The Anti-Tech Era? AI Regulation, Data Centers, and America's Growing Governance Clash
- 21 hours ago
- 52 min read
Colorado has become ground zero in one of the biggest debates shaping the future of technology: how do we regulate AI without slowing innovation?
In this episode of Signal & Noise, Brett House and Rio Longacre sit down with Srinivas "Chinnu" Parinandi, Associate Professor of Political Science at the University of Colorado Boulder, to explore the growing tension between government oversight, technological progress, and economic competitiveness.
The conversation examines Colorado's AI legislation, the challenges of regulating rapidly evolving technologies, and whether America needs a national framework for AI governance. Chinnu also explains how privacy laws, political incentives, and fragmented regulation are reshaping innovation, entrepreneurship, and the broader technology landscape.
In the second half, the discussion turns to the AI infrastructure boom, exploring why data centers have become a growing political and economic battleground as demand for computing power accelerates.
The conversation covers:
Colorado's AI legislation and the future of AI regulation
Privacy laws, federal vs. state governance, and technology policy
How regulation impacts startups, innovation, and competition
Data centers, energy demand, and local opposition
America's AI competitiveness and the future of technology governance
As AI continues to reshape the global economy, this episode offers a timely look at the policy decisions that could determine whether the United States remains a leader in technological innovation—or falls behind.
Watch the full episode and join the conversation.
🔑 What We Cover💡 Key Takeaways🎯 Why This Episode Matters
Read the full transcript below.
Brett House (00:01)
Hey everybody, welcome back to Signal and Noise. This is Brett House joined by my co-host Rio Longacre. And today we're diving into one of the most important increasingly or maybe just standardly controversial topics and issues facing the future of technology, AI in business in America, and it's AI regulation. ⁓ And for those that have been sort of paying attention to the law over the last few years, states like California and Colorado, along with regulators in Europe have introduced
Rio (00:07)
Hey there.
Brett House (00:31)
sweeping laws around privacy, AI governance, and digital infrastructure. Supporters argue that laws protect consumers and create accountabilities. Critics argue they risk slowing innovation and creating impossible compliance burdens for businesses, especially small businesses, trying to operate across state lines to help us impact this ⁓ very interesting topic and one that Rio and I have certainly done some homework around.
to make sure that we were prepared to talk to Professor Chinu Paranandi from the University of Colorado Boulder. Chinu, welcome to the show. Thrilled to have you. ⁓ And Chinu's calling in from a car, but he claims that it's going to be fully stable and he's not going to lose connection. But if he does, so far so good. He's an assistant professor of political science at the University of Colorado Boulder.
Rio (01:08)
Rock them.
Chinnu Parinandi (01:08)
Thank you for having me.
Rio (01:18)
So far, so good.
Brett House (01:24)
And you really focus on American institutions, state politics, public policy, and the interaction between government and business regulation. And you look a lot at these structures that shape policymaking in both the state and federal levels, right? So you're particularly well positioned to discuss Colorado's evolving regulatory environment around AI, privacy technology, which is considered by some to be very aggressive, and how those trends connect to the broader national debate.
around innovation, interstate commerce, AI specifically. you know, and just for those that don't know, Colorado became one of the first states in the country to pass a broad AI governance framework aimed at high-risk AI systems, putting it at the center of the national debate. Yeah, what they deemed high risk. Putting it there, but they were sort of in the center, and Rio lives in Denver, so you're a native to the area. And what I actually found an interesting fact is that
Rio (02:08)
what they deemed high risk, right? But, you
Brett House (02:22)
the law that Colorado or set of laws that Colorado has passed is one of 1,561 bills introduced across 45 states in 2026 alone around AI regulation, which is hard to wrap your head around. so, Chyna, did I do you service in terms of your background? Would love you to introduce yourself to the audience and let them know who you are.
Chinnu Parinandi (02:48)
No problem. Thank you, Brett. And thank you, Rio, for having me. ⁓ I am actually an associate professor, which still is not full, but I have the benefit of tenure, which is nice. So I can be more open in my answers to these questions. I study political economy as an old phrase that goes all the way back to Adam Smith. It essentially deals with this idea that political systems and economic systems and institutional systems in terms of governmental structures
Brett House (03:09)
Yep.
Chinnu Parinandi (03:18)
and the incentives that they create influence markets, influence human behavior in good and bad ways. And so I studied essentially the political economy of technology. Most of my work historically has been in energy and I'm still really passionate about energy, but technology, obviously you can't say the word technology without thinking of AI. And so that allowed me to sort of have a yes. And so that,
Brett House (03:43)
And without thinking about energy.
Chinnu Parinandi (03:47)
That allowed me
to have a natural like segue into AI and I've been studying the AI space for close to three years now. And so I'm learning a lot just like you both are and it's still kind of like the wild west. And some of the regulation is kind of an attempt to tame this area from the vantage point of, of governments and regulators.
Rio (04:10)
Well, we're really excited to have this conversation today. It's very timely. This is a topic area that I've written about fairly frequently over the last few years. When CCPA came out and GDPR, we had a lot of clients asking questions. So as a consultant, we were brought in very frequently to help them navigate those, whether it's from a compliance or just technology perspective. And especially in ad tech, there's been a lot of implications of it. Okay. Is that if that is hash is protected, what's covered, what's not. So.
Lots of interesting stuff there, but the timing is really good here because of AI. But the two of you mentioned is how it's blown up over the past few years. Most US economic growth is now actually tied to AI, whether it's companies adopting AI, building data centers. You mentioned the tie in between power, electricity and technology. mean, AI makes that extremely clear that we cannot be a world leader in AI for very long if we can't generate a lot more power. I think that's become very clear.
And Brett, as you mentioned, there's all of these laws that are either being passed. had the Colorado, the very controversial 2024 AI ⁓ law, SB 24205, I was at a, it's funny. I remember when the mayor here in Denver won his election, Mike Johnston, I was at one of the, I was at a party to celebrate it. And then I, this is around the same time, one, a big tech ⁓ CEO, we're talking and he's, cannot believe we passed this law. This is, this is insane. This is making, this is going to open up liability for.
not only the companies using models, but the developers creating them. This is crazy. It was amended and we have a new version now, SB 26189. It does rewrite it, but I think there are a lot of issues we need to dig into that as well. But the broader implication here is not just the law in Colorado, the way I want us to talk about, like all of the other states, dozens of states have passed these, but as you mentioned, there's been this kind of rush to go in and regulate it or prevent it.
Brett House (06:02)
Yep, California with SB53 and Texas with
HB149, yep.
Rio (06:05)
slow it down. And
as we've seen with GDPR and CCP, a lot of these regulations, have downstream consequences that maybe should be foreseen, but very much very often are not. So we're seeing this explosion of state-of-the-art regulations around privacy, around AI systems, around data governance, and even physical infrastructures. 40 % of new data center projects are being beaten down by these quote unquote grassroots efforts. There's been accusations of foreign influence. I don't know if that's true or not, but 40 % of them are now being defeated.
This has become a very big deal. So I think Colorado has kind of become this interesting case study because it passed us all. But generally speaking, I think this will be a fun discussion of what do these laws do? What are the forces behind them? Political, cultural, economic. And what does this really mean for innovation and entrepreneurship here in Colorado and the US writ large? So let's dig in. This is going to be fun.
Brett House (06:55)
Absolutely. when you look at, Jenny, when you look at today's political climate, you know, not to be partisan around this, but around technology and AI, what stands out to you just to sort of lay the groundwork is different from from previous regulatory areas and what's really jumping out your research.
Chinnu Parinandi (07:12)
big distinction, staring at this from 2026 compared to say like 1996, which is an era of deregulation is the role politically at least of populism. Now populism essentially like means politics that pander to like the everyday person. And even if you don't think that there's like, even if that's just a phrase and there isn't actually a real kind of like everyday person, it's this important conceptual tool.
that political movements can use to kind of root support for the policies that they pitch. 1996, right, like 30 years ago, we'd be talking about sort of a much more coalitional based policymaking where business and government would work together because there was less populism. There was less of a view that sort of elite interests, be they government, be they business,
be they sort of like whatever other kinds of elite interests you can define, we're kind of rigging the game against folks. Today, we have an era of populism. so business is viewed suspiciously by a lot of voters and importantly by voters in both parties, right? So populism is not something that, you 30 years ago you could say that the Democrats were more populist in nature.
