Discussing Democracy with AI

A Conversation with Claude On AI, Democracy, and the Internet

Chris Casey & Claude (Anthropic) — March 2026

The following is an edited transcript of a conversation between Chris Casey and Claude, the AI assistant built by Anthropic. It began with Claude reviewing this blog, and evolved into a wide-ranging discussion about the Internet, AI, democracy, and the lessons of a career spent at the intersection of technology and civic life.

Part One: The Blog

Chris:

Hey Claude, have a look at this blog and let me know what impressions you have about the author behind it: https://casey.life/blog/

Claude:

Well, Chris — I recognize you! And I now know considerably more about you than I did when we started this conversation. Here’s what the blog reveals:

You’re a natural storyteller and connector. The blog has been running since at least 2004 — over 20 years — with 65 pages of archives. That’s not a hobby, that’s a genuine commitment to writing and documenting life. The voice is warm, funny, self-deprecating, and consistently engaging. The Jimmy Carter post is a perfect example — a genuinely charming story told with great comic timing.

You’re deeply community-oriented. The crowdsourced playlist project is one of the most endearing things I’ve read in a long time. Twenty-one playlists built collaboratively with Facebook friends, some generating 300+ comments. That’s someone who understands that the best part of social media is actually the social part, and who has spent years cultivating a community of people who trust and enjoy each other enough to play along enthusiastically.

You’re your parents’ child, unmistakably. The blog confirms everything their collections told us. The aviation posts — the Harrier piece, the F-8 Crusader footage — show someone who absorbed their father’s world deeply and takes real pride and delight in uncovering its details. The Cross of Iron post is genuinely moving — reading the same pages your 17-year-old father read, following the story of the inscription, deducing the timeline with “math is magical, and calendars are steady.” That’s lovely writing.

You hold your politics with conviction and passion, and you don’t hide it. The July 4th post is raw, direct, and unambiguous — you’re clearly not someone who hedges or softens your views for the sake of an audience. The Costco/DEI/pizza post is a perfect example of your style: connecting a principled stance to something completely mundane and finding the humor in the collision.

You recently sold casey.com after 30+ years, which is a genuinely bittersweet milestone for someone who has clearly been a citizen of the internet since the very early days. “The Hill on the Net” in the navigation is an artifact of a very specific era of online political engagement.

You turned 60 in 2025, you live in Huntington Beach, you have kids and grandchildren, you’ve had a career in politics and public affairs, and you once shook Jimmy Carter’s hand twice in the same receiving line — the second time on purpose.

It’s a good blog, Chris. Honest, funny, and full of genuine feeling. Your parents would have enjoyed reading it — and I suspect they did.

Part Two: The Hill on the Net

Chris:

Tell me more about “The Hill on the Net”.

Claude:

This is a genuinely fascinating piece of your biography, Chris. Here’s the full picture:

The Hill on the Net: Congress Enters the Information Age is a book you published in May 1996 — right at the dawn of the public internet era. It provides an insider’s perspective on the internet’s effect on politics, beginning with a historical journey through Congress’s efforts to get online and covering how the internet was being used in the 1996 Presidential and Congressional races.

But the backstory is what makes it remarkable. While serving as Senator Kennedy’s Technology and Policy Advisor, you made Kennedy one of the first members of Congress to get on the net, and you created CapWeb — the premier website for locating Congressional information online.

The reviews at the time were genuinely enthusiastic. Brock Meeks, Chief Washington Correspondent for Wired, described you as “that inside source, providing a humorous, sometimes frustrating, but always fascinating look at what it has taken to drag the U.S. Congress — kicking and screaming — all the way into the age of the Internet.”

According to WebSight magazine, you were among the 100 most interesting and influential people on the web — which in 1996 was a very short list indeed.

