AI can write your audio ad. Here’s what it can’t do.
We use AI. We’re not worried about it. Here’s why.
Let’s get the obvious out of the way: AI can generate an audio ad. Give it a brief, a brand name, and a product message, and it’ll produce a script in seconds. Pair it with a synthetic voice and some library music and you’ve got something that… sounds like an ad.
For some uses, that’s genuinely fine. A functional announcement, a local promo, something that needs to exist quickly and cheaply. There are tools for that now, and they’re getting better.
But here’s the thing. There’s a gap between content that exists and content that performs. And that gap is where humans still make the difference.
What AI is good at
Speed. AI is extraordinarily fast at generating options. First drafts, variations, alternative structures, tone-of-voice experiments - things that used to take hours can happen in minutes.
That’s useful. We use AI tools in our own process, particularly in early-stage ideation and post-production workflows. It makes us faster, and faster is good.
AI is also increasingly useful for creative testing. Running executions through AI-powered analysis to predict how an audience will respond before you commit media spend. That’s a genuine advance that helps everyone make better decisions.
What AI can’t do
It can’t tell you whether a script is any good.
That sounds flippant, but it’s the crux of it. AI can generate a script that’s grammatically correct, on-brief, and structurally sound. What it can’t do is tell you whether it’ll make someone lean in. Whether the pacing will hold attention at 7am on a Tuesday commute. Whether the tone will land with a 35-year-old first-time parent or alienate them.
That judgement - the instinct for what connects with a human audience - comes from experience. From having written, directed, and produced thousands of pieces of audio and seeing what actually works in market, not just what looks right on paper.
Direction is human work
Even with a perfect script, audio performance lives or dies in the direction. The way a voice artist delivers a line. The pause before the key message. The energy shift between the setup and the payoff.
These are micro-decisions that shape how a listener feels, and they require someone in the room who understands what they’re listening for. AI can’t direct a voice performance. It can’t hear that a read is technically correct but emotionally flat. It can’t coax the version that makes the whole thing come alive.
The strategy layer
AI generates content. It doesn’t ask whether the content should exist in the first place.
Should this campaign run on radio or streaming? Does this script need three versions for different dayparts, or one that works across all of them? Is the brief actually right, or should we push back before we waste studio time?
Those are the questions that determine whether a campaign performs. And they require people who understand audio as a discipline: the channels, the audiences, the way creative and media interact. Not just people who can operate the tools.
Where this is heading
AI isn’t going away, and it shouldn’t. It’s already making good audio production faster and more accessible. The brands getting the best results are the ones using it to accelerate the process, not to replace the thinking.
What’s actually being automated is the announcement end of the market - functional, low-stakes audio that never needed deep creative input in the first place. That part of the industry was already a commodity. AI just made it cheaper.
The gap between adequate and genuinely good is getting wider, not narrower. And the brands that invest in the human expertise to close that gap are the ones whose audio will actually perform.
Want to know how we use AI in our process — and where we don’t? We’re happy to walk you through it.
