From Messages to Conversations: AI Agents are Changing how we Find Culture

In the last six months, we’ve seen a surge in traffic at ArtsJournal. That’s great, right? But when I looked at server logs, we found that 70 percent of that surge was machines —bots— not people. We aren’t alone. According to recent reports, automated traffic hit 51 percent of global web activity in December 2025, the first time in a decade that machines outnumbered people online. AI and large language model crawlers alone quadrupled their share of web traffic in eight months, from 2.6 percent to over 10 percent. By the end of last year, there was roughly one AI bot visit for every 31 human visits to a website. A year earlier, the ratio was one in 200.

So is the web being abandoned to the machines? This isn’t a hyperbolic question. It has big implications for publishers who count on human visitors to sell to advertisers, and that’s worth a whole other story.

But there are also big implications for any artist or arts organization who uses the web to find an audience for their work. Earlier this year I wrote about a future where AI agents could act as intermediaries for our cultural consumption, scouting shows, galleries, and concerts we might want, auditioning offerings based on what they know about our preferences. I thought we had three to five years before this became the dominant mode of cultural discovery.

I was wrong about the timeline. The rise of AI as a discovery engine is happening now. And it will upend how the arts will be found as well as help shape the tastes of a great many people. The next visitor to your website is increasingly likely to be a machine, not a person. And that machine is trying to decide whether to recommend you or not.

That sounds ominous. But if arts organizations can get past the initial panic, this shift offers something that a century of arts marketing never could. Not just a new channel for the same old messages, but a potentially different relationship with an audience.

Some Background

Google search queries are already declining. The research firm Gartner predicted that traditional search engine volume would drop 25 percent by 2026 due to AI chatbots and virtual agents, and the prediction is on schedule. More than a third of consumers now begin their searches with AI tools rather than typing into Google, and for many queries, the AI answer is the only answer anyone sees. Roughly 60 percent of Google searches now end without a single click.

This is the new middleware. Not the old civic middleware of local newspapers, arts councils, and community institutions, the connective tissue between creators and communities that has been collapsing for two decades. This is algorithmic middleware that’s been replacing the old kind at a pace that makes the collapse of newspapers look leisurely. It exists between the person who might want to see your show and the information about your show. And unlike a newspaper editor or a friend’s recommendation, it doesn’t convey passion or feeling. It processes metadata.

The immediate consequences are real. More than 60 percent of event websites delete or substantially alter their content within three months of an event ending. That destruction of content is one reason mid-tier arts organizations remain invisible to AI while mega-events compound their advantage.

And we already have a cautionary tale for what happens when you stop here. Call it the Stone Age of algorithmic culture: Spotify and Netflix. Their algorithms have become more conservative over time, not more adventurous, recycling what you already know rather than introducing what you don’t. Spotify’s retention metrics now count for three times more than play volume in the algorithm. A “30-second hook” culture has emerged. If a song doesn’t grab you immediately, the algorithm moves on. The familiar wins while the unknown gets buried.

Nobody set out to make culture blander. But when an algorithm can only optimize for what it can measure, and what it can measure is shallow — clicks, saves, repeat listens — the flattening happens almost in spite of everyone involved, for artists, listeners, even the platforms themselves. And it’s getting worse. In October, Spotify integrated directly into ChatGPT across 145 countries. Can you say stranglehold?

This is the primitive version of machine-mediated culture. The algorithm doesn’t know what a revelatory night at the theater feels like, it knows what people clicked on last time. If we stop here, we deserve what we get.

But here the argument gets more interesting, I think. The crude algorithm worked for music and movies because the friction between recommendation and consumption is essentially zero. Spotify says “try this,” you tap and you’re listening. The transaction cost of a miss is maybe thirty seconds. Netflix suggests a show, you click, and if it’s bad you bail after ten minutes. The algorithm doesn’t need to be sophisticated because it doesn’t need to persuade anyone of anything. It just needs to reduce the sea of options to a manageable list, and the near-zero cost of trying does the rest.

