Radio stations are being warned not to mistake working AI generated software for a secure and sustainable broadcast system.
The warning comes in a blog written by Darren Shortt from The Media Site, looking at the increasing use of AI tools to create playout and other broadcast software.
Darren says more stations are experimenting with software created using services including ChatGPT and Claude, but argues that getting something to work is only one part of running technology relied upon around the clock.
“A piece of software that works and a sustainable, secure broadcast product are two very different things,” he writes.
The blog highlights questions around who maintains the software, fixes problems, monitors security and ensures it works correctly alongside other systems.
Darren describes an example involving an overseas customer which installed its own AI written playout system on a hosted server. According to the blog, the software generated a large amount of network traffic and flooded the network with packets, resulting in The Media Site removing the service.
He says AI generated software can perform its intended task without necessarily understanding the wider environment in which it operates, including shared infrastructure, network security, resource limits and monitoring.
The Media Site says it also uses AI as part of its own development work, but its software is subsequently reviewed, tested, monitored and maintained.
The company has also introduced a policy requiring approval before third party playout, streaming or broadcast software can be installed on its network.
Darren says the message for stations considering AI created systems is to think beyond whether the software works when first installed.
“Just make sure somebody is still there when they break,” he writes.
And yes, we are aware of the irony of an AI image used for this story, which was also written by AI. 🙂
