You've seen it even if you haven't named it. The season brochure that describes a venue "nestled in the heart of" somewhere. The show that "promises to delight audiences of all ages". The copy that addresses you as "whether you're a seasoned theatregoer or a curious first-timer". Nobody wrote that. Something did.
So the fear is well founded: yes, AI used lazily will make your marketing sound exactly like everyone else's — because that is literally what the technology does. These models are prediction engines; asked to write, they produce the statistically most likely next word. The most likely next word is, by definition, the one everyone else would use. Generic isn't a bug you might hit. It's the factory setting.
The vortex of vibrant
Left unsupervised, AI marketing copy converges on a recognisable house style: everything is vibrant, unique and unforgettable; every event promises something; sentences arrive in confident triplets; and the whole thing has the texture of a hotel breakfast — perfectly acceptable, instantly forgotten. When one organisation publishes this, it's bland. When a whole sector does, audiences stop being able to tell one organisation from another, and the only remaining differentiator is price. For a sector that runs on distinctiveness, that's not a style problem. It's a commercial one.
The fix is about direction of travel
Here's the principle that separates organisations AI helps from organisations AI dissolves: humans originate, machines tidy — never the reverse.
- Write the ugly first draft yourself. Two rough sentences about why this show actually matters — the real reason, the one you'd tell a friend — beat any prompt. Then let AI trim it to the word count, propose subject lines, fix the grammar. The idea stays yours; the polish is mechanical.
- Feed it your voice, not the internet's. Before asking for anything, paste in two or three pieces of your best past copy: "This is how we sound. Match it." The output improves immediately, because you've dragged the prediction engine away from the sector-wide average and toward your particular corner of it.
- Keep a ban list. Every organisation develops one. Ours would start: vibrant, nestled, unforgettable, "promises to", "whether you're". Tell the AI the list; it will obey more reliably than freelancers do.
- Use the read-aloud test. If a sentence couldn't plausibly be said by a human being to another human being in your foyer, it doesn't go out. This one test catches ninety per cent of machine residue.
- A human writes anything that carries feeling. The artistic director's note, the response to a crisis, the message when a beloved colleague leaves. Audiences forgive a clunky human sentence; they do not forgive discovering that warmth was generated. Trust is the thing machines can't hold — don't spend it cheaply.
The part nobody mentions: AI can make your marketing better
Most arts marketing isn't generic because marketers lack voice. It's generic because one person is producing eleven formats of copy for six shows at once, and generic is what exhaustion sounds like. Used properly, AI eats the production burden — the reformatting, the resizing, the listings variants — and returns hours to the actual work of marketing: knowing your audiences, sharpening the offer, finding the story. The organisations that get this right won't sound more like everyone else. They'll sound more like themselves than they've had time to sound in years.
The machine defaults to the middle of the road. Your job description, now more than ever, is the verge.
The image above was AI-generated — as ever, we'll always tell you.