An AI “Digital Twin” for the Performing Arts

A few weeks ago, I wrote about “speaking to the art” in museums—using AI to turn passive observation of artwork into an active, contextual dialogue, a way for visitors to find what resonates with them and explore more meaningfully. It is an idea about deepening the experience once the visitor is already in the room.

The performing arts face a bigger challenge, before anyone enters the concert hall or theatre: discovery.

In a museum, the cost of engagement is low. If a painting doesn’t move you, you walk five feet to the next one. In the performing arts, the transaction cost of a “miss” is huge. We ask audiences to commit fifty dollars, a babysitter, parking, and three hours of their Friday night to a specific, unchangeable event.

And what’s the sales pitch? A “listing.”

The standard arts listing is a relic of the print era, one that we just pasted onto the web and failed to update. It consists of a title, a generic publicity photo, a list of names, and a 200-word paragraph of marketing copy that tells you everything about the plot and little about the experience that you’ll have. Some organizations add video and maybe an interview with the director or links to reviews. But these are static and don’t necessarily speak to the visitor’s interests or experience.

We are asking consumers who are accustomed to algorithmic personalization and doing deep-dive research before buying a toaster to go on an expensive blind date with our art. And this is our pitch? This is a systemic failure of imagination. We are broadcasting information when we should be inviting interrogation.

In the evolving world of AI, marketing is moving from getting messages out to engaging in dialog with the consumer. Messages get lost in the Sea of Messages. Persuasion asks what you’re interested in first and engages you in opportunities. Listings, whether in the newspaper, on your social media feed or on your website, are incapable of anything more than conveying information, and they’re only of use, really, to people for whom that information already will mean something.

The Digital Twin

Imagine if, instead of a static web page, every production had a “Digital Twin.”

This isn’t a customer service chatbot designed to tell you where to park or how to donate. It is a specialized Large Language Model trained on the specific DNA of that show or production. It is a model loaded with the script, scores, director’s concept notes, costume sketches, video rehearsal, historical context, archive recordings, and producer interviews and framing of the work.

The potential patron lands on the Digital Twin and instead of reading a blurb, opens a dialogue.

A hesitant parent could ask, “I’m thinking of bringing my teenager to Romeo and Juliet. Is it stuffy?” And instead of a canned marketing response, the Twin—accessing the set designs and director’s notes—might say: “Not this production. We’ve set it in a near-future dystopia to highlight the themes of tribalism. The language is original Shakespeare, but the aesthetic is Cyberpunk. Here is a concept drawing of the set.”

A classical music novice might begin with, “I usually find Mahler overwhelming.” To which the model could reply, “That’s valid. But [Name] has a faster, more transparent approach to the Fifth. Listen to this 30-second audio clip from rehearsal; she’s emphasizing the woodwinds to lighten the texture.”

Or, if you have valuable archives that allow for comparisons or examples the AI can draw on, a curious visitor might get a master class in how this conductor versus that conductor shapes a performance. The point is to follow the visitor’s interests and passions to see where they might engage.

This is no longer a listing; it’s an audition. The potential audience member is auditioning the work to see if they want to invest–their money, their time, their attention.

Art as a Process

The deeper argument for a Digital Twin is not about technology but about how we think about art. Is it simply a product on a stage or is it a process that sets out to explore ideas and create something amazing out of those explorations?

Traditionally, the prevailing wisdom has been to hide the work in progress. We treat the rehearsal room as a black box, worrying if we show the “messy” parts—the sketches, the unfinished sets, the corrections—we will ruin the magic of opening night. Artists didn’t want to be seen as anything other than at their best.

But in the modern creator economy, the process is the magic. Audiences today are obsessed with the how. By keeping the artistic process hidden behind a curtain until the curtain literally goes up, we are withholding the very thing that builds connection. And in an environment of constant attention, where “perfect” performances are a click away and synthetic content increasingly overwhelms, it is the organic provenance of live art that is rare and will increasingly resonate.

A Digital Twin allows us to safely give access to the process. It moves marketing from shouting “BUY NOW” to saying “LOOK HOW.” When a user spends ten minutes asking the model about the costume design, or comparing approaches to Mahler, they aren’t just shopping, they’re engaging. You are creating a micro-relationship with the institution before they’ve spent a dime.

The Better Audience

There is another benefit: it creates a better audience member.

We worry about spoilers, but in most art forms context creates depth. A “cold” audience member—one who bought a ticket based on a generic blurb—spends the first twenty minutes of a performance just trying to figure out the rules of the world they have entered into.

A “warm” audience member—one who has already engaged—arrives already introduced to the language of what they’re going to see or hear. If the model told them, “Listen for the oboe in the second movement; it represents xxx,” they are not spoiled; they are primed. They become active listeners. They feel they’re in the know.

When I was a music critic, I used to study audiences. Even within the realm of classical music, it was fascinating to me how different audiences took on different personalities. Most striking was how some audiences, after some performances, immediately started talking about what they had just heard, while at many more, audiences filed out quietly. I’ve always felt that the measure of success of a performance wasn’t that everybody liked it, but did they have a reaction to it, good or bad, and how strong was that reaction? A better audience member has engaged with the music and has opinions. A worse audience doesn’t know how to engage in the first place.

Then there’s what we can learn

People lie. They do. Ask them in a survey or focus group and they’ll likely give you a projection of the person they like to see themselves as. Their actual behavior is often very different. With a Digital Twin, we can see the conversations, find out what people are asking, get insight into what interests them or not. Well before the performance we can start to see where the ideas resonate, what pricks the curiosity. Again, it’s art as a process. Yes there are issues of privacy, but by making the interactions transparent, these can be addressed.

Future-Proofing

There’s another reason to think about doing this now.

We are rapidly approaching a future where AI agents will act as intermediaries for our cultural consumption. In three to five years, you probably won’t browse the web for tickets at all. You will simply say to your phone, “Find me something weird and visually stunning to do on Friday night, under $60.”

Your personal AI agent will go out and “interview” the APIs of arts and entertainment organizations for everything available. Based on what the AI knows about your tastes and preferences, it will audition the offerings and present choices. This is your Spotify algorithm on steroids.

If your organization only has a static HTML listing with a JPEG and a paragraph of text, your work will be invisible to that agent. Your traditional page has no “hooks” for the machine to grab. But if you have a Digital Twin—a model of your production’s intent, style, and content—your art becomes discoverable and can negotiate on your behalf. The Digital Twin addresses today’s problem of discovery, but we are simultaneously building infrastructure for a machine-mediated future.

Why most arts organizations won’t do this

The technology to do this is actually pretty simple. Building specialized LLMs or RAGs is now routine. And cheap. There is plenty of money out there waiting to fund experiments in AI applications. But I suspect the conceptual shift will be hard for most arts organizations to make. A shift in control. A shift in relationships with audiences, with artists, and with the art itself. It isn’t about just picking up a new technology or tool and putting it to work to do something cool; it requires a mindset that challenges what art can be.