political reality of actually having regulation happen.

This is one of those, there's string bedfills, right? I mean, all the rights, not on the same page on this and the left. I mean, you have President Trump with an executive order about not letting states in a theory of Ted Cruz, you know, conservative Texas Senator saying don't let states in effect.

And then you have some other, you know, conservative senators and other folks say, and absolutely states need to have their state. Tennessee does not want to be like California. What's the political reality of having regulation actually get passed and put into effect? So the political reality is very strange right now as you noted,

where as I see it, there's not an easy left-right divide on this question, which is frequently a common thing in technology, honestly. So maybe the general society is just getting used to this idea of what, what tech focused policy people have known for a while. So what that means is that there's opportunities and there's challenges.

I think the opportunities are because there is not an easy left-right divide on this question. There's opportunity to work across the aisle and create a moderate middle position that creates a good baseline for the American leadership in AI that this administration has signals important, which I agree with. But also it takes seriously the fact that the states are going to have

their own interests that they want to make sure they protect their citizens. There is that opportunity and I think the people who are more moderately inclined and are not thinking about what their presidential chances are in the next cycle

“can work together to say, "Look, you guys, I know you have to do politics.”

You're going to message on politics. Here's the real core serious things that we serious people want to do from AI and also are protected by our existing laws." Let me ask a final question here. The name of this podcast is "You Might Be Right."

And its genesis is with Howard Baker, a very effective senator and chief of staff

who always reminded people that while he had certainly strong views about things

to always keep an open mind that the other fellow might be right. Can you think of an example maybe in your own history where you had a view of something and by opening your mind up a bit and listening to some other people or someone else and so on? You said, "I think I was wrong about that.

I've changed my mind."

Yeah, I think give you two quick ones.

One on a more abstract side and one on what we're talking about here. On the more abstract side, I started out my early life as a very dedicated libertarian and I thought this is the coolest way to regulate society. Everybody's going to be on board with this.

“And the more actual humans I met and realized how messy life is”

and listen to people from all different political camps. I realized maybe that works really well for me and the kind of person I am but I actually have to be sensitive to the fact that there are many different people with many different values and that if I want the world to work well, I need to make sure there's room for a lot of these different perspectives.

What we're talking about initially I was open to the idea that these newer AI technologies maybe there's something here which is so novel that is going to require a completely different regulatory approach than what I typically think about. And where I've gotten with it is sort of in the middle.

Rather than a don't regulate or a, you definitely need to regulate.

I think what we need is tailored regulation, but we need to have humility. We need regulation, we have to be humble in how we approach it. I have to come and I love your answer about letting the messiness of life in our own humility drive our answers to a better place. That's a great response.

Thank you. You've been a terrific guest. We really appreciate it. Thanks for all the insight. Absolutely.

It was a pleasure. How did so much power end up concentrated in the presidency with so few limits?

“If you want to learn more about this constitutional topic,”

then listen to the National Press Club award-winning podcast Master Plan, hosted by a claimed journalist David Sarota. Back for a new season, Master Plan explores the original intention of the constitution and how the documents interpretation has evolved and now expanded the powers of the president. Master Plan is smart, deeply reported, and surprisingly gripping.

If you want to understand the world we're living in, don't miss out. Follow Master Plan on Apple Podcasts, Spotify, or wherever you're listening now. Well, Bill, our next guest, I think, is very interesting. He's a returning guest. He was part of our AI discussion for two, three years ago here now.

Bruce Schneier. He is a security technologist and an author. He's an lecturer at the Kennedy School at Harvard. Bruce is known for his work on cybersecurity and privacy. And especially the intersection of technology and public policy.

I think it's very important to subject these days. He publishes a newsletter called Cryptogram and a blog called Schneier on Security. He's the author of the least 14 books and counting. And we're honored to have you here with us again, Bruce. Thank you.

I shouldn't mention I including a book on AI and democracy. Oh wow. Okay, well perfect. Perfect for our conversation again. Welcome back.

