A Conversation with Harvard Professor Bruce Schneier on AI, Democracy, and Money in Politics

AI is reshaping our world. Money is reshaping our politics. What happens when they collide? Tarik sits down with Harvard Professor Bruce Schneier to unpack the algorithms and dollars quietly rewriting American democracy, and what it'll take to fight back. 

Check out tarikforcongress.com to learn more and get involved.

Transcript

SPEAKER_00 0:00

First of all, I want to thank you, Professor, for joining the call and giving us some of your time. In this call, we have Professor Bush Neyer. He's a public interest technologist and a fellow and lecturer at Harvey Kennedy School and a member of the Electronic Frontier Foundation, which is the leading nonprofit organization defending civil liberties and the digital world. And he's the author of great books like Rewiring Democracy, How AI Will Transform Our Politics, Government, and Citizenship. So, Professor, my first question for you is what do you mean by a public interest technologist?

SPEAKER_01 0:27

So this is not my term. This is a term of art. And this is someone who combines tech and policy. My background is technical. I have a computer science degree, but I teach at a public policy school. I teach at the Harvard Kennedy School. I am teaching people who didn't take math in college, right? Because so much of our policy is tech driven. And it's not just AI policy, it's energy policy, it's defense policy, it's medical policy. All of these things have a strong technical component. And we need people in government who can straddle both. Either technologists who are doing policy, policymakers doing technology, because it's all combined. The internet is everything. Computers are everything. So that's what we mean by that term. And there are a lot of us out there, and and we work uh in government, at NGOs, in corporations, doing tech and policy together. Either the policy of tech or the tech of policy.

SPEAKER_00 1:26

Your generation is one of the most interesting generations because you grew up without computers or internet from the perspective of a public interest technologist who seen this technological advancement. What was your thoughts? What was the public sentiment? And how was the government reaction to it?

SPEAKER_01 1:38

Uh in general, I think of technology in terms of who it empowers. Does it make the already powerful more powerful, or does it empower the less powerful? And the history of computerization is really the quick versus the strong. It is um you know political change in Eastern Europe, it is the dissidents, it is the uh unsophisticated, it is the marginalized who are using this technology to organize, to communicate, to do things they never could do before, right? It empowers the masses. The already powerful, the companies, the governments, are much slower to adopt these technologies. But when they do, they have raw power, more raw power to magnify. So now we see these technologies used for surveillance, for control, for censorship in ways that the marginalized had a lot of trouble pushing back on. Right? So the quick are faster to use the technology, the strong use it more effectively. And I think we're seeing the same thing with AI played out a little bit differently because you have from the beginning the tech monopolies controlling the technology. So you didn't have for computers or the internet. It was much more distributed.

SPEAKER_00 2:55

I've read an article that you've written where you talk about an election that happened in Japan, where AI was used by a candidate who portrayed himself to be representing the youth. You know what I'm talking about, right?

SPEAKER_01 3:04

I do. It's a great story. So uh this is a story of uh a young man named uh Takahira Ano. And last year, or maybe the year before, he's running for governor of Tokyo. He's a programmer, he's not a politician, and what he does, something no one's ever done before, is he has a two-week live stream on YouTube where an AI avatar answers questions. An authorized deep fake answers questions. He comes in fifth out of out of like 30 candidates. It's kind of crazy. Right? So he's the first one to leverage AI as a legitimate campaign tool. Uh this would be, you know, kind of a cute uh story, except a few months ago, I think actually late last year, he wins a seat in Japan's upper house, in their in their, I don't know what it's called, their diet in their upper house. On this sort of pro-technology platform, he's not like left or right, he's future instead of past, is the way he describes it. His party is a new political party. It's called Team Murai, Team Future, and he's building tech tools for democracy. So in Japan, when you're a political party, you get so many seats in the House, you get money from the government. And he's not doing that for campaigning or party building. He's doing it, using that money to build tech tools that he's making available to everybody in government, no matter what the party is. Uh, in the most recent election a couple months ago, his party won a few more seats. So he's got like a dozen seats, which is not a lot, but it's not insignificant. And he is very much building AI tools for listening to constituents, to voters, for crowdsourcing party platform and legislation, for trying to build a responsible mechanism for government using AI. It's a fantastic story to me because we hear all about how AI is terrible for everything. And it turns out, in the hands of someone who wants more democracy and better democracy and more responsive democracy, AI can do that.