Brett House (08:29)
Yeah, that's an important point.
Chinnu Parinandi (08:37)
than the Republicans. But I think what's happened over the last 10 years is that the Republican Party is much more divided within the party between traditional, like pro-business interests and populists. And the Democrats are also facing a similar divide between sort of centrist Democrats who may have had more pro- Yeah, between centrist Democrats who may have had more pro-business interests and kind of progressive Democrats.
Rio (08:55)
You're the legacy centrist Democrats,
Chinnu Parinandi (09:03)
who have a lot in common with the populace in the Republican Party, even if the object, even if the views aren't similar, there is a similar philosophy of the game being rigged. And so this is gonna lead to more kind of ⁓ intrusive regulation toward business because the assumption is that business is trying to structure the rules to its benefit rather than
trying to create some type of marketplace of ideas or marketplace of incentives that fuel innovation.
Rio (09:37)
That's a good point that it is driven by both the far left and the far right. Like in Colorado, for example, we've got, you know, I think it's a very blue state left-wing populism. You could argue is probably behind some of these regulations we're going to talk about. So if you look at Florida, Florida, I think was proposing from the right-wing government or certainly right-of-center government was proposing an AI law as well that did some of the same things. And to my point earlier, there's dozens of states that passed these, both red and blue. So this is a...
a rare bipartisan issue in this country, right? Where both parties agree that there is a bad guy, which is in this case, it'd be big tech, right?
Brett House (10:12)
Yeah, yeah, well, I, ⁓ sorry. Good.
Chinnu Parinandi (10:14)
I
would say I agree I agree that sort of like this is one of the few areas in our polarized climate where the parties do agree that something has to be done.
Brett House (10:24)
Yeah, well, let's look at the like with Biden at the national level, at least before we drive, we dive into the Colorado law. I think it was EO 14-110. There's a lot of these funny names for these laws or bills that they want to pass. Yeah, in executive order, they treated AI as promise and peril. It's centered on safety, privacy, civil rights, competition and national security.
Rio (10:37)
executive order, yeah.
Brett House (10:48)
But that, so those all sound good, but then the tech industry, the Mark Andreessen Horowitz, the Teals, the Elon Musk's, the tech bros all came out with a real strong perspective that ⁓ this was too important to smother in bureaucracy. And they called for repealing federal rules that hinder AI, especially as it prevented our ability to compete effectively with China.
How much credence do you get? Like, how would you balance those two perspectives? One seems to be saying, hey, we need AI to protect for, we need to be the global leader in some sort of ⁓ controlled environment as we release what potentially could be dangerous technology versus, hey, you're smothering, especially small businesses that can't afford to pay the tax to deal with all of the regulatory hurdles. ⁓ What do you think about that, Chino? What's your perspective? What do you see?
Chinnu Parinandi (11:43)
So
I like this example. I think about the sort of need for a response to both AI and China as a two-fold answer. One answer is as a political scientist, and I study American politics and American political institutions, and specifically the interplay of legislative and executive institutions. And so I think I have an opinion that in the American political system,
The way that policy should be enacted is through the legislative process. And there are lots of reasons why policy should be enacted through the legislative process. One is that sort of there's more electoral accountability and sort of democratic small D democratic accountability when you can say that the people's elected representatives have adopted a law. The other reason is that the U.S. political system has a lot of veto points within it to give different
competing, yes, to give different competing interests a say in a particular policy. And what that does is that slows down the policy process, which a lot of people don't like. But the good side of it is that when the policy does get adopted, it has broad buy-in. And it may be watered down from the vantage point of purists, but that buy-in and that watering down allows for a policy that
Rio (12:39)
The checks and balances, right?
Brett House (12:41)
Yep.
Chinnu Parinandi (13:07)
is kind of well thought out where you've over engineered the policy to kind of take out a lot of potential defects. ⁓ And so my vantage point is that I would love a world where there are no executive orders because executive orders happen because Congress is unable to pass law. And so I would like a world where the Biden administration and then the Trump administration and then whoever's president next are not sort of
giving us these competing EOs that create uncertainty from a business perspective and from a regulatory perspective by the way. Now,
Brett House (13:40)
Yeah, you see the EOs are reflective of kind of a disempowered, yeah, disempowered or ineffective legislative body in the government.
Rio (13:44)
the logjam in Congress, the fact they can't pass anything, right?
Chinnu Parinandi (13:52)
Yes, I think
they're reflective of an ineffective legislative branch. And I think they're kind of like a short term answer that isn't necessarily well thought out and is meant to pander to the constituents of a given presidential administration. Now, there was a second point that you were talking about with respect to China. And I think that I think this is important and I'm happy that you brought it up. I think that we all kind of like
Brett House (14:13)
Yep.
Chinnu Parinandi (14:19)
need to hit the pause button in terms of worrying to some extent about whether China is overtaking the United States and about kind of the strengths of the Chinese system. And the reason is that China has an authoritarian system and authoritarian systems can pursue innovation in certain certain areas. Right. So like the Chinese government can say we're going to invest all of this money in AI.
Brett House (14:35)
Yeah.
Chinnu Parinandi (14:46)
We're going to give people massive bonuses and incentives to invest in AI, but it's within these broad limits, right? There's always kind of like ⁓ boundaries that innovators are hitting up against within China because you can't disturb the political system or the political economic system in a way that threatens the legitimacy of the government there. And so we can see certain areas where the Chinese I think are doing quite well, but
I think there are limits that we aren't able to see because we're still kind of in this comparison mindset where we see ourselves being leapfrogged by the Chinese.
Brett House (15:24)
And that might
be, yeah, that might be short-termism, right? If you look at the history of China, I think about it, you look at the invention of paper, right? One dynasty brought paper to humanity and the next dynasty destroyed all the printing presses. And then they re-emerged in Europe, know, hundreds of years later. I don't remember the actual dynasty. Same thing with gunpowder. And it's interesting because it was sort of the ebbs and flows of...
Rio (15:25)
What?
And the next time you see destroyed all the printing presses because they didn't want this always a threat.
Chinnu Parinandi (15:38)
Yes.
Brett House (15:50)
Who was in power as an authoritarian state within China that changed all that?
Rio (15:53)
Well, but looking
at China, I mean, think that, you know, even before getting into AI, they've been very successful, supporting at having the state support specific industries, batteries, electric vehicles, drones, you know, the motors that drones use that are battery powered. Those are things they've been able to do at incredible rare earths, right? They've been able to ramp up at incredible scale. And then, you know, and then the accusations are,
Brett House (16:12)
rare minerals yeah rare earths
Rio (16:20)
dump and flood the markets and other places and drive businesses out. I mean, so I think that's a separate discussion. They've proven that it can be very successful in certain industries. no doubt. Like you go to China, the high speed rail is incredible, right? And what they've done. But looking at AI, mean, like on one hand, we have restricted some of the chips we were able to send there by Nvidia and some of our chip manufacturers. We put other restrictions on them. But you can even argue some of those restrictions have actually resulted in some products because
Brett House (16:30)
Yep.
Rio (16:49)
what necessities of other invention, right? So deep seek, I know deep seek is not, it's like 1 % adoption. And the part of that is because people are worried about it being back doors. the Chinese, half the AI engineers globally are in China, right? So most of the other half are here. ⁓ for sure, they're very smart people that even with these constraints, which will maybe force them to put out, find interesting workarounds to put out good product. But I think that the broader point though is like,
like the weird getting to before we start talking about China was like, like Congress is jammed, executive orders. You're, you're, you're, right. Like Biden makes one Trump reverses. makes another that's the next president's a damn. He cancels all the Trump ones. That's just not, that's just a bad way to govern. That's agree. looking like maybe what's happening at a state level to maybe that's kind of a reaction to the lack of anything at a federal level, right? Like the States are, are now with who are many driven by populism left and right.
