INTERVIEW SERIES: Media, Culture, and Artificial Intelligence INTERVIEWER: Gökhan Çolak INTERVIEWEE: Matthew Salzano

AI, Society & Human Behavior

Your research explores how people shape technology and how technology shapes people. At what point does technology stop being a tool and begin influencing human behavior itself?

I’m not sure I accept the premise here: that at some point technology is simply a neutral tool that somehow doesn’t influence human behavior! Not to be the cliché media scholar, but to quote McLuhan—the medium is the message. Our media, especially our dominant media of a time (like the internet today), change our cultural priorities, norms, and content in a profound way. McLuhan’s example was the electric light changing our experience of time by making “night” less relevant; our example may be smartphones altering our memory. That’s not to say that how people use technologies doesn’t (re)shape particular media. It’s also not to say that media technologies emerge from beyond culture to influence it—media technologies are also products of culture. So, the influence of technology is malleable but inescapable. There’s no moment where any media—radio, TV, Facebook, Generative AI (GenAI)—was simply a tool that a human used without being influenced.

We’re increasingly surrounded by machine-generated content. Are we entering an era where human communication becomes overwhelmed by artificial communication?

Not long after ChatGPT was publicly released, Matthew Kirschenbaum forewarned what he termed the textpocalypse in a great essay in The Atlantic. His concern was that, as GenAI models begin interacting with each other, the web would be overwhelmed with synthetic (that is, not human-written) text. If that synthetic text continues to be challenging to identify and exclude, then text would cease to be a trusted medium for communication.

I’ve spent a lot of time with students in the 3 years since this was published discussing whether we have “reached” textpocalypse. In general, the consensus my students and I share is: no, but we keep getting closer. At the moment, so much AI generated content is still either (1) clockable, i.e., you know it when you see it, or (2) human-initiated such that you still can’t quite call it completely “artificial communication,” even if you don’t like the human-machine collaboration.

That said, textpocalypse and my strict definition of “artificial communication” may miss the point. I illustrated this to students in my “AI and Creativity” course with the example of Fruit Love Island on TikTok. This AI-generated show imitated the dating show franchise “Love Island,” but featured fruit-people instead of humans. It’s racked up millions of views, and it was just made by one person. The actual show Love Island takes more than 400 people to produce.

Fruit Love Island isn’t fully artificial (it wasn’t autonomously created by bots) but it feels like a harbinger of what could be to come. Instead of employing a crew of skilled humans who are working hopefully fulfilling creative industry jobs (often union jobs, too!), creative products could be much more-machine led, still popular, and cheap in comparison. Every day on your local television you can see results of this with AI-generated advertising that previously required human crews, actors, editors, writers, and so on. So, this is a different kind of overwhelm—maybe not in the content of human communication but certainly in its material conditions.

Many digital platforms promise participation and connection, yet people often feel exhausted and emotionally disconnected online. Why do you think that contradiction exists?

I hope you’ll read my book, Just Participating: Reformatting Critical Sensibilities When There’s Just Too Much, when it’s published by NYU Press (hopefully in 2028) to learn more about my perspective on this! In the book, I argue that our digital platforms and political culture have created the conditions where participation is an unquestionably valorized end of itself.

That demand to participate has always existed, especially in social justice circles, but digital media intensified that demand by constantly nudging us to participate and presenting us with endless seemingly- urgent reasons to participate. I call this constant demand to participate the participatory imperative, and it feels like it’s “just too much,” as I put in the title—we can’t participate in anything and everything all of the time. To add to that exhaustion, it feels like nothing changes: the book is mostly focused on cases of social justice advocacy during the first Trump presidency, and here we are in the second. We want to make change, but our existing models for participation are exhausting and ineffective in the face of our digital and political conditions. So, in the book I try to offer a new way to approach participation conceptually and practically to help reckon with the participatory imperative.

Critical AI & Ethics

The term “Critical AI” is becoming more visible in academic and cultural discussions. What does Critical AI actually mean to you?