SPEAKER_00 5:01

The funny thing, if I try to run a similar campaign as his campaign here in America, I'll probably lose the last slide because of the sentiment that we have, especially among the youth, towards AI.

SPEAKER_01 5:10

It is interesting. I I very much, when I when I when I talk about this, I tell people don't conflate AI as a technology with the for-profit companies that are controlling the technology. A lot of people's, I think, very real fears and reservations about AI. Are really fears and reservations about the AI companies? It's not the tech, it's the humans controlling the tech. AI, as we said, is a power-enhancing technology. Do you want that power to go to Google and Meta and Elon Musk and Anthropic and Open AI? Or do you want that power to go to democracy, to the people? Right? That's the question. It's not the tech, it's who's wielding it, why and how.

SPEAKER_00 5:54

So shifting back to American politics, what do you think politicians in DC are missing about AI?

SPEAKER_01 6:01

So I think that, you know, confusing the uh tech with the politics. I think uh we're not taking seriously enough the job loss and revenue loss. Now, recently we've seen uh some changes. I mean, Bernie Sanders, just like last week, proposed that the US government take a stake in these AI companies. Elizabeth Warren, I think, is a better proposal, proposed basically a token tax, an energy excise tax on these AI companies. I like Warren's proposal better. I think in Sanders' proposal, we become complicit in these companies. And I don't want government to be in the role of investor or capitalist. I want government to be government. But the idea of an excise tax, that that how do we as a society capture some of the profits generated on the fruits of all of our labor? That seems like a good idea. So so I I think these ideas to be taken more seriously. How to do this? Uh, certainly in the US, uh, we're not thinking about regulation at all. I mean, even uh, and you see administration, right? Trump had an idea that a few weeks ago that all new models have to go to the government for approval. Someone whispers in his ear, and two days ago we got the actual order, which was pretty pleased. It would be nice if the companies would show it to the government, but no requirement, don't worry about it. We need actual serious regulation. Here, the EU is better than the US. I mean, I think we can complain about some of the details, but they're trying in a way the US is not. And the last thing that I think really needs to be taken seriously and not is how fast this is all changing. If you knew how AI worked four months ago, it doesn't work like that anymore. If you know how it works today, it's not gonna work like that in four months. The technology is changing faster than I think anybody who is not deeply seeped in it realizes.

SPEAKER_00 7:39

How can government intervene and enforce regulations on AI while at the same time doesn't make the government too powerful, especially when it comes to surveillance and data collection?

SPEAKER_01 7:47

You know, actually, I'm less worried about the government being powerful. I think the companies are too powerful. And right now, I mean, especially in the US, right? Income inequality, money in politics, all the reason democracy doesn't work. The problem is that money drives politics. The problem is these companies are too big. Single biggest thing is breaking up the tech monopolies. I mean, none of this is going to happen, so we can just talk about it, but breaking up the tech monopolies would do wonders for competition, for price, for innovation, for all the things that we claim we want. That would be the first thing.

SPEAKER_00 8:16

It's funny you mentioned it because the thesis of my campaign is that money and politics is the root of all issues that we are facing in today's society.

SPEAKER_01 8:22

And I think you're right. And income inequality is the root of that.

SPEAKER_00 8:26

The root of money and politics.

