In this letter to the editor, Aaron Read responds to Rojith Thomas’ feature “The Predictive Engineer.” Radio World welcomes letters to the editor on this or any story. Email [email protected].
Arguably the greatest challenge any engineer faces is knowing when a new technology is worth it, and when it’s not. Fortunately for us, when it comes to generative artificial intelligence (gAI) and Large Language Models (LLM), the challenge is easy: It’s not worth it.
To begin with, the entire AI industry is a bubble that literally is bursting as we speak. Ed Zitron has been a pioneer in covering this very well, but the only evidence you need is that in May and June of 2026, AI companies started changing their billing models from flat subscriptions that disguised the true cost, to per-token-billing that suddenly meant massive price increases. Within weeks, months at the most, any AI tool to do the examinations Rojith Thomas suggests will become far too costly to justify the limited ROI it can provide.
But really this is more about overlooking simple, tried-and-true solutions … in favor of the flashiest lights and loudest bells and whistles.
If you want your radio station to stay on the air no matter what? There’s a simple way to achieve that: redundancy. Backup transmitters, auxiliary antennas … or even fully redundant licensed aux sites, backup power generators, backup audio processors, backup studio/transmitter links, STL devices that perform “bit splicing” across multiple ISP’s, redundant automation systems and automatic switchover via silence detection. And remote control + telemetry systems have never been more powerful, flexible and “smart” enough for an engineer to program in specific tasks to be done in response to certain meter/status readings (e.g. “macros”).
[Related: Read the Radio World Ebook “Optimize Your Air Chain”]
Notice I said “simple,” which it is … but it’s not necessarily “easy,” and most likely it’s not “cheap” either. Essentially you are building an entire, second radio station to operate in a “soft parallel” or “warm standby” to the first. That’ll cost money, and quite possibly more money than it is worth! How do you determine the ROI?
For a commercial radio station, it’s easy: How much will being off the air cost you in terms of missed spots and make-goods? Maybe apply some X factor of lost listeners during the downtime if you like. Certain stations, like sports outlets, might also have an X factor of lost revenue if the outage is during live game coverage, too. If you’re a major-market station with a healthy ratings share? Being off the air for just one hour could cost you thousands of dollars, maybe tens of thousands. The ROI of redundancy starts getting pretty obvious, pretty quickly.
For noncoms it’s a little tougher. The bigger stations can still use money as a determinant: Maybe it’s called “underwriting” instead of “sales,” but money is money. You could also calculate how much if the station goes down during a pledge drive, too. And you may have the advantage of access to certain grant funding and/or major gifts for redundant technology solutions.
But a lot of it is going to be trying to assign an objective number (lost revenue) to a subjective concept (the station’s mission). It’s still doable, but you’re more likely to explicitly need your CEO’s help, along with your major gifts officer and/or membership leader. Help them decide what the dollar value is to being off the air. Besides knowing what your redundancy budget will now be, you’ve performed “CYA” when and if your station does go off-the-air and they’re antsy about it.
Of course, the flip side here for any station, commercial or noncom, is that this calculus might very well tell you the station just isn’t important enough to have ROI to justify all that expensive redundancy. That, I would argue, is indicative of a failure in how that station is being operated, writ large. Or perhaps more kindly, it’s indicative that the station’s signal just isn’t economically viable in 2026, period. But at least you’ll have the numbers you need to justify the decision … even if the decision is to forego any redundancy.
You’ll notice something totally absent from this entire process: AI. Given that artificial intelligence frequently fails miserably when used to substitute for the judgment of skilled professionals, why bother using it when there’s already a perfectly functional system that we’ve had for many decades?
Aaron Read is a radio broadcast engineer with more than 25 years of experience in the field. You can reach him at [email protected].
[Check Out More Letters at Radio World’s Reader’s Forum Section]
