"A Door Slams. Happily.": AI-Generated Creative Gone Wrong — and How to Avoid It

Put yourself in the shoes of a small business owner. You and your team do a lot of things well, but marketing has never really been a priority.

Then along comes AI. And making your own marketing, which used to seem out of reach and scope, suddenly seems within your grasp. You ask AI and make a radio or TV commercial based on your website and sales materials, and it does. 

The output is formatted. The paragraphs are structured. There are sound effects and visuals indicated and a certain convincing confidence to it all. And if you don't know what ‘good’ looks like (in a script, in a brief, in a creative strategy), you might just think you've got something.

Trouble is, you don’t. 

But you won't know that until someone who does know what ‘good’ looks like sees it. And by then, you've already submitted it.

AI is one of the most powerful tools available to marketers and communicators right now. That's not hyperbole — it's useful, fast, and genuinely capable of making smart people smarter. 

But that last part is the key qualifier. And when that expertise isn't in the room, what gets produced isn't marketing. It's the convincing shape of marketing. And that's arguably worse than nothing.

Wait... You’re not a production house. Why do you care?

True, Show and Tell does very little in-house video and audio production. We rely on great partners to do that, because they do it best (and our clients deserve the best). Sometimes those partners are private studios. Other times, they’re the broadcasters themselves.

We’re sticking our nose into this conversation because we value those partnerships. They’re our peers, and we’re often working together to make great marketing. 

Many of us cut our teeth on the production side of the business. Some of us collaborate closely and regularly with these partners (in the ‘old days’, that meant hours and days together in dimly-lit editing booths, which we’re mutually glad has changed.... the snacks were always great, though).

Our responsibility is to set them up for success: with a great creative foundation that’s clear and executable. The partnership depends on it. So we take the responsibility to heart.

And when we learn about real cases where our partners have been handed what’s clearly AI-generated creative direction (formatted, but incoherent and unproduceable), we take it personally.

Because we love advertising. We want there to be more great advertising out there, whether we produce it or not (‘a rising tide’, and all that). And in our view, when poorly-conceived AI dreck makes it to market under the guise of advertising, it tarnishes a medium we’re extremely passionate about.

Plus, we have a lot of opinions. And the forum on which to share them. Welcome, BTW.

We're seeing unedited, AI-generated creative in the wild — and it's not pretty

Here's what it looks like in practice when an AI-generated script being put into production, without critical evaluation or review.

Radio scripts submitted to station copywriters — professionals who have spent careers crafting and producing commercial audio — are delivered by clients with script directions like "SFX: door slams happily." Or, for a financial services company: "In the distance, a dog barks." For no discernible reason. Adding nothing. Making no sense.

These aren't outliers. They’re real examples, shared with us by our real partners (anonymized for obvious reasons). And they're arriving at production houses every week. Often every day.

They're immediately recognizable for what they are: a brief (or more likely, a prompt) dumped into ChatGPT (probably the least creative of the AI platforms, for what it's worth), with zero critical evaluation before submission.

TV treatments aren't faring much better. We've seen agency-produced scripts — actual paid agency output — with no scene direction changes, no close-up callouts, none of the detail the production team needs to actually execute the work. Instead, professional-looking formatting masks a lack of professional rigour.  Scenes that have no logical connection to the next. Action completely out of sync with voice over. 

One producer showed us a script from which they couldn’t answer a basic, essential question: how many talent do I need to hire to film this spot? 

If you don’t value your marketing, why should they?

These are the kind of things that tells a producer immediately: nobody who knows how TV works touched this.

So the motivation to elevate your work drops. Because what’s the point of elevating the work if you’ve demonstrated you won’t know or appreciate the difference? 

You'll still get something back. It just won't be good. Because you don’t value your marketing. So why should they?

You can't evaluate what you don't understand

If marketing isn't your core capability (for most small businesses, it isn't), that's completely fine. But you probably don’t have the experience to tell the difference between a script that works and one that just looks like it should.

AI knows this about you. Not literally, but functionally. It's a people-pleaser by design. Ask it to write a radio script that hits ten key messages in thirty seconds, and it will do exactly that. It will give you ten messages in thirty seconds.

But it won't tell you that nobody will absorb any of them. It won't tell you that thirty seconds is roughly 75 words and you've just asked for a legal disclaimer, not a commercial.

It completes the task. It doesn't evaluate the brief.

Mediocrity training

But wait, it gets worse (for marketers, and those posing as them). 

Every mediocre piece of AI-generated content that gets created and shared is training future AI on mediocrity. The models learn from what's out there. Most of what's out there (that is, most publicly available creative work, most ads, most scripts), isn’t good to begin with. 

AI can't tell the difference between a Cannes Lion and a 2 a.m. car dealership spot. It draws from both, indiscriminately. So when you ask it to write you a commercial, you're getting (at best) an average of the whole pile. Which is, statistically, mostly bad.

You're not just getting slop. You're proliferating it.
 

3 ways to use AI to get better creative (without producing crappy content)

None of this is an argument against using AI. We recently shared 7 different use cases in our own agency, where the technology is helping us work faster, better and smarter. Instead consider this an argument for using AI to improve what you’re great at, not filling a gap you can’t critically evaluate. 

Here are three specific ways that AI can help anyone improve their marketing — even (or perhaps, especially) non-marketers.

Use AI to sharpen your brief, not to write your creative. 

You know your business, your customers, your goals. AI is good at organizing that information, identifying contradictions, and helping you articulate what you actually want. A tight, clear brief handed to a creative professional is worth ten times more than a poorly-interpreted AI-generated script. It makes that creative’s job faster and easier, and makes your work better.

Before you send anything, evaluate it in a fresh conversation. 

Seriously: generate your output, open a new chat, paste it in, and ask AI to be ruthlessly critical. Ask it whether the key message is actually landing. Ask it whether a script can be read comfortably in thirty seconds. Ask it what's missing. It'll tell you things it never would have flagged if you'd asked it to create and evaluate in the same breath. It's not a perfect filter, but it catches the obvious problems before your name is attached to them.

Know when to use AI — and when to use a professional.

If you're working with a station producer, a creative team, or an agency, they likely already have AI tools built into their process. They know how to prompt well, evaluate critically, and reject the garbage before it gets anywhere near production. Give them a great brief, and get out of the way. 

Let’s build your AI adoption strategy

All of this boils down to our core perspective: AI is best used to amplify your existing abilities, not as a substitute for what you don’t know.

The good news: there's a version of AI adoption that actually works. One that plays to the tool's genuine strengths, keeps humans with the right expertise in the loop, and produces output your team (which includes your creative partners) can actually use.

We've helped clients build exactly that kind of AI adoption strategy. Not a blanket "everyone use this tool" mandate, but a strategy that makes sense for how your business actually works.

If you want to figure out what that looks like for you, book a free discovery call with us. We'll help you build an AI adoption plan your team will actually embrace — and that won’t erode your relationship with producers. 

Or worse,  with your customers.