SPEAKER_01 8:27

Right. I mean, money and politics, I think, has its own problems. I mean, the fact that uh I mean citizens united uh kind of enshrine the notion that uh you know money can buy politics. But it really is the income inequality that you know these mega billionaires can buy politicians with pocket change. There's a uh uh writer, Hamilton Nolan, I like reading. He writes a lot about labor. And he makes he wrote an essay a few months ago. They basically said, look, anything you can buy with a billion dollars, you shouldn't have. Nothing costs a billion dollars that you should have. That's like an election. Nothing normal costs a billion dollars. Only scary things cost a billion dollars. And and you know, we have people on our planet, in our country, from a billion dollars is okay, I can spend that.

SPEAKER_00 9:07

I'm assuming the biggest fear you have right now is not necessarily the influence of AI is the influence of money in politics that give too much power to corporations who have control of AI.

SPEAKER_01 9:16

That's right. I mean, it's remember, AI is not scary, AI doesn't do anything, it's the humans controlling the AI that do things. So, yes, I really do fear uh the people, the humans.

SPEAKER_00 9:28

I think one of the recent articles that you've written is about uh anthropic creating a tool to find issues in software, and then they decided not to make that tool available for the public. Why do you think uh it's important to pay attention to it?

SPEAKER_01 9:40

So, first I want to say that it's non-anthropic. This is a fantastic PR move for the company, because everyone now says anthropic is the best. They are one of the models. So what so but the basic story is that the good models, and there are many of them, some are open source, some come from Europe, some you can run on your computer, some are anthropic, some are open AI, some are others, are very good at finding vulnerabilities in software. This has benefits for attackers and defenders. Attackers can find vulnerabilities and break into systems, defenders can find vulnerabilities and patch them, thereby securing systems. And now we are sort of living through all of that uh turmoil. It turns out these frontier models, the uh the Claude Mythos was the OpenAI model, also the GPT model, are very, very expensive to run. They're extremely energy inefficient. And uh I think that's the real reason that uh Anthropic didn't release their model uh to the public. They made a virtual out of necessity. But we're gonna be living in a world where finding these vulnerabilities is uh is cheap. And that's gonna be interesting to watch. Right? It'll cause a lot of security because companies will use these models to find vulnerabilities and patch their systems. I mean, last month Firefox, the browser, patched, I think, 271 vulnerabilities found by AI. Use Firefox, it is way more secure than it was two months ago. That's AI. It does good and bad. You have to always have to look at both.

SPEAKER_00 11:05

Imagine a world where we were able to eliminate the influence of money in politics. And now AI is a tool that you use for the people and by the people. How would that world look like?

SPEAKER_01 11:16

So it's interesting. Uh, in our book on AI and democracy, we talk a lot about something called public AI, the notion of an AI model that is not controlled by a for-profit corporation, either by a government or an NGO or a university or some consortium, like an AI model that was built not in the profit motive. And that this would be a really valuable, not replacement for corporate AI, but addition into the milieu. We write about this uh in our book. Uh it was aspirational, but in October it actually happened. Switzerland released a public AI model called Appertuse. It was uh built by uh consortium universities, funding from the government. It uses uh no uh illegally copyrighted material, it uses no poorly paid global south labor. It was entirely made using renewable hydropower, so it's clean AI. Uh, they use an existing data center, so no new rare earth minerals were mined to make this AI. It might be the first fully ethical AI on the planet. Now, it's not as good as the frontier because they weren't aiming for the frontier. That's what the capitalists do. They wanted to build a useful model at a price point and performance point that would work for everybody. And they did that. Uh, they're gonna come out with a new model, I think, later this month. But here's an example of how we could build AI in a different way. And so I want to see that. I want to see real uh regulation on what these models can do, transparency requirements, disclosure requirements. I want I want some kind of universal access. I want the the companies that are making orange lots of money destroying jobs to somehow figure out how to give back. That's some of the uh token tax that we've been talking about. There's a bunch of things we can do if we choose to. I mean, right now we're designing these AI systems for the near-term financial benefit of a bunch of white male tech billionaires. It's like a dumb way to organize society. We could do better. It's unlikely we will anytime soon, but it's certainly possible.

SPEAKER_00 13:09

What piece of advice you would give to the youth who are watching this interview?