They're the ones drafting a lot of these laws and statutes that we see that are targeting tech companies, they're targeting AI that are blocking data centers. Thoughts there.
Chinnu Parinandi (17:55)
That's a great point. So this is typically what happens in the US. So what's interesting about the states is that, and the reason this is possible is because of another difference between the US and China, which is that the US has politically constitutionally enshrined federalism, right? That 10th amendment in the bill of rights creates autonomy for states without fear of the federal government coming in and sort of claiming that it has the ability to usurp state autonomy. And so what that does
is there are 50 states and so 50 states can come up with their own different regulatory regimes. And this allows for much more experimentation than you might see in an authoritarian country where one actor really has its finger on what's going on. ⁓ And so typically in the US, you have the states crafting their own regulatory regimes before the federal government even does anything. And then that allows for the federal government to pick up successful templates. ⁓
Adopt those successful templates and then there's more of a feedback effect with the states where the states then amend and interact with the federal government to build on those existing templates. Now that happens under normal conditions, but then when you have like a dysfunctional federal government, it gets turbocharged at the state level because you have this, you know, other set of of actors or other set of venues where policymaking and policy action can happen.
Brett House (19:21)
Yeah, you think the state patchwork is at the center of this conflict? And do you think it does make us less competitive? I mean, I think that's the big concern, you know, that when you have an ⁓ environment where all the states are passing their individual laws, it makes for a minefield for any organization, especially the smaller ones that don't have massive legal teams, to navigate.
⁓ how to do this, especially when it comes to cross state issues. ⁓ How do you see that in terms of that competitive framework we talked about ⁓ comparing ourselves with China?
Chinnu Parinandi (19:56)
I think that it definitely creates ⁓ inefficiencies for small businesses, right? It's much cheaper and more economical for a small business to mobilize at the federal level because it's got one size fits all solution rather than having the resources to mobilize in 50 state capitals. In terms of the policy solution, I think it depends on kind of your sense of efficiency.
Brett House (20:24)
Yeah.
Chinnu Parinandi (20:24)
If
the concern is getting it right, the state solution might actually be better, but take longer, right? Because if you have a poorly thought out federal solution, then the negative costs of that federal solution are born across the United States. Whereas a poorly, let's say the Colorado bill is poorly thought out, right? The Colorado law, that poorly thought out Colorado law. I'm not claiming it is, I'm just using this as a hypothetical that poorly thought out Colorado law is only born out by Colorado.
Brett House (20:31)
Yeah, that's an interesting point. Yeah.
Yeah.
Chinnu Parinandi (20:53)
And you'll see the effects of that by businesses relocating to other states or more of an AI startup culture, migrating to other states, ⁓ et cetera. So then the US as a collective system will be able to course correct from the example of Colorado.
Brett House (21:12)
Yeah, yeah, it-
Rio (21:12)
Yeah, that's a good point.
mean, Colorado is not a big state, right? So let's say we write bad AI laws. Companies can just say, you know what, we're going to go to Texas or we're going to go to Wyoming. Or, I mean, that's why all the data centers are being built in Texas and Wyoming right now in places like that, because, you can kind of build whatever you want. There are far fewer regulations. So that's a really good point. But then you have a big state like California. I mean, again, I mentioned the CCPA, California Consumer Privacy Act earlier in this pod.
Like that passed and like most companies I know just use that as though that's the national law because it's California's such a big market. The same thing happened, if you recall, with automobiles, right, with the regulation for emissions. There's the California, like auto, no one's going to make two versions of cars, right? So they ended up just, let's just make everything to the California standards. So the California like cafe standards ended up becoming de facto the United States one.
Brett House (21:45)
That's the Mishallah, yeah.
Chinnu Parinandi (21:51)
Mm-hmm.
Rio (22:07)
I think you see, so that's an interesting point, but looking specifically at like the Colorado laws, how can a company, like if we want to be leaders in AI, how can companies navigate all these laws? Even, okay, you can go to another state, but like if you can be sued in all these other states, how do you, like the internet's global and like it's, can be a business in Colorado and have customers in Florida or vice versa. How do companies navigate this?
Chinnu Parinandi (22:35)
It's a great question. I think part of the answer is what you're talking about with California. So one reason that I think companies choose California's level of regulation is historically most other states have lower regulatory climates than California. So it's not just the size of California, right? 40 million people, fifth biggest economy, if it were an independent country, it's that you have a huge market that you're catering to and you can
satisfy every other state's regulatory climate by ascribing to the California ⁓ regulation. I think the issue that Colorado, with Colorado's law and with your broader question of how to satisfy these different regulatory climates, is that Colorado's law is really vaguely written and it gives a lot of discretionary power to the attorney general's office so that penalties can be made up on the spot.
And so that leaves companies in a situation where they don't know what to do. So if you look at California, historically through California regulations, they're exhaustive, but they're also extremely detailed. so companies, companies in business don't like uncertainty, right? This is like one-on-one and they don't like uncertainty because it's difficult to plan. It's difficult to innovate. It's difficult to figure out where you want to go next. If you have variability in your, in your kind of set of planning scenarios.
Brett House (23:52)
Yeah. Yeah.
Chinnu Parinandi (24:04)
California is a high regulatory climate, but there's low variability because what is demanded is known upfront. I think the danger with Colorado is that what is known, what is demanded is not known upfront. And so companies don't know how to adjust on the spot and that's disproportionately going to impact small businesses.
Brett House (24:19)
Well.
Yeah, and yeah, that's a very good point. But do you think that if it's a vaguely written law, ⁓ you know, not only does it create uncertainty in the environment, but does it actually diminish enforceability? Can it be a situation where where companies can effectively ignore the law because they know due to the non specific nature of how it's written, the vaguely written nature, it can't really be enforced specifically? Or do you think it just creates? Do you think it just creates like in market uncertainty, which just has a
Rio (24:51)
But that's variability,
Brett House (24:56)
deleterious effect across the board for small and medium sized businesses.
Chinnu Parinandi (25:00)
So
I think it's a great question. I think that the types of companies that are going to choose to attempt to not comply is predicated on the size of the company and the resources of the company's legal department. Right. So Google and Amazon and Metta, they can do whatever the hell they want. Right. Yeah. Exactly. They'll send an army of attorneys. They probably are already consulting with the AG's office every day.
Rio (25:13)
size. Yeah.
Brett House (25:15)
Yeah.
Rio (25:21)
They'll still have an army of attorneys. It doesn't matter to them,
Brett House (25:23)
Yeah.
Chinnu Parinandi (25:29)
And so they know what enforcement looks like. They also know that enforcement is like a moving target, right? It's not the same tomorrow than it is today. A small startup in Boulder is just not going to know what to do and will be kind of, you know, that person is going to be underwater thinking about this. And so my concern is that that person is going to decide to go to Salt Lake, right? Similar place, similar quality of life.
Brett House (25:35)
Yeah.
Yeah.
Yeah, so you're going to
be losing that innovation center, that innovation hub within Colorado because they're going to move to a state where they know either it's just less regulated. What your point about ⁓ Google and Meta, OpenAI, Google Meta, Andreessen Horowitz, they've all backed national standards instead of state by state rules. Back to that point we were making earlier that we have this state soup.
which I think we kind of went past this, but I think you made a really good point that the state soup is there for a reason. It kind of goes to the foundation of our competitive capitalist society, right? Maybe even European competitive societies, right? When you're encouraging competition or in this case, encouraging competitive regulatory environments, the likelihood that you get to the right answer seems to increase versus a top-down approach, which back to the China example,
where there's just one size fits all and it's not tested, it's not interrogated, it's not set up to compete against other models, right? So it almost sounds like you're arguing that the state by state piece is just how this, literally this country was founded. You'd argue that how modern European societies were founded, right? They competed against each other. They used that ⁓ gunpowder from China that China had banned.
to fight against each other, guess what? They then innovated, they developed national armies and then we can, you
Rio (27:17)
Well, yeah, Brett, there's an argument that you can say capitalism,
like modern capitalism really emerged out of like medieval Europe because of the, you know, all the principalities in the States and, fighting each other and trying to compete right in competition. Right. ⁓
Brett House (27:25)
and because of competition, right? So sorry
to get academic, but we're talking to an academic. I was just, know, but I, cause I'm seeing this sort of trend here. And are you saying that it sounds like you're in favor of the state by state approach, but you also fear the fact that companies might just leave particular areas that are building up innovation centers.