When I say “Critical AI” I mean Critical as in Critical Theory, as Jennifer Sano- Franchini has articulated. Critical theorists go beyond ethical or responsible AI (which, in my view, is often more focused on specific uses or principles—which can be useful or useless) and focus on power, values, and action in relationship to AI. Here, I mean “AI” in a big sense: not only the technical qualities of machine learning, natural language processing, large language models, etc., and their specific interfaces and instances, but also the social, cultural, and rhetorical life of “AI” as a concept. Critical AI asks sociotechnical questions like: who benefits from AI and who suffers? Whose agency is reduced and whose is uplifted by AI, whether its specific instances or its grand narratives? What political philosophies and hierarchies of values are given priority in the way we talk about and interact with AI? How do we intervene and take action in/with AI to create a more socially just world, one where people can truly flourish?

AI systems are often presented as neutral technologies, but are algorithms ever truly neutral?

Absolutely not! I would highly recommend three books on this topic: Weapons of Math Destruction by Cathy O’Neil, Algorithms of Oppression by Safiya Noble, and If…Then: Algorithmic Power and Politics by Taina Bucher. O’Neil explains the basics of data science so clearly, and across a diverse set of examples shows how “data” gets used to neutralize decidedly not-neutral processes that are weaponized against people, especially those already marginalized. Whereas O’Neil takes a broad scope, Noble importantly digs into specific problems of digital oppression: she explains further how these biased algorithms specifically oppress women of color in search engines like Google. Finally, Bucher’s focus is on how algorithms reformat social life in profound ways that reshape human communication (in ways that are beneficials for the profit-motives of creators of algorithms).

Who gets protected — and who gets left behind — in the rapid development of AI systems?

One of my favorite moments of the last year is when graduating college students booed and heckled speakers who attempted to speak positively about AI. One specific clip stands out to me: when the real estate development executive Gloria Caulfield told University of Central Florida students “the rise of artificial intelligence is the next Industrial Revolution.” Her comparison was somehow meant to be uplifting, adding (to more boos) “AI capabilities are in the palm of our hands.” Her comparison, I fear, is apt but not encouraging. For the AI capabilities are “in the palm of our hands” in the same way the factory line brought us mass produced products—convenient, but at what cost? It was mostly to the benefit of the factory owners, to some benefit for those with wealth to afford the outputs of the factory line made more cheaply and readily available, but at great cost to the classes of people who were now suffering: alienated from their crafts, left to poverty, or working a painful and soulless factory line job.

Media, Culture & Power

Social media platforms and algorithms increasingly shape public discourse. Do you think digital platforms are becoming a new form of invisible cultural power?

Digital platforms have now long been a form of invisible cultural power. As a communication scholar, my proximity to communication practitioners has always made this clear to me in a way that may be less true for folks who haven’t seen “behind the scenes” of media. Take, for example, “Search engine optimization” (SEO), the practice of improving a webpage’s ranking in search results. This has been taught to communication and marketing students for ~20 years for early adopters, and certainly widespread in the last 10 years. SEO recognizes that search engines shape culture by ranking some results higher than others, and attempts to manipulate that shaping to be more favorable for a given webpage (itself an artifact of culture). And it’s not just about making sure your webpage looks good—you also need to make sure your webpage is getting linked to and referenced in order to improve your rankings, or you may want to uplift something new to down-rank something incriminating. Look no further than how celebrities allegedly stage public stunts or rename products specifically to bury negative search results! The need to modify rankings for a client on Google may have meant that you or I consumed cultural artifacts about that client, unbeknownst to us that it was some small part of a larger strategy to improve search results.

Now, SEO is getting more complicated—people use GenAI like ChatGPT, Claude, or Gemini instead of Google Search, and Google has “AI overviews” people may read/click instead of looking at search results. In the public relations and marketing industries, folks are already talking about “Generative Engine Optimization.” Muck Rack, for example, has shared their research about what models are most likely to cite in their responses. It’s totally different than what helps with SEO — a new sort of invisible power — and so communication practitioners have to shift their work accordingly.

I’ve chosen to focus my answer on search intentionally, instead of the more common social media examples, because I think that it’s become pretty well-known and accepted that creators on social media are influenced by algorithmic incentives. I think it’s important to recognize that even before Facebook was widely used, digital platforms like Google were shaping our culture, regardless if we used Google to find something (and thus interacted with its PageRank algorithm) or if we consumed cultural artifacts meant to improve someone’s search result ranking.

In a world dominated by constant notifications, information overload, and endless engagement, what happens to human attention and critical thinking?