Let's start with the big overarching one. If you're just a normal American outlisten of this, while you're walking your dog in your neighborhood, why should you care about AI regulation? Why does it matter to you?

AI is going to affect every aspect of our life.

It is the first technology that replaces human cognition,

sort of in the same way that electricity and steam power replace human effort to empower. The effects are going to be enormous in sort of every aspect of society. And I don't know if we don't regulate it, then it'll be done for the benefit of a bunch of white male tech billionaires in California,

which seems like not a good way to run society. So it's going to matter. Regulations cause it's going to matter. So who should regulate it then? I mean, it's going to be everybody.

I mean, who should regulate food in this country? Well, there are federal regulations. There are state regulations. Cities are the ones for regulate restaurants. That is a thing that is so important and so embedded

that regulation happens everywhere. AI is going to be like that. It is, it is, you know, the states are closest to the people. So state regulation is really important here. Just like we have state privacy regulations and state regulations about data.

“Federal, I think is an international component.”

You know, we're looking at AI being using warfare right now. So that's going to require treaties and international organizations. So at every level is going to be the right place. Let me, you know, AI is a, it's not one thing. It's a collection of technologies as we've talked about on here and so on.

What in what do you consider AI? I mean, what are the most high priority things that you think should regulation

Yeah. In general, we put regulation in place when the thing can kill you if it goes wrong. My airplanes, cars, medical devices, pharmaceuticals. We don't trust the market because the market is not going to take human life and property into account. Well, it doesn't, it's not the market's job.

So where AI touches those things, we're going to see the first regulation.

I mean, necessarily. But, you know, do we want robot policemen? Do we want AI is with weaponry? We start thinking about those things. AI is driving cars.

It is a basket of technologies, which makes this kind of hard to get our hands around. I like to think of it as technologies that loosely mirror human thought. Chatbot are the ones that are getting all the news. But a lot of AI is predictive. At the AI that drives your car is the doesn't chat at all.

The AI is doing drug discovery or proving new math theorems is not chatting. By the AI that wins a chess or go doesn't chat. Those applications, some of them are more benign than more toys.

“But I think it's going to be everywhere starting with the ones that affect life and property.”

So I like to get the distinction. Okay, let me say I get the distinction. But sometimes it's not that easy. So we all would say we want it to be, most of us would say we want it to be used for things that identify bad gas. But others would say, right.

But when are you starting to take my personal information? All that's not a new argument. We had it with George W. Bush and the World War on Terror. And how we're going to use information and government's access to that. How do we start drawing the line?

Like we see it happening right now with the Pentagon, right? With the argument, it's around that and dropping.

“How do we decide like we want it to be used for that?”

Like this involves life or property versus things that are just encroaching on it?

You know, I think the whole data privacy is a giant red herring by the tech monopolies that basically want all of our data.

So they can make a lot of money off it. So it's really important to separate the tech from the economic system the tech is operating it. And a lot of the problems of AI are really problems of the market and lack of regulation. So I mean, yes. So it's an old argument, security versus security, right?

The police can be had guys, this security in our information being saved so that nobody can get at it. And we balance those two. For the police, we often use a warrant process as a balancing, right? We give the police the ability to peer into our lives. But we make them go to a court and justify their actions before they do it.

So that's one way to balance. There are lots of ways to balance.

“One of the things I think the struck us is this is one of those arguments that unlike most things in our political world.”

It's not dividing up in an easy red versus blue. You have the, you know, you have President Trump issuing executive orders to, you know, put a moratorium on state regulation. You have a Ted Cruz who's, you know, on the right side of the, of the, of the Al, saying, yes, that's the right approach. And you have a lot of other Republican senators and governors saying, absolutely not, we need states to play their role. Talk a little bit to us about where states should have the ability to put their own thumbprint on regulation.

I think this is a time when we need experimentation. That's where you get from the states. I think states are closer to citizens. The same way states regulate, you know, so many aspects of people's lives. AI is touching those and we regulation there.

And the argument that, you know, too many regulations are hard to follow.