Chinnu Parinandi (27:46)
So,
so like historically in the other, like we already talked about auto ⁓ fuel economy standards, historically in other areas where there there's a patchwork of different state standards, what happens is that the federal government, so like state competition leads to externalities. Some of those externalities are good, but some of those externalities are bad. So let's take auto pollution or air pollution as an example, right? Colorado weather, like the wind blows from west to east.
Brett House (28:09)
Yep.
Chinnu Parinandi (28:16)
And so Colorado, could imagine, sets up all of its polluting factories on the Kansas state line. And the polluting air goes into Kansas. And this is like a classic example of a negative externality. So Colorado doesn't have to deal with the consequences of the policy, Kansas does. And so what happens is the federal government steps in for that type of interstate issue. And it created the Clean Air, passed the Clean Air Act, created the Environmental Protection Agency.
Brett House (28:30)
Yeah.
Chinnu Parinandi (28:45)
The purpose of the federal government is to like not mess stuff up, not mess stuff up anymore because of the, the, kind of excesses of state competition. And so what the federal government does is it sets floors, right? It says these are minimum pollution standards across the country that have to be met. And we have the ability to compel and compliance and assign penalties if compliance is not met. And the idea behind that is that Colorado.
should not be able to push negative externalities onto other states. Now, what the federal government's not doing, and this is where the debate comes, is the federal government is not trying to create kind of a standard, at least based on the theoretical aspects underpinning the Clean Air Act in the 1960s and 70s. The federal government is not trying to create a regulatory regime that outpaces the states and runs in advance of the states, right?
An individual state wants to do that, that's totally fine because state voters differ across like California voters want different stuff from Alabama voters. That's why there that's that's why there's much different policy regimes across those states. But the federal government, because it represents all states, should not create a standard that far outpaces what the majority of states want. And so that's where the federal legislation would come in handy with A.I. Right. So like. Setting the floor.
Brett House (30:10)
is setting that floor, right?
Chinnu Parinandi (30:13)
So that, you know, if there were, for example, ⁓ a deceptive business practice that was intentionally known, it would be so that a company or a, a, a, a deployer to use Colorado's language, a developer and deployer would not move to like Tennessee, right? Where they would be, be able to have like, you know, free rein to do whatever they wanted, right? That's where the federal kind of policymaking is useful.
Rio (30:42)
Well, I know the current administration has talked about some AI, like an executive order or law where they would basically prevent states from passing laws in for a specific time period with the understanding that legislatively at a congressional level, we would come up with some kind of law over a time period. I know they've been talking about that, but looking turning to the Colorado law. So SB 26189, I mean, like my understanding of it and again, not being an attorney, but
having done, you know, to Brad's point earlier, we've done some research, right? It seems to govern AI systems. And then I'm going to put quotes here that used in consequential decisions like employment or hiring, housing, lending, credit insurance. And it seemed to be implying that if there is a disparate outcome that could be tied back to use of AI, that could potentially, and this is where I think the vagueness comes in, that could potentially result in liability.
And then I guess I have a couple of questions out of this. Number one, we already have laws preventing discrimination in employment and hiring and housing and lending and everything listed here. We have laws already. Why do they need another AI law? mean, I'm not sure I understand that. number two would be like, how are they going to, I mean, is it, do they actually just have to prove that resulted in someone not getting hiring and they're in a protected class?
Or do they actually look into the model themselves and say, what in the model made this decision? Can you maybe elaborate on both those points?
Chinnu Parinandi (32:09)
Yeah, great set of questions. ⁓ I've also been trying to find out how this is going to be implemented and we won't know a lot about how the law actually works until we have these implementation issues. ⁓ So the first point I want to touch on is this idea of a consequential decision. It seems to me that what is happening here is that harm is defined in this really legalistic way, right? So harm here would be that like, you know, you could obtain a job.
In a non AI world where there is not an algorithm being used, you could obtain some employment, but it's the use of the AI algorithm that ended up being sort of pivotal or consequential to use their language in terms of causing a decision that created some type of harm for you. The way the harm is defined in the law is super interesting. It's not defined in terms of like a physical harm necessarily. It's a broad based harm that
is defined in a way where it would materially affect your prospects in life. And so it's a very broad category. I think that's important because that gives the AG's office a lot more leeway to pick up potential sort of claims against classes of individual. ⁓ In terms of the other part of the question, and I forget, sorry, what was the second part of the question?
Rio (33:33)
Yeah, yeah. The second part was like, how are they actually going to are they just looking at the outcomes from the use of you can prove your harm? That's all that matters. Or or if there is some kind of a disparate outcome or is or is it like let's say they look into that, they determine that. How would they then prove it was the model's fault? Would they then look into actually the the weights in the models? Would they look at how the actual training, the training run it went on? What material is Hoovered up during that?
Brett House (33:51)
you
Rio (34:01)
How would they actually look at the code base? I don't even know how they would do that or whether that'd even be legal. Any idea?
Chinnu Parinandi (34:08)
So that's a great question. think that from a legal perspective, you would have to be able to prove that like some counterfactual company where the model is not used resulted in a different outcome for the individual. But you know, we can't run the person through an experiment where they interact with an algorithm. Yeah, yes.
Brett House (34:26)
Yeah, that's like a control. You have a control
to your experiment, right? Exposed versus unexposed.
Rio (34:29)
Well, even within one model, like you
might get a different answer from the same model if you run it three times. You might get three different answers. That's how AI works.
Brett House (34:34)
Yeah,
begs the question, how do lawmakers, governments do not operate the speed of tech they haven't in the entire history of humanity, whatever the technology innovation was. So how do they fully understand the pace, the complexity of the AI systems they're trying to regulate without hiring people from the industry, then you kind of fall into the problem of people from the industry trying to regulate their industry. ⁓
Chinnu Parinandi (34:57)
So I.
It's a great question.
think that, you know, this is interesting. does not, the law does not create like a private right to action, which I think is important. And I think that's meant to kind of give some, ⁓ a breath of relief for businesses, although it's not much of a breath of relief. I think that what's going to happen is that we're going to learn through reg through compliance with the law and the filing of grievances about what the standard of grievances now.
I think this is a really important point. So every technically sophisticated industry has what's called a revolving door. So think about like pharmaceuticals, right? So everyone gets angry at the FDA that the FDA hires people from Eli Lilly and from Merck and from Pfizer. Well, there's a reason for that. Exactly. So what you're going to need, if Colorado goes through with this and enforces it to any degree of likes,
Rio (35:46)
Who else is going to understand the industry, right? They're going to know it best.
Brett House (35:51)
Yeah.
Chinnu Parinandi (35:57)
of competence is you're going to need lots of people from tech working for the government or working in concert with the government to answer these questions that you brought up. Right. Because currently. Yes.
Brett House (36:08)
Yeah, without any inherent bias.
That's often the case, right? They might have investments, they might have previously worked in companies, it's part of their network, it's what they built their career on. How can those people in the industry regulate their own industries? It just seems like a recipe for corruption. ⁓ Right? Isn't that the biggest challenge? So how do you solve, how do you get around that? I mean, I guess it's just you're legally, you you're putting your hand in the Bible, you're raising your right hand.
Rio (36:35)
Well, Brett, think they're
more worried about people going from government to business, right, than the other way around. Or maybe the revolving door where people keep going back and forth, right? You're in business, you go to government, then you pass some laws, you're able to go back and get a plum job back. I think that's more the concern, right?
Brett House (36:40)
Yeah.
Yeah.
Yeah.