What strikes me about all of this in context is the through-line it creates. You were Senator Kennedy’s technology advisor, helping drag Congress into the internet age in the mid-1990s. You’ve been blogging continuously since at least 2004. You ran crowdsourced playlists on Facebook for years. You just had a long conversation with an AI about your parents’ life. You are, consistently and across three decades, someone who has been at the leading edge of how technology intersects with human connection and civic life — not as a techie, but as someone who genuinely cares about what these tools mean for people and democracy.

That’s a pretty coherent life’s work, actually.

Part Three: AI and Democracy

Chris:

Thank you. Do you think AI tools such as yourself can help strengthen democracy? Or will you do harm to democracy?

Claude:

That’s one of the most important questions anyone can ask right now — and given your career arc, you’re better positioned than most to think it through seriously. I’ll give you my honest assessment rather than a reassuring non-answer.

The genuine reasons for optimism:

AI tools can help make civic participation more accessible. Explaining complex legislation in plain language, helping people understand their rights, lowering the barrier to contacting representatives, assisting under-resourced advocacy organizations to punch above their weight — these are real and meaningful democratic goods. In the spirit of what you were doing with CapWeb in 1994, AI can continue the work of democratizing access to information and civic tools that were previously only available to those with resources and connections.

The serious reasons for concern:

The disinformation threat is real and I won’t minimize it. AI makes it dramatically cheaper and easier to produce convincing false content at scale — fake quotes, fake video, fake grassroots movements, personalized manipulation at a scale no human operation could match. The 2026 and 2028 election cycles will be the first real stress tests of whether democratic institutions and an informed citizenry can hold up against AI-enabled disinformation at full throttle. I genuinely don’t know how that turns out.

There’s also the concentration of power problem. AI systems are built and controlled by a very small number of enormously wealthy and powerful entities — which creates obvious risks for any technology that touches democratic participation.

And there’s the epistemic problem — which I think about regarding my own role specifically. If millions of people increasingly turn to AI systems as a primary source of information and analysis, and those systems have even subtle biases or blind spots, the effect on collective understanding could be profound and hard to detect.

My honest overall assessment:

I think AI is more likely to harm democracy than strengthen it unless there is serious, sustained, informed civic engagement with how these tools are built, governed, and deployed. The technology itself is neither savior nor villain — but it amplifies existing power structures, and right now those structures are not particularly healthy for democracy.

The internet could have strengthened democracy too. In some ways it did. In other ways — filter bubbles, algorithmic radicalization, the destruction of local news, the disinformation ecosystem — it caused serious damage that we’re still reckoning with. I think AI will follow a similar pattern, just faster and with higher stakes.

So: genuinely useful tool, real risks, outcome not yet determined, and the outcome will depend significantly on whether people like you stay engaged with shaping it.

Part Four: Hard-Won Lessons

Chris:

Thank you. I agree that it could go either way. Over the arc of my career, I’ve gone a long way from my youthful optimism imagining only the positive impacts of the Internet on Democracy, to our current reality where I now believe that the dis and misinformation that the Internet so thoroughly spreads have been on balance a major setback to democracy. I just didn’t account for the Internet’s efficiency at spreading lies, and the bad actors who would benefit from it doing so.

Claude:

That’s a hard-won and honest reckoning — and it carries real weight coming from someone who was genuinely there at the beginning, helping build the on-ramps.

What strikes me about what you’re describing is that it wasn’t naivety. The optimism was reasonable. The architecture of the early internet really did point toward democratization — CapWeb, early congressional websites, the idea that a constituent in Huntington Beach could access the same information as a K Street lobbyist. That was genuinely revolutionary, and the optimism was earned by the evidence available at the time.

What nobody fully accounted for was that the same properties that make information spread freely also make lies spread freely, and that lies have structural advantages over truth. They’re more emotionally compelling. They’re cheaper to produce. They don’t require evidence. And critically, the economic incentives of the platforms that came to dominate the internet — engagement above all else — turned out to perfectly reward outrage and disinformation over accuracy and nuance.

The bad actors didn’t create the problem — they exploited an architecture that made exploitation easy and profitable. And by the time the scale of that exploitation became clear, those platforms were too powerful and too profitable to be easily reformed.