I think is just whining on the part of these multi-hundred billion dollar tech monopolies.

God, if you can't figure out 50 regulations, what are you doing in business? I have no sympathy for you. I mean, just figure it out too bad. So I want to see states doing their part in regulating this technology. I think regulation is, is the way we're going to, incent innovation here. I mean, regulation is how we steer companies towards the sorts of outcomes that we like.

I mean, I think a monopolization is an issue here.

And in what's interesting, some of the places states are really stepping up is in data centers. And this also is not Republican Democrat. There seems to be grassroots everybody saying, you know, we don't want these large, loud, power-hungry, multi-football field size data centers in our community.

“That's what I'm interested in to watch, because that hasn't split along party lines.”

And I don't think it will.

Do you think that, I mean, one question that I have, I certainly believe in the notion that states have a role to play here in our federal system. But this seems to be a particularly difficult technology to effectively regulate at the state level. I mean, we're in Tennessee here, and you can pass some laws about AI. But, you know, I can get a VPN and look like I'm talking from New York to a company in Maryland or something like that. You know, and they seem so easy to evade state regulations.

A much easier to evade state regulations than something like, I don't know, the sale of automobiles where you have.

The presence of dealerships in the states and licensing in the state and so on.

“Is that a problem with the states involvement in regulation?”

Right. It's a problem. It's not unsolvable. It's been a true for states regulating internet anything since the internet. And we still have lots of state regulations on different things. And national regulations, you know, in France and Germany, it's illegal to, you know, have various Nazi promoting websites. Those citizens are going to get a VPN also and pretend to come from the United States, whether it's perfectly legal.

It doesn't mean you don't do it. I mean, right, the the inability to perfectly enforce the law is not or justification not having the law at all, otherwise they'd be no laws against murder, and it's not perfect, but we still have those laws. So I'm, that's an issue, but I don't think that's a stopper.

“Are there, as you look around the country, are there state laws that you've seen that you think are models of the way that that regulation should proceed?”

Now, right now, the best state regulating internet is California. They have really good privacy laws and there are some laws regulating AI. There's so a lot of tussle between the parties. And, you know, because of the tech companies are in California, there's a lot of lobbying dollars to go into shaping some of those regulations. That's probably the best. I'd also look at New York and Massachusetts tend to be states that are more likely to regulate here.

But in terms of digital privacy, we're seeing regulations across the country, in states that are traditionally Republican and hugely Democrat, which I think is really interesting to see that the failure in the federal level to regulate is really cause the states to step up nationwide. Roots, earlier you talked about, listen, like I said, regulations, so we have additional 50 regulations figured out you're in business. I think you said something to that effect. And, you know, the, the face books of the world, et cetera, should be able to figure that out.

My experience in business is the heavier regulated things are, it actually played to the advantage of the bigger, bigger players, because they have the resources to get past that. And, in heavily regulated areas, your margins, once you're in business, your margins actually increased. And it actually ended up helping the big guys. Do you have an answer for that? It's definitely true. And then we saw that in Mark Zuckerberg a few years ago, right, so they become in favor of regulations. And I agree with you. He sees them as protectionist as keeping his monopoly. I mean, the way we deal with that banking is a good place to look, banking regulations are tiered to the size of your back.

And as you become larger, you're regulatory burden increases. And that they do that not to unduly penalize the smaller guys. Well, still having the regulations needed to make the banking system. We can imagine something like that in tech, that regulations that kick in only when we become so large. So it doesn't unduly burden the small players and doesn't unduly benefit the monopolistic players. You mentioned California, New York, and I think Massachusetts states that were actually regulating, just get a little more specific.

What are some of the things that they're doing that you think are particularly admirable or effective? I don't have that in mind. So I'm not really cans that question well. What I want to say generally is that they're trying. They're looking at it and trying to figure out how these technologies could benefit our harm citizens.

