What’s Actually Earning Its Keep, AI and Innovation

There’s an old rule in live production that never gets written down but everyone understands: every seat on the crew has to be earned. Nobody is along for the ride because they might be useful someday. You’re on the crew because the show doesn’t work without you.

I’ve started holding technology to the same standard. Not “is it impressive?” Not “is it the future?” Just: has it earned its seat? After a couple of years of watching the newest generation of intelligent tooling move through real media operations — not keynotes, not pilots, but daily production — here’s my honest scorecard.

On the bus

Transcription and captioning at scale have earned their seat — with a human in the review chair for broadcast, exactly where judgment belongs. Metadata generation and archive tagging have earned theirs twice over, turning dark archives into searchable, monetizable ones at a fraction of what manual logging ever cost. Highlight and clip generation for social has become genuinely load-bearing in sports. Translation and localization workflows that were cost-prohibitive five years ago are opening markets that never justified the spend. And quality-control assistance — flagging issues for humans to review rather than replacing the review — has quietly become one of the best hires in the plant.

Notice what every one of those has in common: volume work, checked output, and a human still holding the judgment. The tool does the tonnage. The person does the truth.

But there are some other categories of content, and generative basis where application of this is truly innovating, and creating images that are compelling, immersive storytelling, and creation of a view, some frame, and images that would have otherwise not been feasible near live. But they come from the bottom up of adoption from the crew, not from the top down push because it was bought.

Still auditioning

“Fully automated” live production without meaningful human oversight has not earned a seat, and in my view isn’t close — live is precisely where the cost of error is highest and the time to catch it is zero. Anything that claims to replace editorial judgment about what matters and why is still selling the showroom version. And a personal tell I’ve learned to trust: any tool demonstrated only on perfect, well-lit, clean-audio content is a tool that has never met a rainy night feed from a compound at the edge of the connectivity map.

How the seat gets earned

Here’s the pattern I keep seeing in the organizations getting real returns, and it will sound familiar to anyone who lived through tapeless, through HD, through IP: they didn’t start with the boldest strategy. They picked the boring, high-volume problem their people already hated, instrumented it honestly, and expanded from evidence. The revolution never announced itself. It just showed up in the budget, one earned seat at a time.

That’s how every transition I’ve lived through actually worked. The technology that survived wasn’t the technology that demoed best or some sales person convinced someone to buy on concept. It was the technology the crew stopped noticing — because it had simply become part of how the show gets made.

So let’s drop a waypoint: stop asking what’s revolutionary. Start asking what’s earned. The bus leaves for the next show either way — and the roster, as always, tells you everything.