SPEAKER_01 13:14

Uh that uh to really think about this. I mean, I we're living in a very, very tumultuous time, and I get it. You know, I mean, I'm I'm teaching here and I have students who see their career path disappear. And what do we do now? And we need to figure this out. We need to, as a group, move politics, democracy into the 21st century. Right now, it is stuck in kind of the mid-1800s. That's the way how these systems were built. They were built when travel and communications were hard. They were built for an era of industrialization. They're not working in the information age. And we need to fix that. So I want people to get involved. It matters. Though the worst thing I see is for young people to say today to say, oh, elections don't matter. Republicans and Democrats are all the same. I get it. Democrats suck, Republicans suck more. You might want to edit that out.

SPEAKER_00 14:06

Yeah, okay. Let's say we were able to eliminate money in politics. Give me three top regulations should be implemented immediately afterwards.

SPEAKER_01 14:17

Uh biggest thing is antitrust. The biggest thing, the single most effective thing we can do is break up the tech monopolies. A hundred percent. Nothing would do more and do better than to break up the tech monopolies. So that is the first thing. Uh God, after that, you know, I want uh broad regulation uh protecting employment. I mean, too much of tech is not really tech, but is figuring out how to get around uh employee protection laws. I think basically Uber is an entire scam to get around worker protection laws. And a lot of this gig economy is. I think we need to fix that in general. So labor law writ large. And then after that, I think uh data protections. I mean, the reason I mean surveillance being a business model is kind of crazy, like spying on people and then manipulating them as a business model. Why do we allow that? Well, that's Google, right? That's everybody who, you know, this whole advertising ecosystem.

SPEAKER_00 15:08

Could you talk more about surveillance? I think you you were at the time when Snowden um first exposed the government for you know surveillance. How scary is surveillance?

SPEAKER_01 15:20

So surveillance is not new, right? I mean, like, you know, this device has been spying on you since you put it in your pocket. Uh Facebook, Google have been spying on you since they uh came into existence. And and that kind of broad surveillance, you're right, was around in the Snowden's time, which is 2014. It's over a decade ago. And there, the government and corporations could take these big databases and figure out where you go, what you buy, who you talk to. All this is the metadata, right, of all of these, uh all of these things. Uh what's different is with AI is AI can listen to the actual conversations. Right. So surveillance will say that you and I are having a Zoom right now. Right. And the date, the time, who's on the call, that's all surveillance. The spying is what we're talking about. And before AI, a human had an eavesdrop on this conversation to know what we're talking about. Now an AI can.

SPEAKER_00 16:06

I've read one of your articles where you said how the language that AI uses is the language from textbook and you know from the text, and that might influence the way we speak. If this is like one implication of using AI in our life and its implication on our society.

SPEAKER_01 16:20

No, it's not just AI, it's autocomplete in general, which is older than AI. And the notion that our language is affected by what we read and what's suggested to us. So if uh you know I'm typing an email and the email provider, let's say Google, you know, pops up suggestions of what I could say, it's easy to just to click OK. And that changes my language. And you receiving my email, see this new language, which isn't really me, it was the computer, and that changes your language. So we are now being affected by what the computer is saying. Uh AI supercharges that. AI can write the entire email, and I just click OK, or maybe I change a few words. All right, so now it is really the AI speaking and AI talk that is becoming dominant. And I think we see this. You go on Reddit, you look at uh, you know, comments on Substack, a lot of them written by AI.

SPEAKER_00 17:08

Written by AI. I think one thing that's also been written by AI is student assignment and articles in schools. So I'm totally true.