This is a question that scholars are fiercely divided upon, especially regarding GenAI. I’m of two conflicting minds about it. On the one hand, I think our attention is obviously reshaped and redirected in ways that are not always positive. For example, we have all sorts of reports that young people don’t/can’t read books, and I’m simply not willing to view that as a neutral cultural change. In particular, the novel is a technology of mass culture I’m not willing to lose, as it gives people the chance to empathize in a way that is distinct from other mass media and, as philosopher Richard Rorty might say, is of specific importance for democracy. And, I know from my own lived experience that it takes a lot of effort to not unquestioningly trust information I receive on social media or from a chatbot and instead go out of my way to check the information—I’m inclined, sometimes by the design of digital media, to bypass my critical thinking rather than activate it in those moments.

On the other hand, I think that digital media has lead to a flourishing of attention and critical thinking even if misdirected. Spending hours on TikTok may not be as engaged as spending hours reading a book, but it does mean one could be engaging with a global community of diverse perspectives! But the algorithm doesn’t always provide content that leads to meaningful interaction—and in fact we know that it sometimes provides the polarizing things that undermine democratic participation rather than support it. That said, it sometimes does provide us with content that stimulates our critical capacities. I think about how much analysis I encountered on social media of my two favorite media events of the last year or so: Heated Rivalry and Love Island USA. The people long for literary analysis, as I saw in many comments across the long videos analyzing dialogue, characterization, shot composition, narrative, and so on. This sort of communal, interpretive work is still desired even if it’s different than in other media eras. But it also has a dark side: the proliferation of conspiracy theories online is “critical thinking” gone wrong, an over-application of the desire to interpret in community.

So, these new technologies are changing our attention and critical thinking, and we’re still figuring out how to catch up. In his book The Gutenberg Parenthesis: The Age of Print and Its Lessons for the Age of the Internet, my friend and colleague Jeff Jarvis puts this so clearly: It took 150 years from the invention of print to the products we now take for granted as print media, like the novel, essay, or newspaper. We are only 30 years into the web-browser-enabled internet, which, as Jarvis writes: “In Gutenberg’s timeline, that puts us at roughly the year 1480.” We are in the infant stages of this media era, if it is truly as transformative as the likes of print, radio, and television. Accordingly, I am wary of taking too strong of a perspective on what will happen to our attention and critical thinking. I see GenAI as a continuation of the internet’s media era (as cable TV was to television), a new technology made possible by a previous one (the internet) that is both continuous and disruptive. We must keep experimenting, deliberating, and deciding on what works and what doesn’t—and see what happens.

The Future of Artificial Intelligence and Communication Technologies

Do you believe AI will ultimately democratize knowledge and communication — or concentrate power even further?

If the story of the internet so far is any indication, I think it will concentrate power even further. But this isn’t about the technology itself—it’s not about LLMs, or NLP, or ML—it’s about the political economic conditions of the technology. Let’s start with the internet as a historic parallel (that I’m simplifying greatly): it’s originally a hobbyist web with virtually no gatekeeping. If you can host your website and write HTML, you can be included. People found communities of websites and blogs.

There was so much hope from leading thinkers that this was a democratic medium, a revolutionary one. And early social media encouraged this, as it was at the center of social movements like the Arab Spring and Occupy Wall Street. However, we’ve seen the internet (especially social media) become totally corporate and platformed; we’ve seen the mass manipulation it has made possible; and we’ve seen who has economically benefited. It’s concentrated power in the hands of a few Silicon Valley elites, not democratically expanded access to communication. I think we’re on a similar track with AI, as a few (Google, Microsoft/OpenAI, Anthropic, Meta) power brokers control so much of what we consider “AI.”

At the same time, it wouldn’t be fair to say the internet hasn’t democratized knowledge in some sense. We look to Wikipedia for basic information (like to confirm when SEO started!), we learn how to change a tire on YouTube, and we get excellent recommendations from Reddit. So I’m not totally opposed to the idea that AI could have some role in democratizing knowledge, even as I remain extremely skeptical and concerned at the moment about its role in estranging humans from each other and undermining expertise.

When future generations look back at this moment in technological history, what do you think they will say we misunderstood most about AI and digital culture?

What do I think people in the 2000s misunderstood about the internet? I think it would be cruel to say to people that they were mistaken for thinking it was a democratic technology, because they genuinely felt it was at the time—and frankly, I think it was at least more democratic at the time. Similarly, I think future generations have to forgive us for whatever naïve things we believe about AI and digital culture, because we’re simply working with what we’ve got.