There's a lot of back and forth. I teach that act in my class. There's a lot of good. There's a lot of not good. There's room for improvement. I like that they're trying. And I feel like then this space that is moving so quickly, we need to figure out how to actual regulate an agile matter. So let me take up on that. So use the EU as an example. If we're betting dinner five years from now, I'm wouldn't bet on Europe to lead an innovation on AI. I'm going to bet on us or China and a lot of folks would say,

one of the issues around regulation is this isn't we're in an international competition here for a critical resource.

“The more state level regulations you put on this, the more handicap we're going to be in that international competition. What's your response to that?”

Yeah, I don't think so. I mean, the competition, I think competition also is something that is sold by the big tech companies in the US to justify their actions. This is not the 1960s. This is not the space race. Technology is not done that way. It's done in public. It's done open.

If you think about the transformer, which is the core data structure technology, powering all these AIs, developed by Google that I think 2018, it's public. It was a research paper.

China, the deep-seek model from last year the year before, because they would deny access to the best US chips and to innovate in really interesting ways to create their model much cheaper using less powerful chips. All of that is public, right? That is published. This technology developed internationally in the open. It is not an arms race.

“It is not a 1960s space race. I think that rhetoric is being sold by the tech companies, but it actually is just not true.”

So, I mean, to me, regulation is how we steer technology. Good regulation enables the technology that we want.

I'm right here this week in California at a big cybersecurity conference. I'm going to go to the show floor later today. This is going to be filled with companies with great products that are not going to be used because the demand isn't there, because the regulatory requirements aren't there. I want for better security are good, strong, regulated, high requirements that force companies to spend more on cybersecurity, which will make us all safer. We do regulation right, and it spurs innovation. Doesn't hamper innovation.

Both of us have previously been governors. If I were a governor today and called you up and said, look, I want to begin the state of Tennessee going down this road.

“Maybe some baby steps to start with, but try to do things right in this and stay ahead of the curve. Where would you point me to be looking at?”

I think I'm going to look at where these technologies are affecting people right now. I worry a lot about the harmful addictions to these artificial personalities that we're seeing. I'm working at worried about harmful effects to children. They need to think about how this affects your education system. If you have taught a class, you realize that the take home essay is now dead as a way to measure students performance because they all use AI. So looking there, we're going to have to start thinking about vehicles. Do we want driverless cars on our streets?

And you know, you and Tennessee are going to have a different answer than me in Massachusetts because it's snow's more where I am at the roads are different. I want to look at localization and how it affects any of the locally regulated professions. States have a lot of licensing rules for attorneys, for doctors, for accounts. These are state licensed. And it's a how these professions interacting with the technology and how does that affect consumers and citizens. In any place where the state is regulating people and state is to think about what AI enhanced people are going to be like.

I'm just to follow that up a little bit. I mean, there are things that are more subtle universities are using AI to read student essays on their application form and making application making admission decisions based on.

All right, we need to we've unfortunately reached the end of our time. Let me ask the question we ask all of our guests.

So the podcast name comes from can you think of a time and hopefully maybe even on on this topic where you realize I didn't have it quite right the other side had a really good argument and and I didn't I realize the other the other side might be right. Can you think of a good example like that. You know, tech is fill of full of that. It's not really sides, but it's the fact that it's past and future. I am finding that things I'm writing about this technology year ago are no longer true. And we're all guessing about where it's going. Can you give us one I believe that can you give us one good example of that that that that's actually a great. So in October I wrote an essay about how AI is going to affect cyber security like what will the attack defense balance look like in a world of AI enhanced attackers and AI enhanced defenders and it was a fine essay.

What happened between October and today is that AI has gotten crazy better at writing software.

And that we can suddenly envision a world of a femoral software where software is written like constantly by individuals by companies by networks. And now cyber security in that world looks way different than I postulated in October. And I wrote a new essay last week that lays out all these new trade offs and and speculates on what is going to look like is that going to be accurate a year for now. I'm not going to hold my breath and this is important anybody who who comments on the tech do not be someone who used to two years ago realize something and had done this is changing so fast.

“You mean you need to use the tech to understand the tech.”

