Lately · Experiment · May 2026
The FIFA Files
I gave AI a podcast to make, mostly to find out where it would fall over.
The podcast is The FIFA Files, forty episodes (yikes) on how football's governing body got bought, sold and quietly let off the hook. I picked FIFA because the story is genuinely massive. Decades of court filings, indictments & leaked emails. The sort of thing that needs a newsroom. I wanted to see how far one person and a few tools could get on their own.
Quite far, but definitely not all the way. Claude built the website. NotebookLM swallowed every document I could feed it and turned it into something I could actually navigate, a sort of production bible I could ask questions of. That's a useful tool and I'd recommend doing something similar on your next creative project. It allows you to experiment with ideas, ask questions about things you forget mid-project and add other context you wouldn't normally have time to consider to give a more rounded story. The audio summaries are also helpful to learn and absorb knowledge.
Claude and Gemini helped frame the episodes. On the heavy lifting it's brilliant. Reading a mountain of context. Building a home for the work. Fishing clips out of a long edit. All the jobs nobody dreams about doing.
It gets you about ninety percent there. Then it really falls down. The artwork is average at best. The production and delivery is basic. You can see the exact point where a human with a bit of taste needed to step in. Same with the site. It could be nicer. I leaned it on FIFA's own colours and pushed it about as far as a Saturday allows. Every one of those creative decisions was mine. The AI didn't have a single opinion about what the thing should be. That still needs humans in the room who actually care.
Which is the whole point, really. I came up in radio, and the bit I love is people discussing how to make something good, what story is relevant, experimenting with sounds and format. That messy back and forth is the job. It's also what makes the thing worth listening to. Automate it away and I start to wonder why anyone would press play… or why you'd want to make it in the first place.
Here's the bit that surprised me. I didn't much enjoy making the podcast. Once AI was doing the heavy lifting, the creation itself became a bit of a chore, because I wasn't making any real creative decisions. The part I actually loved was the opposite end: working out how to set the project up, how to problem solve with AI, how to get it carrying the boring load well. Hand too much over and the thing still gets made, but the value you take from making it quietly drains away.
Which is a more interesting question than the usual one. It's not just what AI can do. It's what AI and humans should do, and where the enjoyment actually lives. I don't think you give that up for the sake of it.
So: brilliant experiment, and I'll happily keep the grunt work automated forever. I'll just use that saved time to work on the parts that matter and that I get the most enjoyment from. Have a listen and see if you can hear the ninety percent.
The FIFA Files →Written by Ben Jay, Head of Video at Trustpilot, who spends his spare time experimenting with what AI can and can't do for creative work. More experiments → · Get in touch →