SPEAKER_01 17:15

I I I teach here and I guarantee you that most of the stuff I read is written by AI. You know, I don't know what to do about that. I think we all have to figure out education, I need to figure it out. I tell my students that having an AI write your assignment is like having a robot lift weights to you at the gym. Like, why are you bothering? The whole point of lifting weights is not to move the heavy things from here to there. It's the act of lifting makes your muscles stronger. I mean, the reason I assign my students the policy memo to write is not because we need more policy memos, it's because the act of writing a policy memo makes you smarter. And if you're gonna outsource that in AI to an AI, like why are you coming to grad school? But I get it. I mean, I mean, I think the problem is, you know, goes back. If the goal of writing an assignment is to get a good grade, and the goal of getting a good grade in that assignment is to get a good grade in the class, and the goal of the that is to get a good GPA, to get a good job, to pay off the student loans you had to take out to go to school in the first place, then cheating is a perfectly reasonable strategy. And so I actually blame the system that forces students to take out obscene loans to go to school.

SPEAKER_00 18:20

But do you imagine a future where AI could be used to help us um represent the society even more, to help us have a fair elections?

SPEAKER_01 18:29

I mean, 100%. We talk about a lot of that in our book. Uh, and it's certainly possible. Again, AI is a tool for the humans who uh who wield it. If the humans want more democracy, AI will help that. If the humans want less democracy, AI will help that too.

SPEAKER_00 18:42

What message would you send for a Congress representative right now? Uh, who are not talking about AI much because they're afraid that saying anything about it might kick them from office because you know the meta attacks and so on and so forth?

SPEAKER_01 18:53

I mean, it's hard. We we need to push back on that, but I get it. Right? I mean, money and politics is the problem. The influence of these corporations is a problem. That is a core problem. I mean, I I want to solve that. And maybe we have to solve that before we can have any AI regulation.

SPEAKER_00 19:08

In one of your articles, you you wrote about um who can harness AI for campaigns, and you talked about Obama and how he was uh you know the first presidential candidate to use social media and his advantage. So when we talk about AI and campaigning, how do you imagine that?

SPEAKER_01 19:24

You know, so I think about uh that Japanese story, right? I think of uh Anasan and uh him using an AI avatar. Uh right now there's a shorter number of AI tools in uh for campaigning. There are AI tools for uh connecting candidates with donors, there are AI tools to help candidates run for office, uh collect signatures, produce a web page, position papers. And here, don't think senator, don't think president, think city council, think school board, think people with no money, no expertise, no time. But AI tools will enable more people to run for office. I think it's phenomenal for democracy. Uh there are AI tools for get-to-vote campaigns, for door knocking, for canvassing, right, to make that more effective and more efficient. So I think AI can affect all sorts of uh other areas of politics. We're seeing you know AI-powered uh media and not the horrible stuff, not the deep fakes, not the not the propaganda, but real things. You know, producing position papers, connecting with voters. Can an AI have a conversation with a voter instead of and you know, it's it's sort of not the same as like you talking to a voter directly as the candidate, but it's better than the you handing the person a piece of paper and they read it. It's a conversation with an AI avatar. I mean, a lot of this, I think there's gonna be some uh uh acceptance barriers, but I think that's all changing. I think young people are more likely to accept these things, and I think it's going to I think it's gonna change.

SPEAKER_00 20:43

Because now our district is considered one of the most educated districts and top universities. And so what would you tell them if you have a message to them?

SPEAKER_01 20:51

You know, this is the time to stay involved. And uh, I mean things that are happening are important. I mean, everyone always says this is the most important generation, the most important election, the most important this. And the reason is because it's the only one you can affect.

unknown 21:01

Yeah. Right?

SPEAKER_01 21:02

The ones in the past are over, the ones in the future aren't here yet. This is the most important election because the one that's happening now. And and just remember that.

SPEAKER_00 21:08

The last question I have for you is where can people read more about your work?

SPEAKER_01 21:14

So I'm easy to find. Uh, you type Bruce and Iron into uh Google or whatever your search engine is, and you'll get my website where all my writing is. Uh my essays, my blog posts, you read about my books, it's all there. I'm not on social media because I'm weird, but my website has everything.

SPEAKER_00 21:28

Yeah, thank you so much, Professor, for joining us for this call. It was really interesting to hear your own perspective.

SPEAKER_01 21:33

Thank you, and good luck.

SPEAKER_00 21:34

Thank you.