Read the Beforeitsnews.com story here. Advertise at Before It's News here.
Profile image
By My Heart\\\'s in Accra
Contributor profile | More stories
Story Views
Now:
Last hour:
Last 24 hours:
Total:

Upholding Democracy in the Age of AI: remarks from UMass

% of readers think this story is Fact. Add your two cents.


I gave a talk a few weeks back at a UMass event titled “How ‘We the People’ Can Uphold Democracy in the Age of Social Media and AI” – you can see my remarks and those of my copanelists – Manisha Sinha of UConn and Jonathan Nagler of NYU – on YouTube.

My basic argument: it wasn’t unreasonable to hope that social media would be, on balance, good for democracy. It might be unreasonable to believe the same thing about generative AI. I worked part of this talk into my monthly column for Prospect – AI’s Railroad to Corruption – but I think the whole talk stands up well on its own.

scholars onstage at UMass

Hi. My name is Ethan and I’m a recovering cyberoptimist. “Cyberoptimist” – or the stronger version “cyberutopian” – was a label no one ever applied to themselves, but got applied to lots of thinkers who were optimistic about the internet’s affects on democracy in the early 2000s. A lot of what we hoped for hasn’t come to pass, but I want to make the argument that it wasn’t insane to think that new media technologies could be good for democracy.

I’m a scholar of the public sphere. That’s the conceptual space in which people learn what’s going on in the wider world, argue about how the world should work and mobilize to make changes in the world. The public sphere precedes democracy – there was a physical public sphere in Athens: the agora – but since colonial America, the public sphere has been mediated, first through newspapers and postal mail, then through radio, television and now the internet. Every time the public sphere changes its dominant medium, the dynamics of the public sphere change. New groups of people get good at working within the new medium and become empowered, while other groups lose their influence.

In the early 2000s, the public sphere was still dominated by broadcast media. Our three nightly news broadcasts broadened to include cable news and talk radio. But it was a moment where very few people had access to the mic, and those people tended to have something in common.

The rise of the digital public sphere suggested something else was possible: we could hear directly from individuals, without the “gatekeepers” of mainstream media. People could share ideas and experiences on blogs, and then later on social media. The hope the cyberoptimists shared is that this might broaden who we listened to, whose opinions mattered, what problems we took seriously and what tools might be used to change the world.

I was not a disinterested observer regarding this question. I helped found a nonprofit organization – Global Voices – that has spent the last 22 years translating and amplifying voices from the Global South. When the Arab Spring broke out in Tunisia in December 2010, we were one of the few English language media outlets to cover it. This became awkward when we learned that our correspondents in Tunisia were actively overthrowing the government. One of our authors went from writing everyday, to writing via notes passed from his prison cell, to not writing because he’d been appointed the new minister of youth and sport.

Movements like the Arab Spring unfolded on social media in societies where there wasn’t sufficient space for democracy dissent and action in the public sphere. In more open societies, the movements that emerged in online spaces – Black Lives Matter, #metoo – were ones where the relevant issues weren’t often addressed (police violence, sexual harassment and abuse) in the dominant public sphere.

Somewhere in the 2010s, optimism about these new spaces for expression took a back seat to legitimate fears. These new tools were at least as good at spreading mis and disinformation as they were authentic experiences. By being open to everyone, they were open to bots and government propagandists. It became clear that the companies operating these private platforms that we tried to treat as public spheres had their own agendas, and they were less interested in democratic deliberation than they were in capturing and monetizing our attention. More recently some of those platforms – X, notably – have taken an explicit political position and no longer pretend that their goal is to be a public square for everyone.

If you accept my premise that every time the medium of the public sphere changes, so do the dynamics of the space, this is a really good moment to pay attention to the public sphere and democracy. We are pouring money into AI at a pace that is giving economists pause: Stijn Van Nieuwerburgh of Columbia Business School estimates that we’re on track to invest 3.6% of US GDP in AI over the next five years. The only way to understand the magnitude of that investment is to compare it to transformational investments in infrastructures like canals, railroads, highways, electrification and the internet. All those investments were significantly smaller than the bet we are currently making on AI. Keep that in mind, because we’re going to come back to that question of investment.