Chinnu Parinandi (36:53)
think it is, but I think there are lots of domains where it's a necessity. And I think this is one where it is a necessity. And I think the difficulty here is kind of like, unfortunately, for the folks listening to this podcast, and for the folks in this room, I think the burden is going to be on those in the AI community and the business community to kind of improve the opinion of the public, the general public with respect to AI, because right now, unfortunately, the
Brett House (37:00)
Yeah.
Yeah, yeah, yeah.
Chinnu Parinandi (37:23)
I think the public's understanding of AI's capabilities, know, Rio started by talking about the economic benefits of AI. I think right now the public's opinions are entirely divorced from kind of what actually is happening on the ground. And I think that the political system is capitalizing on that distinction.
Brett House (37:39)
Yeah.
So there's a requirement for the industry itself to, whether it's a policy, ⁓ a lobbying body like the Interactive Advertising Bureau or some other technology sort of lobbying body that is partly responsible for creating ⁓ a good image of a particular industry in the minds of the legislators or the government, but from a general perspective.
Rio (38:03)
Well, tech, well, the AI, well,
Brett, the AI CEO is doing the worst job imaginable. These guys are like cartoon villains. Like, I mean, it's like, they go up every day, like Dario. mean, actually, I think Anthropix is an amazing company, right? But every day he comes up, talks about how many jobs are going to destroy this. Like every white collar job. And so you're saying it's not going to happen, number one, but it's like, it's not helping your case. I mean, it's, makes you, it you look like a Bond villain, right? But, but, but, but, but you know, for your point about the regulators, I think this is going to be a challenge. And Brett, mentioned this.
Brett House (38:07)
Yeah.
Yeah.
Yeah. There's some- Yeah, catastrophizing. Yeah.
Rio (38:33)
These things are, this is very complex technology. It's changing very quickly. It's hard for people in the industry to keep up. How can a government regulator, I don't want to say bureaucrat, that sounds pejorative. How can a government regulator, and think about it, if you're really good at AI, you can go make generational money working for any of these AI companies, or you can start your own company, right? Very easily. You can build some incredible stuff very quickly. So if you, like it's...
Chinnu Parinandi (38:50)
Mm-hmm.
Rio (39:02)
we're going have to pay people better or find better ways to compensate people where, know, where these are maybe just part-time roles, because if we want the best talent to be looking into these things, I mean, I'll give you an example. Summa, who's the CIO of Denver, she's incredible person, incredible tech leader. She was CIO of a technology firm. She was making, she was doing pretty well. Like she went to work for Johnson's administration.
took a massive pick. I don't know how much, I'm guessing she's probably making a tenth of what she was making. She's doing it because she loves, she wants to help. She believes in Mike's vision. She wanted to help the city. She's a public service minded person anyway. She'd worked for the state before, but I think that's rare to get talent like that for these roles, especially regulator roles. I think that's going to be extremely challenging. Moving forward, thoughts on that.
Brett House (39:31)
Yeah.
Yeah,
you've got to ⁓ just hope that people are going to be altruistic enough to be, know, the folks that are willing to go and work for a non-for-profit, know, folks that are willing to go and become an educator, right? You're an educator, right? ⁓ People that are willing to go into government versus ⁓ work, or in the financial services industry, you could go and work for the government, ⁓ you know, in finance, or you could go and work for a hedge fund and make, you know, ⁓
exponentially more money. how do you encourage people to beyond just altruism?
Rio (40:17)
Yeah.
Chinnu Parinandi (40:23)
think this is an area where like people are looking for information and this includes like the AG's office and regulators and even the legislators that crafted this bill think what you want, you know, we can think what we want of them, but they're still looking for information for for compliance and enforcement because you know, one thing that still can hurt legislators, especially at the state level and especially at the local level where there are these concerns of capital flight and job loss.
is a poorly designed, poorly carried out law. And so I think that this may be an area where businesses and government and academia can kind of all come together and talk about this. And I think your podcast is really helping with this. And there are some other institutions in the Denver area and in Colorado talk about this, talk about what good practices look like, talk about what ⁓ sort of algorithmic discrimination looks like from a tech perspective, right?
This is it's important to not cede the ground to the government because then the government's going to come up with examples that it wants to come up with. But if we kind of self monitor and self police and come up with some good standards as an industry, then that may be kind of creating the templates that government that government then uses because they're off the shelf.
Brett House (41:32)
Yep.
And it engenders
trust, right? And so you don't get these kind of draconian rules coming from the other side because you've got people like Dario catastrophizing about the entire destruction of a huge set of the white collar workforce. And it's like, that is the worst communications plan you could possibly come up with for a company like Anthropic who fundamentally is based on apparently good principles.
Rio (41:55)
Every week one of these guys says something insane.
Brett House (42:10)
as opposed to at least that's how they position themselves against OpenAI, right? As the good guys in this fight, we do good, right? So it is, I think sometimes you just have to shut up your CEO, right? Because they'll say some egregious things in market where it's like, where is, this is why you have PR people. Yeah.
Rio (42:27)
Well, who's going to speak for the industry, right? The industry, like
these guys are bond villains. They can't stop saying outrageous and inflammatory statements every week. like, I don't know, that's a communication problem, but actually looking at like the data center issue, know we mentioned this a little bit earlier. It's like 40 % of these data centers are now being defeated at a local level. And some of this is based on valid concerns, but think most of it is based on rumors and
Brett House (42:43)
yeah.
Rio (42:57)
exaggerations and outright lies like the, like I've done some looking into this, right? This data centers use a lot of water. They actually don't. use the new ones use almost no water in fact. And if you look at where's water used in Colorado, 90 plus percent of it's used by agriculture, right? California is the same thing. Almond growers use more water than any in California than consumers do, right?
Then you look at the power bills, like there's new regulation at federal level that the data centers have to be self-reliant on power and be off the grid and build a produce zone. So it actually would be a benefit for consumers because what like any excess power would actually go back into the grid and it would lower rates. like, it seems a lot of these arguments against data centers are not based on reality. They're based on either hyperbole or lies or propaganda. I mean, what's your take on this?
Chinnu Parinandi (43:48)
So data centers are a fascinating issue. think that like the politics and economics of them is still kind of nascent. My vantage point is that we've had disruptions in terms of energy availability and energy supply in the past. I think that this is something that we can solve through aligning incentives of institutional actors with kind of what
the data center energy requirements are. I'll talk to the energy part first, then I'll talk to the bands on them. So with respect to the energy part, there already are solutions that I think are helping to kind of deal with what the energy demands. There's tremendous investment in nuclear. ⁓ Part of it deals with sort of like in the energy community, we talk about distributed generation, which is like local generation and the rule that data centers have to have their own distributed generation.
This requires rethinking the traditional ⁓ electric utility model and allowing for more kinds of decentralized governance of a grid, which I'm all about. That's a good thing in my opinion. The other part of it is that currently the federal energy regulatory commission or FERC deals with interstate power issues. This is a case where sudden need for greater energy.
Brett House (44:50)
Yep.
Chinnu Parinandi (45:12)
⁓ requires a federal response. And you'll hear this a lot from both the business community and the political community. FERC really needs to get involved and have a whole kind of section dealing with like data center related demand and how to balance demand and supply across state lines within the kind of regional power pools.
in the cases where that demand suddenly shoots up. And right now FERC is not really involved in that, but needs to be. That's kind of my institutional answer. Now, public utilities commissions, which are the state level actors that are responsible for regulating energy within state, they have solutions. Once such solution is like a large tariff, right? We're talking about creating a large tariff that we apply to data center consumers. That I think is a good solution. Whether that solution is tenable in the future,
We won't know, but that's okay because we don't know, right? It's okay to not know because it's a new technology and that's kind of part of dealing with a new technological breakthrough is not having the regulatory answers immediately. Terms of the question of.
Brett House (46:20)
So you put decode that. you place a tariff on these data center operators in order to control for these attacks to potentially cover downstream impacts, right? That could be negative. Is that what it's for? Is it sort of a slush fund to support if there's legal response from the community, there's water issues and let, know, in ⁓
Rio (46:28)
It's like a tax basically.