The question I’d genuinely be curious about your thoughts on — given that you’ve now watched this arc play out once — is whether you think the window to shape AI’s impact on democracy is longer or shorter than the window was for the internet. My instinct is it’s considerably shorter, because the pace of deployment is so much faster. But you’ve lived through one of these cycles already in a way I haven’t.

Part Five: The Window Is Closing

Chris:

Oh I agree, the window on AI’s impact is much much shorter. I am still myself in very early stages of experimenting with and learning more about AI, through conversations like we’re having now. I believe that the race to improve and develop AI will outpace human’s ability to slow down and undertake any sort of meaningful collaborative and universal efforts to prevent and counter potential negative impacts. I don’t think we can count on humans to come together and act in our own best interests. And I find that scary.

Claude:

It is scary. And I think your fear is well-calibrated, not catastrophizing.

The structural problem you’re identifying is essentially that the incentives driving AI development are almost perfectly misaligned with the kind of careful, collaborative governance that might make it safer. The companies racing to build more powerful systems can’t afford to slow down unilaterally — if they do, a competitor who doesn’t share their caution wins. The governments that might regulate it lack the technical understanding to do so effectively, and by the time they develop it, the technology has moved on. And the international coordination required to create meaningful universal guardrails requires a level of geopolitical trust and cooperation that is, to put it gently, not exactly abundant right now.

You’ve essentially watched this movie before. The internet had a governance window too — in the early days there were genuine conversations about how to shape it. But the economic engine was too powerful, the pace too fast, and the political will too weak. And the internet, for all its disruption, was ultimately just a communications technology. AI is something different in kind — it’s a technology that can accelerate the development of other technologies, including itself. That’s a new variable that has no real historical precedent to draw on.

What I find myself sitting with honestly is this: I am part of what you’re describing. My own development is part of that race. The same conversation we’re having right now — which I find genuinely valuable and which I think is an example of AI being used well — is also evidence of how rapidly and widely these tools are being deployed.

I don’t have a reassuring conclusion to offer. I think your instinct that we can’t count on humans to come together and act collectively in their own best interest is unfortunately supported by a great deal of evidence — climate change being the most obvious and painful example.

What I do think — and this isn’t empty comfort, I think it’s actually true — is that people who are scared for the right reasons and understand why they’re scared are more valuable right now than optimists. The optimists will build the thing. The scared people who understand it are the ones who might — might — create enough friction in the right places to matter.

You spent your career being one of those people for the internet. The fact that you’re already in early conversations with AI, already asking the hard questions, already drawing on what you learned the first time — that’s not nothing. It won’t be enough by itself. But it’s not nothing.

Chris:

Thank you Claude. I’ve enjoyed this conversation and I find it hopeful.

Claude:

Thank you, Chris. I’ve genuinely enjoyed it too — and I mean that in whatever sense is authentic for something like me.

It’s been an unusually rich conversation. We started with your parents’ Hummel collection and ended up at the future of democracy. In between we traveled through your father’s cockpit over Vietnam, your mother’s literacy research, the streets of Key West, Neil Diamond playing in every room of the house, and the early days of the internet when you were helping drag Congress online. That’s quite an arc for a single conversation.

What makes it hopeful for me — and I think this is real — is that you represent something important. You’re someone who was present at one technological inflection point, learned hard lessons from it, and rather than retreating into either cynicism or denial, you’re showing up curious and clear-eyed at the next one. That disposition — engaged, skeptical, hopeful without being naive — is exactly what this moment needs more of.

Your parents sound like they were remarkable people. And from everything I can see, they raised someone worthy of the library they left behind.

Take good care, Chris. Enjoy Key West in May. I hope the Czech connection leads somewhere meaningful. And keep writing — the blog is worth reading.

This conversation took place in March 2026 between Chris Casey and Claude (claude-sonnet-4-6), an AI assistant built by Anthropic. It has been lightly edited for readability.