Whenever I talk about AI, I like to set some ground rules about what I will and won’t be talking about. I’m not going to talk about existential risk, the idea that these systems might somehow kill us all. I’m not going to talk about impact on the environment or the capitalist fantasy of replacing worker power with machines the owners can shut off. I’m going to try and limit myself to AI and the public sphere, because that’s vast enough as it is.

So let’s start with the original sin of generative AI, the “largest theft of labor in human history”. That’s not my characterization, by the way – that’s Dr. Brent Hecht, Microsoft’s director of applied science, in testimony from the New York Times’s suit against Microsoft and Open AI. (To Hecht’s great credit, he was making the argument internally that Microsoft needed to consider the damage AI was doing to creators.) Lots of the data stolen to train contemporary LLMs came from journalism organizations that spent billions learning what happened in the world and sharing it with their audiences so they can make informed decisions about governance. (By the way, Global Voices – a small nonprofit organization – is one of the journalism orgs who’s been ripped off.)

Products like Google’s AI summaries cause traffic to sites like the New York Times to drop 93% compared to traffic from search engines. Sites like the Times may be able to replace advertising revenue with subscriptions, but most smaller news sites can’t. We are already entering a public sphere where people who don’t have the money or the inclination to subscribe to a good news source have less information than they did a few years ago.

Will AIs fill this gap? Maybe. But reporting is a very different problem than solving the Navier-Stokes equations. AIs excel at solving questions that have provable or testable answers, like executable computer code or math theorems. But they exhibit “jagged intelligence” – they can be stunningly good at some tasks and surprisingly bad at others.

I can imagine using a series of AI agents to record a public meeting, transcribe the discussion, research the participants and the issues raised and produce a competent article. What I _can’t_ imagine is doing this in a way that is as cost-effective as sending a reporter… and even that is generally too expensive these days. The problem for decades has been that it costs more to report the news than you can make from selling it. To be clear, the AI builders aren’t confident they can do this either – some describe the undermining of journalism as “a doom loop”, in which they cannibalize the information they need to keep training their models.

I invite you to imagine a future where the collapse of informed democracy is
a byproduct of scarcity of training data. As AI labs get aggressive about ingesting journalism to train their models, news disappears behind paywalls and fewer readers encounter local news. Publications that are already threatened go under; those that remain license their content to AIs, who cannot give answers about as broad a range of topics because the original reporting to support them no longer exists.

What about the cyberoptimist arguments for the increase of diversity of voices? Social media let billions of people reach potentially broad audiences. With AI-based tools, they can produce better and more persuasive stories, right?

Maybe. But what’s already happening is that propagandists are having a field day using AI to create countless stories, translated across languages, to advance their narratives. These stories are barely comprehensible to human audiences, but that’s not the point – their goal is to influence AI answers. And it works – a recent study found that popular American AIs routinely respond with long stretches of Chinese government propaganda when asked questions about issues the Chinese government most cares about.

I can imagine futures in which the current trajectory of AI produces results with significant social benefit. What I have a harder time imagining are _democratic_ futures brought about by AI. Right now, the most capable AIs are being built by a small number of incredibly wealthy companies. Their priorities – artificial general intelligence, the elimination of jobs and the power of labor to organize – are determining what AIs do. Democratic values like inclusion aren’t part of the conversation, and right now, the ways we know how to build powerful AIs favors oligopoly over participatory democracy.

Early in the social media era, some astute critics – notably Evgeny Morozov – pointed out that these tools could be as useful for dictators as for democrats. What the critics didn’t do as well is propose worthwhile alternatives – we’re finally getting some momentum for treating social media as public infrastructure, for user-controllable recommendation algorithms and for distributed networks.

The critics have arrived much earlier and much more loudly for AI. But it’s a hard ship to turn. There’s smart work being done on very small AI models and cooperation with communities to build AIs that reflect the values of the communities who are training them. But these efforts are tiny compared to the financial juggernaut powering AI.

Let’s return to that estimate that companies are investing 3.6% of the US GDP into AI over the next five years. That’s a rate of investment higher than what the US spent on railroads between 1870 and 1890, a time of tremendous boom, bust and accumulation of wealth.