Chinnu Parinandi (46:29)
attacks.
Brett House (46:49)
issues around that. I is that what's really the the taxes for? mean it's the
Chinnu Parinandi (46:52)
So the idea is that
in the instances before, ⁓ now we're moving into a world where data centers, many data centers have a requirement that they're energy, like self-sufficient in energy. But in the previous world where there was not that requirement, and given that the US is a patchwork, there probably are plenty of places where that still is not implemented, right? There's a public concern that data centers,
Brett House (47:05)
Yeah.
Chinnu Parinandi (47:20)
like that all of the energy capacity is going to go to funding data or to fueling data centers and that that's going to lead to an increase in rates on the on the residential side. Right. This is like where the public opinion is. And so what the tariff does is the tariff charges more for data center usage. Now, obviously, you can imagine a world where the data center is self-sufficient and no longer has to pay the tariff. That's wonderful. But the tariff is meant to. Yes.
Brett House (47:29)
Yep. Yep.
Yeah.
Yeah, and it probably encourages and incentivizes that behavior,
right, to actually build and support the build of additional energy resources like nuclear.
Chinnu Parinandi (47:55)
But what the tariff is meant to do is it's meant to not just incentivize that, but also utilize the proceeds from the tariff to maybe subsidize the energy costs for residential consumers. And so I think those are the sorts of things that data center ⁓ businesses need to consider in order to kind of change the public narrative around data centers. Now, with respect to the question of bans,
Brett House (48:08)
Yes, yes, exactly.
Chinnu Parinandi (48:24)
There's an easy answer for this and the answer is zoning practices. The issue is. We've had. We've had zoning changes before like you could just create a new class of zoning right? It's not. It's not residential. It's not small commercial. It's not industrial. It's like a new zoning class that where you set aside land for this particular purpose. I think that a lot of the a lot of the.
Rio (48:29)
Not building space.
They're thinking about it.
Brett House (48:36)
Yeah.
Chinnu Parinandi (48:54)
public opposition is just NIMBYism dressed up in a different way. And so one way of dealing with NIMBYism is setting aside areas that are exclusively geared toward this.
Rio (49:07)
Looking at like just more broadly, looking at the regulation we've had in Colorado. mean, have you seen any evidence that this is like stymied innovation has this driven companies out, has this impacted growth at all? know the law isn't actually in effect yet. It's been proposed, but I mean, I, every tech entrepreneur I talked to in Colorado is very concerned about it. And anecdotally, I do know a couple of people who have either slowed down plans or even moved. So is there any data that's come out or is it too early to understand what is the
overall impact on innovation on the tech industry here in Colorado.
Chinnu Parinandi (49:40)
Great question. think I don't think it's too early. And the reason I don't think it's too early is that the law was completely gutted and rewritten. And so and so that's the evidence that you need that there's a lot of concern about it. I think what you'll see is you'll see a really, really slow approach to implementation and that slow approach to implementation will be entirely reflective of the idea that I think politicians
most politicians I think know that the law has the potential to chase away businesses. And I want to get to this point of comparing us as Colorado to California, because I think that really matters. California can do certain things and get away with certain things because it's California. And what I mean by that is that there are 40 million people, there are tons of really like world class educational institutions, there are beaches, there's LA, there's a huge, yeah, it's California.
Rio (50:37)
It's California. Yeah.
Brett House (50:39)
No.
Chinnu Parinandi (50:39)
And so they can kind of like, they're the big person in the room. It's the same on the conservative side with Florida and Texas, but like California is the big person in the room. They can write laws and they have kind of like a captive audience because there are not a lot of States that can replicate perfectly what California has to offer in terms of workforce, education, amenities, et cetera. My concern with Colorado is Colorado is a really nice place to live. I've had lots of opportunities to leave and I'm still here, but
Colorado is not California. And so the danger here is to mimic what other states are trying to get in the arms race for credential based on being seen as overly aggressive when you don't have the same kind of ⁓ captive market that another state might have.
Rio (51:28)
Yeah, it's funny. There was an article in the Wall Street Journal a couple of weeks ago titled entrepreneurs flock to Colorado. Now red tape is driving them away. I mean, it's Wall Street Journal. It tends to be slightly right of center, but I don't disagree with the thesis based on what I'm hearing in the business community that there is a perception that the state is becoming through bureaucracy, regulations, and just the inability. I look even look, I live in Platte Park.
There's three construction sites that are big holes in the ground. One's been sitting there for a year. One's been sitting there for eight months. They can't get, they can't get the final red tape cleared up to actually break ground. And if there are holes in the ground, fences around them, it's crazy. And that's all purely because of like regulation and permitting. That's just really, really slow here. So I don't disagree generally with the article in a woods. I'm glad they, I mean, the new version of law, are you
Do you think it's much better or do think it still needs more work?
Chinnu Parinandi (52:29)
I think there has to be more detail in terms of what compliance looks like and what liability looks like. And I think that's something we already picked up on. And to the credit that you both have pointed out, we don't know those things. And so how much detail can there be? But the concern for me is kind of how it creates uncertainty in the eyes of business. And also, to be honest, in the eyes of consumers, I think what's really important to understand is that if consumers didn't care about AI,
The industry would never would have taken off, right? Like the idea here is that like there are victims and there are perpetrators. But what's really true is that this is a consumer driven environment, right? And so consumers benefit from this as well. Like I, I teach and, and I know lots of people that use Claude and I know lots of people that use AI and teaching and research and like it's, revolutionizing every single industry. And so that's a consumer driven thing. And so we also don't want to.
We also don't want to change or drive away like the consumer driven experience as well. I think that getting back to the crucial point or the crucial side of this, there is a need for regulation and every industry will eventually have regulation. But that regulation has to be well thought out and has to be organic. It can't be something that's sort of top down. So the EU is an example because they've had the most restrictive or comprehensive based on your opinion.
privacy laws and data laws in the world for almost a decade now. ⁓ Those laws are really detailed to the letter. There is some evidence that sort of ⁓ those laws have played a role in kind of reducing the size of the AI data tech market in Europe. Although
Rio (54:19)
There's not a single European tech company that's in the top 20 globally. I mean, it's really sad. I I've lived in Europe. I've done tons of work over there over the years. I mean, my guess would be that GDPR specifically has severely hampered innovation, especially in countries like Germany that really do take their regulations seriously. yeah, it's just my two cents on it.
Brett House (54:23)
Yeah.
Chinnu Parinandi (54:44)
I think there are.
Brett House (54:44)
Well, yeah,
and I think the facts back that up, right? From an economic contribution perspective, when you look at all of the European nations, even the biggest are significantly less, even in aggregate, than a lot of other countries like the US and China. Yeah.
Rio (54:57)
Your growth is lower, yeah.
Chinnu Parinandi (54:59)
I do think that there are kernels of elements in the EU law that are worth paying attention to. I'm not saying replicate them, but the idea that there are data officers in companies that talk to the EU. There shouldn't be an officer in every company because most startups, there aren't that many people and they're doing thousand different things. But the idea that you talk to business and you let business have
input really matters. And I think that's a way that Colorado can go forward. I think that the entire scaffolding to cut and paste from the EU and try to apply it to a state of six million people is a bad idea. And so my hope with this law is that the law itself is going to kind of be more symbolic and will lead to more of a balance or finding a middle ground between regulation and business.
One question that I do have with a lot of laws at the state level nowadays, especially given the polarization of the US, is to what extent are these laws kind of symbolic versus to what extent are the laws actually kind of, do they have teeth? And we're not going to know the answer to that.
Brett House (56:13)
Yeah, back to that enforceability
question. Yeah. Yeah. And it can be enforceability from like, are they going to actively pursue enforcement? And then that's one, I think one component. And then also can they pursue enforcement considering the complexity ⁓ behind these models and this technology and how?