The analog arts are the ultimate unfamiliar experience because they’re a different experience every night. While Spotify optimizes for stasis (keeping you in a mood), an AI agent can optimize for transformation—which is a core value proposition of the performing arts.

We’re asking someone to make a significant commitment of tickets, time and traffic hassle to get there. The gap between discovery and decision is enormous. A crude recommendation — “people who liked X also liked Y” — doesn’t do it. If a bot is the “first audience,” the website needs to stop being a brochure and start being a knowledge base. You need something that can answer questions, address hesitations, and meet a specific person’s curiosity. You need, in other words, not a better algorithm but a conversation.

From Messages to Persuasion

And that’s the opportunity, I think. The instinct will be to treat the Rise-of-the-AI Bots as the next iteration of SEO, adapting to the new rules, optimizing the metadata, and tagging everything properly. And yes, these are the table stakes. If your website can’t be crawled and categorized, you’re invisible.

But stopping there misses the opportunity and the revolution.

For a century, arts marketing has been a broadcasting operation. You craft a message — a listing, an ad, a social media post — and fling it as widely as possible. It’s a message in a bottle thrown into a Sea of Messages, hoping some tiny percentage of bottles wash up on the right beaches. The entire model assumes that the job is getting the message out. Reach equals results. And that model, like the Spotify algorithm, works well enough when the cost of trying is low. But it has always been a poor fit for the performing arts, where every ticket is an act of faith.

The shift AI agents represent isn’t a new channel for the same old messages. It’s a shift from messaging to persuasion — from broadcasting to dialogue. A listing says: “Romeo and Juliet. Friday at 8. $55, starring so-and-so.” It conveys information but it’s useful only to people for whom that information already means something. An AI agent mediating between your organization and a potential audience member can do something a listing never could: find out what that person actually cares about and make the case accordingly. The hesitant parent gets a different conversation than the Shakespeare scholar. The classical music novice who finds Mahler overwhelming gets a different entry point than the subscriber who has heard thirty recordings of the Fifth.

This isn’t metadata optimization. This is the difference between a billboard and a conversation. And the performing arts, the art form with the highest transaction cost and the hardest discovery problem, may have the most to gain from it.

Art as a Process, a Community, Not Product

Perhaps the opportunity goes deeper still, because it has to do with how we think about what art is.

We used to treat the performance as the discrete thing, the artistic act. But the internet and social media have been rewriting relationships between artist and audience for two decades. The artistic “experience” now extends to either side of the performance — the anticipation, the creation, the reaction, the culture that forms around the work. We know that audiences are fascinated by process. They research endlessly. They have an enormous appetite for going deeper into what they love. The community that gets built around the art is also part of the art.

The problem has always been scale. A marketing department can write a listing, send an email blast, post a social media clip, but it can’t have ten thousand different conversations with ten thousand different people about what makes this production worth their time.

AI agents can. I don’t mean that agents substitute for people. Who, really, wants to converse with a bot if they don’t have to? But they do want their specific, niche questions answered instantly without digging through a FAQ page. An AI agent can show you or point you in the direction of what you’re interested in. Show you the video, respond with a relevant history or storyline.

This is where the explosion of personal assistant bots could be interesting. Not as a threat that flattens culture into tags and metadata, but as the first technology that could let an arts organization engage deeply and individually with anyone who’s curious at scale. As I argued in my earlier post, what organizations need isn’t a better-tagged web page. It’s something more like a Digital Twin of their productions, an AI presence loaded with the DNA of the work that can meet you where your curiosity lives. Not a chatbot that tells you where to park. An intelligence that can make the case for why this production, this night, is worth your time and your fifty dollars and your babysitter and make it differently for every person who asks.

Those artists and organizations that can engage with what their audiences actually care about will build deeper relationships with the people who find them. And the people who find them will arrive already invested, already part of the community around the work, already “better” audience members.

The first audience for your art is becoming a machine. The question isn’t just how to optimize for that machine, it’s what you give it to say, and whether what it says is worth a conversation.