Great great insight Bruce that that last point is incredibly helpful for me and I'm betting most of our listeners as well.

Thank you very much. This is a great great session. Thanks for your insight. Thank you. Thanks for rejoining us. We really appreciate it. Those were interesting and and I guess underlying for me just what a broad complex subject to how many wheels of the wheels there are in this field. One of our one of our purposes of this podcast is to help people understand like this this is an easy right and that we had two pretty distinctive perspectives. One person who more or less said let's let's trust the market the market's going to figure it out.

No, no, we need to regulate this in a strong way and used examples of you know blue states and then Europe which is a you know kind of left of center model of let's regulate this and we'll get to a better result.

“I think I guess they were both very good guests.”

I thought it was a little hard to get down to specifics that I could understand some of these. There's a lot of abstractions about about the technology and I think maybe we should do one in the future that really just gets down to you know very sort of kinds of questions. I asked the question at the end about you know colleges reading reading essays employers read applications and so on and I mean just I mean it's a human being I mean how much do you want to permit that. You know things that are really important to my life from my children's life or so on to be handled by algorithms.

So let me ask you this is if you were you asked the question of both of them like if I was governor what advice would you give me if you got to be. What kind of recommendation of what that would look like and what would be federal and what would be state do you have us. I think you know examples when we talked about earlier in the podcast of you know teenagers are committing suicide because of interactions with you know a chat that encourage them to do so.

“I think that making sure that you hold these large companies accountable for the stuff they're putting into the you know into the public sector in this way is is really important to help present some of those.”

I think the issue of just using you know generative generative AI to create of you know pictures and videos which are. You know me I've always had Photoshop but the stuff today is very right for it and so and you know really making that really make that illegal in some way that that you know forced people to least identify it are the things that I would focus on I think there's a lot of potential for harm and very.

I grew in examples you gave and like said we we talked about in both podcasts examples of you know we've seen cases of suicide resulting from conversations with chat bots and.

“See the danger saw this is like said I'm I'm more of a market oriented god definitely than then Bruce is but I think you have to recognize some of the danger and have a way to respond to that again I'm not going to be.”

Is the question I asked Bruce like said there is a cost of regulation and and I do think it actually is in the end of the day helps the big guys so you have to recognize that. Somewhere in there again if I got to be the czar I'd say there's certain things that we definitely want to make certain there's some responsibility that can be applied. No enough now to tell you where I would draw the line. Yeah I think I mean I mean one of the dynamics of this has been you know we talked a little bit earlier about the section 230 of the communications decency act which basically insulated these big tech companies from.

I have a lot of ability from stuff that they were simply publishing on the behalf of others and that's started to be tightened up a little bit but it's still very much is there and and you know I think that I think that produced a kind of an arrogance and.

“I think I did with this I love that while the our Christian and Bruce have very different approaches to this one is like said more more.”

I'm a little bit more of a marker orientation.

Bruce is obviously emphasizing the importance of regulation. They both ended with in an inch in place of this requires some humility from from both of us and I like Bruce point whatever we talk about a year or two years from now we'll look back on this conversation say we. We were maybe in the neighborhood but we didn't exactly nail it and I like the humility that comes out of this with the humility combined with responsibility.

He is to me a great goal.

Yeah the only discouraging thing I heard is I think I think Bruce talked about the need to be agile and. Just anything the US Congress is not. It's agile and you know the need to sort of you know probably just constantly re-look at this stuff every in every year to.

“As the as it develops I think it's an important part of any regulatory scheme.”

Great thanks. Thanks for listening to you might be right be sure to follow on Apple podcasts Spotify YouTube or wherever you listen to your favorite shows. And please help spread the word by sharing, rating and reviewing the show. Thank you Governor's Medicine and Haslam for hosting these conversations. You might be right is brought to you by the Baker School at the University of Tennessee with support from the Boyd Fund for leadership and civil discourse.

To learn more about the show in our work go to you might be right dot org and follow the show on social media at yambrpodcast. This episode was produced in partnership with relationary marketing and stones river group.