Radio's Slop Detector Is The Key To Better AI Results
Slop isn’t new. It’s not an AI thing. Although AI does take the slop category to a new level, mass-production of mediocrity has been around since the first caveman drew a buffalo on a cave wall and his buddy said, ‘Eh, the legs look weird.’
We’re surrounded by slop that we’ve made peace with. Like the chain restaurant where the food looks delicious but tastes like cardboard, and the ultra-cool, slick-looking piece of furniture that looked great on the website and fell apart in six months.
Slop, especially in AI, isn’t really about a lack of effort. It’s about gatekeepers deciding that close enough is the goal. Think about the drive-thru burger engineered to be just good enough. Or the station image sweeper that checks all the boxes, is technically correct and completely forgettable.
Radio people have a slop detector that most industries never develop. In our world, it's a survival skill. Social media influencers who post a mediocre video here and there might see a dip in their reach. A restaurant that serves a bad dish gets beaten up in a Google review. Those are recoverable. But when a radio personality sleepwalks through a break risks losing the listener’s trust that took years to build.
The connection between a radio personality and their audience is a personal experience. We’re in the business of building relationships. Slop, in any form, puts a crack in something that took a very long time to build.
The frustrating irony is that the people best equipped to get great results from AI are the same ones walking away convinced it doesn't work. Radio personalities have a high standard. They hear the slop immediately, reject it immediately, and reasonably conclude that the tool isn't up to the job.
“Get back to me when this genius in a box can do something more than waste my time.” That’s a direct quote from an early conversation I had with DJ, who now uses AI daily. The tipping point came with the understanding that the problem wasn’t the tool, it was the lack of conversation with the tool.
AI doesn't know what your audience wants at 6:45 on a Tuesday morning. It doesn't know your history, the sweet spot and the edges of your content strategy, or why that joke lands in Birmingham but dies in Boston. But you do. Not sharing it gives AI the green light to guess, as it hops the bullet train to Slopville.
“How in the H-E double hockey sticks am I supposed to teach it what’s important? It took me years to understand it myself.” Another direct quote. You can probably imagine the reaction when I suggested using AI to help the AI learn what it needs to know. I won’t share that quote; this is a family show.
Your next steps: Create knowledge documents that guide AI’s responses. Make sure they cover your brand identity, content strategy, target listener persona for the station, and personality, style, and tone for DJs.
How to get help from AI. Prompt: Use this process to create a brand identity knowledge document for a radio station: interview me to learn the most important information required to create a comprehensive brand identity overview. Do not create the questions all at once. Create and ask questions one at a time. Think about my answers and what they reveal to help create the next question. Continue this process until you have gathered the information required to write a comprehensive brand identity overview.
Add the document to a prompt, a custom GPT, Gem, Project, etc. You’ve taken the first step in helping AI start from a place of context instead of a cold guess, and something it never had before: a reason to get you right.
On Friday, 3/20, at 3 pm at CRS, I’ll share a step-by-step framework you can use to create a comprehensive knowledge document library in a single day. I hope to see you there, or in my in-box the week after: lindy@lindymedia.net.