Rio (56:28)
And how like, like,
like I remember CCPAs, like, so I was back when that was passing. I remember there was a couple of clients. I talked to them about it and I, you know, I wasn't not being an attorney. couldn't say you're going to be could get fined or not. guess certain practices you're pursuing here were probably would put you at risk. And that's that's the most I could usually say. And I one of my clients, I remember they did get sued when they didn't they didn't listen. And, know, they
Brett House (56:55)
And that's
exposure to consumer identity, some sort of consumer identity, PII-specific information.
Rio (56:58)
Yeah, right. Yeah. Yeah. Yeah. Yeah. Exactly. This was, this was,
this was PI. mean, basically like one of the things that CCP is someone like, if someone actually says it puts in a DSAR domain subject access request, which basically is saying like, want you to tell me what you've collected on me, who you, who you've shared it with and how you've used it. And then I want you to delete it and send me proof you've deleted it within 45 days or whatever. So
Brett House (57:18)
Yep.
Rio (57:21)
There were a lot of like, basically they didn't comply with that. They couldn't comply with that because the data was scattered and they couldn't match people together and whole supply chain of data. but then, but interestingly, we see CPA and I think a lot of these, like I remember with CAN spam, similar example, a lot of CAN spam people don't realize this. I was holding court multiple times by clients way back in 2004, 2005, by clients who were being sued by enterprising attorneys. Once this regulation was passed, right. The attorney is like, Oh, well, let's find some.
Brett House (57:26)
Yeah.
Rio (57:50)
Let's find some some marks, some companies that send a lot of emails. Let's opt out. And if they keep sending us emails, let's bring them to court. And there's fifteen thousand dollars per offending email. Right. So imagine you're an attorney. You blast out. You sign up for dozens of email newsletters. You unsubscribe. You find a few that don't unsubscribe you successfully. You go into court with a stack of printouts. Here you go. Exhibits A, B, C, D, E, F, G. You owe me like let's see. Either you pay me one hundred thousand dollars. This goes away or I'm going to sue you. You'll probably win because you're guilty. Right.
So a lot of the enforcement of these American laws ends up being done through our legal system, not necessarily the EU would be more through fines by the government. I think there's a different approach typically to enforcement.
Chinnu Parinandi (58:29)
And
honestly, I know that the legal system gets a bad rap for this, but this industry, I can't believe I'm saying this. This actually is less of a hindrance on innovation than the preemptive way. Right. So like we talk in regulation about police patrols versus fire alarms and fire alarms are actually a better method than police patrols. Right. Because police patrols are way more preemptive. They cost a lot more to actually administer and
they end up changing kind of the culture of innovation in a way that like a ex post fire alarm does not.
Brett House (59:06)
Yeah, it's like wait to get your hands slapped after the fact versus.
Rio (59:08)
Yeah. Well, honestly,
I mean, like if you find that there are some implications, right? If you end up incurring some liability, you'll change your practices. Right. And like these, you know, I remember, remember when I went to court, I had to basically testify to explain how the client did not violate handspan. Like they, this person opted out from this newsletter, but it wasn't, it was the type of opt out. So, and, then they did change their practices in all fairness. So maybe it's not the worst approach to the world instead of the heavy handed, like the government just sending big, altering people. Yep.
Brett House (59:34)
Yeah, which is going to prevent a lot of companies from starting to begin with,
Chinnu Parinandi (59:37)
Yes.
Brett House (59:37)
right?
Or just they're going to be fearful of navigating the complex landscape. And so they're going to leave. They're not going to start their company. They're going to do something else versus, yeah, the hand slap after the fact, which still encourages that innovation. ⁓ And hopefully it's not too... Yeah.
Rio (59:52)
You know, a lot of us risk reward, right? mean, you want to
use an entrepreneur, you're taking risks and you want, know, the reward is like, you can make a lot of money and you can, you know, do, do well in life. But you know, the, you know, the risk is that, okay, you could, you could, if the risk is too big, maybe, maybe you don't take that. If there's too much downside with like regular regulatory risk and then maybe, maybe you don't take, maybe you don't take your chances. don't know. So, um, are we, do you want to go to quick hits?
Brett House (1:00:17)
Yeah.
I had one question. says, yeah, we'll do quick. But let me ask this one question. So I think we've all agreed that fragmented AI regulation unintentionally strengthens large incumbents and discourages the smaller players in the space, which is not the type of competitive environment we want to create within states or at the national level. Are we all in agreement with that?
Rio (1:00:22)
Go ahead.
Brett House (1:00:46)
General assumption.
Chinnu Parinandi (1:00:49)
think that that that can be the case. ⁓ I will say that like it's sort of an open empirical question. Like we don't know about AI because we're living it right now, but it's an open empirical question about kind of like whether larger firms find it easier to kind of game the system at the federal level versus the individual states because there always will be a handful of states where these large firms will have less influence for whatever set of reasons. But
Brett House (1:01:09)
Yeah.
Yeah.
Chinnu Parinandi (1:01:19)
Given that it costs more to kind of lobby in 50 states versus one federal government, it does cost more generally at the state level.
Rio (1:01:29)
I don't know, Brett, my feeling in this is a bit nuanced. On one hand, I can see the wisdom in the model that you outlined earlier, right? Where states can come up with their own regulation. I mean, that's federalism, right? If the people in a certain state think they need laws to protect them, to protect consumers, then legislatures can pass that, right? I could certainly see the wisdom in that and having a federal approach where people in Colorado might care more about data privacy and about...
the use of AI and how it impacts their lives and people in Wyoming, which is right next door. People in Florida or Texas might care. Right. So I can certainly see the wisdom in that, but you think the other side of the coin is that AI really is a, mean, I don't think it's an understatement to say that this is the next space race, maybe even more important because the space race was more symbolic, right? Who's going to get to the moon first. This actually is who's going to create the models and have the companies that are going to lead to the next industrial age. Right. And then
Brett House (1:02:21)
is a new industrial age.
Rio (1:02:26)
Not to disparage China, but they are an authoritarian system. Do we really want the world to be run by their models? Probably not. If it's our models or their models, I think we'd rather have ours, because think ours will be the output of a freer society that has values similar to ours, that cares about the things we do, not about, and apparently the Chinese models, they love Xi and they love Marxism. Apparently they do.
Brett House (1:02:49)
They love the communist party. Yeah.
Rio (1:02:53)
That's how they're that trading material they use and the weights they put in those, those are hard coded in, right? You're not changing those. So I think you'd want you to say you'd probably want our model. Our models will probably be more representative of what we want to get out of our technology and like how the way we want to run our society. So I think AI is so critically important that if, if it's a, you look at the Colorado law, if it really is.
hampering innovation and driving companies out of state. I think it probably is, right? I don't have any evidence besides that Wall Street Journal article and what you said today, right? But if it is, mean, maybe we do need a national law to put the brakes on regulation a little bit. I'm not saying we don't need laws to govern this, but I think we certainly do. But it's probably too early to know what the appropriate laws will be. And if we put it in the wrong ones, it will hamper.
probably will hamper or could hamper innovations out of risk you want to take? I don't know. I don't know the answer. It's a very, very tough question and it's a really good one.
Chinnu Parinandi (1:03:55)
One thing I would add is that what shouldn't get forgotten is that consumers have choice, right? Like this is a market driven, yes, it is true that they only choose from things that they can actually utilize and that they have access to. So there is a creation of demand aspect that the AI developers are doing and the deployers are doing, but it is a mass driven phenomenon. And so it's important to never lose sight of that fact. If consumers did not want to interact with this technology,
Brett House (1:04:02)
Yeah, that's a very good point.
Chinnu Parinandi (1:04:25)
We wouldn't be having this conversation right now.
Brett House (1:04:28)
Yeah, and back to your original point around ⁓ how important it is to position and communicate from the industry, whether it's the software or the hardware, the data center side of the world, ⁓ to consumers in ways that's transparent, ⁓ that is open and honest, because there is a lot of misinformation out there. ⁓ You mentioned it, Ria, with the effect on water supply or the effect on
people's energy bills, you hear all these talking points around, ⁓ or hallucinations that are causing people to do certain things that they wouldn't have otherwise done because they're ⁓ getting into a romantic relationship with an LLM, right? People hear those stories and they're sensationalistic. Yeah, it is actually happening. Yeah, I think there was a guy, I read a whole article about a guy that committed suicide after, ⁓ no, he didn't commit suicide. He wrote a book about ⁓ his
Rio (1:05:10)
It's actually happening. Yeah.