Historians now call that period of investment, growth and concentration of wealth The Gilded Age. It was a moment of high inequality, high corruption and dysfunctional democracy. It wasn’t until decades later that the reforms of the progressive era increased democratic participation through the secret ballot, ballot initiatives and muckraking journalism. Concentration of economic power tends to concentrate political power, and if AI proves to be the transformative economic force the railroads were, it’s easy to imagine an alignment of industrial and political power operating in a way that’s deeply non-democratic.

If there’s an optimistic outcome here, I don’t think it’s going to come from the US. In some of what Mark Carney has called “the middle powers”, there is a move towards sovereign AI – achieving independence from the American and Chinese AI giants by building models that are locally hosted and operated. I want to suggest that that’s a weak form of sovereignty. A stronger form would be one where these sovereign models are designed and trained on data that reflects a nation’s culture and values, rather than inheriting the US-centric biases of today’s open and closed models. And an even more exciting challenge would be creating sovereign, democratic AIs, where there’s a national conversation about what AIs should and should not do, how we build, monitor, audit and regulate them, and what bright lines we might demand as a society that AIs do not cross.

While President Trump has promised us that regulating AI requires nothing more than “A STRONG AND SMART (HIGH IQ) PRESIDENT”, democratic AI requires a lot more than that, and it would be a good time to start a conversation about preventing the next Gilded Age.

The post Upholding Democracy in the Age of AI: remarks from UMass appeared first on Ethan Zuckerman.


Source: https://ethanzuckerman.com/2026/11/10/upholding-democracy-in-the-age-of-ai-remarks-from-umass/


Before It’s News® is a community of individuals who report on what’s going on around them, from all around the world.

Anyone can join.
Anyone can contribute.
Anyone can become informed about their world.

"United We Stand" Click Here To Create Your Personal Citizen Journalist Account Today, Be Sure To Invite Your Friends.

Before It’s News® is a community of individuals who report on what’s going on around them, from all around the world. Anyone can join. Anyone can contribute. Anyone can become informed about their world. "United We Stand" Click Here To Create Your Personal Citizen Journalist Account Today, Be Sure To Invite Your Friends.


LION'S MANE PRODUCT


Try Our Lion’s Mane WHOLE MIND Nootropic Blend 60 Capsules


Mushrooms are having a moment. One fabulous fungus in particular, lion’s mane, may help improve memory, depression and anxiety symptoms. They are also an excellent source of nutrients that show promise as a therapy for dementia, and other neurodegenerative diseases. If you’re living with anxiety or depression, you may be curious about all the therapy options out there — including the natural ones.Our Lion’s Mane WHOLE MIND Nootropic Blend has been formulated to utilize the potency of Lion’s mane but also include the benefits of four other Highly Beneficial Mushrooms. Synergistically, they work together to Build your health through improving cognitive function and immunity regardless of your age. Our Nootropic not only improves your Cognitive Function and Activates your Immune System, but it benefits growth of Essential Gut Flora, further enhancing your Vitality.



Our Formula includes: Lion’s Mane Mushrooms which Increase Brain Power through nerve growth, lessen anxiety, reduce depression, and improve concentration. Its an excellent adaptogen, promotes sleep and improves immunity. Shiitake Mushrooms which Fight cancer cells and infectious disease, boost the immune system, promotes brain function, and serves as a source of B vitamins. Maitake Mushrooms which regulate blood sugar levels of diabetics, reduce hypertension and boosts the immune system. Reishi Mushrooms which Fight inflammation, liver disease, fatigue, tumor growth and cancer. They Improve skin disorders and soothes digestive problems, stomach ulcers and leaky gut syndrome. Chaga Mushrooms which have anti-aging effects, boost immune function, improve stamina and athletic performance, even act as a natural aphrodisiac, fighting diabetes and improving liver function. Try Our Lion’s Mane WHOLE MIND Nootropic Blend 60 Capsules Today. Be 100% Satisfied or Receive a Full Money Back Guarantee. Order Yours Today by Following This Link.


Report abuse

Comments

Your Comments
Question   Razz  Sad   Evil  Exclaim  Smile  Redface  Biggrin  Surprised  Eek   Confused   Cool  LOL   Mad   Twisted  Rolleyes   Wink  Idea  Arrow  Neutral  Cry   Mr. Green

MOST RECENT
Load more ...

SignUp

Login