On a recent episode of Drew Carey’s SiriusXM show, he caught flak from listeners after they figured out he was using an AI voice replica of himself to do the show. Carey admitted he used a text-to-speech AI program to voice scripts written by ChatGPT as an experiement. He told Engadget, “I violated a rule from Radio 101. The reason FM stations and treasured radio stations still make money is because people like the personality of the DJs.”
After that Engadget interview and Radio Ink’s coverage, Futuri CEO Daniel Anstandig has released his response in defense of AI in a LinkedIn blog post.
In February, Futuri debuted RadioGPT – combining GPT-4 technology with Futuri’s AI-driven targeted story discovery and social content system, TopicPulse. It scans social media and other sources to identify trending topics and creates audio using AI voices. RadioGPT’s AI can also be trained to a station’s existing personalities’ voices.
Anstandig opened, “While it’s easy to dismiss this as a cautionary tale against the use of AI in radio, we believe that the issue isn’t the technology itself, but rather how it was applied and communicated to the audience.”
He continued on the technology Carey was using – all basic versions of AI tech. Anstandig says had he gone “all-in” with “a far more sophisticated and humanistic AI language model, the outcome could have been drastically different,” adding, “Quality counts!”
He also blames Carey’s lack of honesty about his use of AI upfront for the poor reception. “I believe that in the event a personality uses AI for an entire show for one reason or another, there should be transparency with the audience.”
Anstandig finished by saying that AI has the ability to enhance radio programming, especially in dayparts lacking human talent. “Although Drew Carey’s experiment with AI may have backfired, it should not discourage the radio industry from exploring the benefits of AI in content creation.”
Read Anstandig’s full response here.
I absolutely agree with Mr. Anstandig’s take on this topic. Even thought he’s in a position to be biased toward support of AI, he’s also exactly correct.
Think of stations you’ve heard that were automated in one way or another – even from back in the days of “Satellite Music Network” and the variety of liner slots they provided. It wasn’t difficult to find broadcasters who were sloppy or careless in implementing early automation systems – and just about as many today. Automated (and AI-generated) radio can sound pretty awful. But, it can also sound just about perfect. Careful listening and adjusting of any automatic system is required for it to perform well.
I’m looking forward to AI-generated voices getting better and better, with even more contextual understanding and knowledge among broadcasters of how to prompt and otherwise manipulate the instructions going in, so the results coming out are very human, indeed.