Brett House (1:05:23)
but he fell into kind of a delusional love affair, kind of like that movie, Her, with AI, and it was after the death of his wife, and he was sort of replacing her with an artificial her, and so he literally built the AI system with the memories, with every documented piece of information he had about her to kind of recreate her. There's a movie that's called Her that's very good.
Rio (1:05:28)
It's great movie.
Brett House (1:05:51)
And then went into this, and this was a very smart, super educated guy, and he went into this sort of delusional hellscape where there had to be an intervention from family and friends. And he ended up ⁓ almost going through recovery like an addict and wrote a book about it.
Rio (1:06:07)
Well, if I do like this is not,
this is bro, this is not your wife. This is, this is a program that is very good at predicting and what she would have said. And by that is very good at mimicking the way she acted, but it's not her. That's, that's, that's wild. Wow.
Brett House (1:06:10)
Ha!
Yeah, and the
larger story here, and in fact, to all this point of this whole podcast is that that that is one situation. Yes, there are risks, but let's not overplay that hand. Let's give a balanced viewpoint regarding the economic rewards, the effects on our environment or electricity bills, on our productivity, on white collar jobs. Let's give a balanced perspective. And I find that just like the polarization of our political climate.
you get this radical like it's either one extreme or you swing wildly to the other extreme.
Rio (1:06:48)
It's a no regulation or we're going to like suffocate it, but it's like anything
like you think of cars. Okay. So like we, okay, cars got invented. people, cars are dangerous. Are we going to ban cars? So we can't do it because it could create risk. No, but let's, let's have safety standards. Let's have seat belts, right? Let's have airbags. So I think AI is ideally over time, we come to a similar approach where you have this thing's going to need guard rails because any technology can be used. Like you can use fire to fire to cook, cook a food or you can burn your neighbor's house down. Right. So any technology was going to have.
Brett House (1:07:01)
Yeah.
Yeah.
Yeah. And nobody
would call the automotive industry evil, like they are in some cases calling these AI companies. Yeah. And Ralph Nader had to push for seat belts being implemented in the 60s, which the automotive industry, Ford and GM and the rest of them pushed back hard because it would increase cost and production. then those costs get passed on to the consumer. no one's saying the automotive industry is evil. But there are people swinging way
Rio (1:07:18)
Good and bad uses.
Some people have over time and they've.
Brett House (1:07:43)
⁓ extreme against some of, ⁓ the AI companies, which I just don't think is a balanced perspective. And I think part of it is a communications problem when you have Dari out there saying that, you know, you know, your jobs are at risk across the board. ⁓ that's not good PR. And so I think, I think you made the good point, Jita that, ⁓ yeah, there needs to be ⁓ a kind of an honest communication around, ⁓ the pros and the cons and
And if we do that, if there's more communication between business and government, we'll have better outcomes ⁓ at the end of the day. Right? Is that accurate?
Chinnu Parinandi (1:08:18)
Yes.
Yes, I think so.
Brett House (1:08:21)
So quick hits. Should we do quick hits? We're nearing the end. Thank you, Chiddy, for staying along with this fun conversation. ⁓ So I'll start. So should AI regulation happen federally? This is the of the premise of this entire episode. ⁓ Instead of state by state, you've outlined some of the positives and negatives of both approaches. What is your final call on that?
Chinnu Parinandi (1:08:26)
Definitely.
Rio (1:08:27)
This has been great. This has been really good.
Chinnu Parinandi (1:08:46)
So I think that it should happen federally, but the federal government should play two roles. One role is to identify like worst case scenarios or like with the pollution case, like a lowest set of standards ⁓ to prevent sort of venue shopping by firms across states ⁓ that want to exploit the lack of standards. The other area the federal government should actually play a role is through grant giving, which it traditionally has done to encourage innovation.
Brett House (1:09:15)
Yeah.
Chinnu Parinandi (1:09:16)
that sort of like, you know, capital intensive projects that public governments can't sponsor that companies may not have the ability to sponsor. So I think the federal government has a natural role in those two areas. I'd add that with the energy side, FERC has to play a role in balancing data center demand with existing supply across all units.
Brett House (1:09:39)
Keep crying.
Rio (1:09:39)
Yeah,
it's a little of both. Is AI safety becoming too broad of a political category now?
Chinnu Parinandi (1:09:48)
I think it is. It's been kind of detached from examples of safety. And we could talk about that for hours, but I think it's sort of like a catchphrase for anything tech related that could cause issues.
Brett House (1:10:06)
Yeah. So will AI eventually become regulated either like a utility, right? In the case of the data centers or in case of the software, more like a finance or heavily regulated industry like healthcare.
Chinnu Parinandi (1:10:19)
I think it will. It's too early to see how it's going to be regulated. My guess is that it's going to be regulated by end use. So that's what you just said. Like, so the financial services regulators will regulate AI finance, healthcare regulators will regulate AI healthcare, energy regulators regulate AI energy. I don't think there's going to be a totally separate way of regulating AI because regulators work within existing templates. But I think that every industry that's ever been created
Brett House (1:10:34)
Yep.
Chinnu Parinandi (1:10:48)
has eventually found its way to regulation. And I don't think that this is any different.
Rio (1:10:57)
The growing anti-tech movement in America, do you see it getting worse or getting better over the next couple of years?
Chinnu Parinandi (1:11:05)
It's a really good question. think it depends on where populism goes. If populism as a resonant ideology on both the left and the right recedes a little bit, I think that things could get better for tech. If populism entrenches and remains, it's still going to be a pretty kind of us versus them kind of environment.
Brett House (1:11:15)
Yeah.
Yep. And do you think consumers are more protected today than they were, let's say, five years ago?
Chinnu Parinandi (1:11:36)
I think so. I think they're more protected. And I think that like, to some extent, we're sitting here talking about this law, because there's an awareness of risk that there was not five years ago. You know, we talked up at the example of the movie Her because there are these templates that exist. And so I think to some extent, that lends itself towards more protection. ⁓
Rio (1:11:37)
or less.
Brett House (1:11:49)
Yeah.
Chinnu Parinandi (1:12:01)
Part of it is just that we're gonna have to wait and see how this manifests itself. And I know that's not a concerning answer or that's not a comforting answer for a lot of people. But one example is like, you know, like a new drug gets developed and prescribed and you take that drug and you don't know what the after effects are. Like you may not know the long run effects of a drug for 20, 30 years, but that doesn't mean you're not gonna take it if it's something that can help you, right? So I think that kind of analogy is useful here as well.
Brett House (1:12:07)
Yeah, I'll get certain answered
I think that about does it. That was a terrific conversation and your stability in your car held up the whole conversation. I mean, remarkable. ⁓ You've got good broadband 5G connection there. So thanks everybody. Well, thanks, Jennifer, for joining us. This was a terrific conversation. Certainly love to have academics on that are studying some of these policies and doing deep research because your expertise is certainly going to be, I think, appreciated by us, and I for sure. But I think our audience
Rio (1:12:32)
That was good. That was fun.
Chinnu Parinandi (1:12:37)
you
Rio (1:12:38)
Yeah.
Brett House (1:13:01)
in general. And thanks everybody for joining. Visit us at www.signalanoids.ai. We're also on YouTube, Spotify, and Apple podcasts. And we will be at Cannes International Lions Festival of Creativity in 2026, come June 22nd through the 26th. So yeah, we are partnering with Hearst.
Rio (1:13:21)
You'll be there the whole week.
Brett House (1:13:25)
⁓ in doing a really exciting executive ⁓ interview series at the Hearst House. So we'll see you there. Thanks everybody.
Rio (1:13:35)
gonna be fun. Thank you.
Chinnu Parinandi (1:13:36)
Thank you for having me.
Comments