Comment from Wayne Stamm
I’ve been watching with some interest the articles and debate around AI replacing radio talent.
Shelly Palmer (pictured) has been taking an in-depth look at ChatGPT and has a resource page that is worth exploring.
Shelly also suggests taking out a subscription for ChatGPT for US$20/ month in order to fully understand what it can do. The subscription can be cancelled at any time.
Recently Futuri launched RadioGPT™, the world’s first AI-driven localised radio content solution, and it is worth a listen.
It isn’t all that bad, though there are questions about how much content it will be able to access before regurgitating the same items. Like everything it will depend on how much scope the programmers give the AI to explore in order to get content.
Then there is Craig Bruce’s article on our sister publication Radio Today, AI isn’t about to replace YOU.
He makes some great points including things that AI can’t do:
- Relate to their audience through shared experiences.
- Listen to their co-host and respond in the moment.
- Know their co-host well enough to be able to ask a relevant question or support with a relevant prompt during a shared conversation.
- Take the piss out of their co-host.
- Pause for comedic effect.
- Be vulnerable. Share something from their lives that might be painful or difficult to talk about.
- Talk to callers and guests with empathy.
- Have human experiences with their families.
- Love someone, their dog, children, footy team.
- Feel and express emotions – Adoration, Amusement, Anger, Awe, Confusion, Contempt, Desire, Disappointment, Distress, Fear, Interest and Sadness.
He also had these on his list, and I’m not sure he is correct about the following, as on RadioGPT they are getting close to getting this covered.
- Know the name of the local business owner who is in the news this week.
It can, if you give it the access to local news
- Respond in the moment with a funny remark.
We are already hearing RadioGPT try some added humour, and while the AI currently has no sense of timing, or the pregnant pause, it can tell a joke as well as a lot of the onair talent running a music shift around Australia today.
RadioGPT also has access to most everything ever written about music and artists, and uses it to entertain its audience while linking it to songs.
But sometimes it can go horribly wrong depending on where the AI is getting its information.
Craig is one of the great coaches that radio in this country has had, and while I agree with his sentiments when it comes to AI competing with Breakfast and Drive shifts around the country, there is an opening for regional and even metro markets outside those shifts.
Nights, afternoons and mid-dawn in particular would be an ideal target in regional markets for something like RadioGPT.
As Shelly Palmer says,
I am reminded of a conversation I overheard a long time ago in a galaxy far, far away in a cantina near Mos Eisley Spaceport on Tatooine.
‘Hey! We don’t serve their kind here!’
‘What?’
‘Your droids. They’ll have to wait outside. We don’t want them here.’And yet… these droids were the droids we were looking for. They were the AI co-workers that helped save the galaxy (alongside Leia, Luke, Han, and Chewie).
Galactic politics aside, technophobia isn’t going to stop (or even slow down) the disruptive innovation empowered by large language models, generative AI, or any other type of machine learning. It’s just going to disadvantage those who don’t have access to it.
Already the Writers Guild of America (WGA) and the Alliance of Motion Picture and Television Producers (AMPTP) is concerned enough to be looking to “regulate use of material produced using artificial intelligence or similar technologies”, and voice over artists are concerned about their voices being used as models for AI.
Radio GPT doesn’t appear to have the inventiveness of a live person, nor the passion.
The following is one of my favourite clips from Radiodays Europe 2022 featuring Stephanie Hirst from Hits Radio in the UK, and it shows what a real person can do.
If radio talent is content to sit the chair and not be inventive and relatable to the audience, then AI might find a way in.
It is slick, and it is no worse than some stations I heard while recently touring both here and in the US where poor voice tracking seems to have become the norm.
It is hard to be spontaneous when you are voice tracking, but it seems to me, that even on some of the major metro stations, voice tracking is way too obvious, especially on weekend shifts.
I think what we need is a whole lot more Craig’s guiding talent around the country.
Look at his list above and see how much of this you are doing as an onair talent, and how much you can add to what you are doing now to make sure AI stays away from your shift.
And look at this as a challenge
I would consider the fourth point at this link https://www.shellypalmer.com/2023/02/ai-detectors-unreliable/ .
Who is accountable for any biases and prejudices produced by AI-generated presenters?
The "Elton OOPS" track illustrates the announcer referring to the artist's personal sexual preference which was IRRELEVANT to presenting the artist's music.
It may well also be dangerous to refer to, or use, derogatory language about the artist's ethnic origin or disability.
It is dangerous to the licence holder who may be accountable to anti-discrimination and human rights tribunals.
On the other hand the program output may be include biographical information which may include the ethnic/cultural origins and/or disability about artists for example Dr Gurrumul (RIP), Jeff St. John (RIP), Ian Drury (RIP).
As I commented on a recent article on this site about AI bots used for the analysis and presentation of program content the danger of many stations using the same bots resulting in a bland presentation in a saturated market.
It seems that AI presented info will sound OK for linking tracks with announcements and minimal or nil commentary.
But where is the spontaneity of a live Ward Pally Austin or Doug. Mulray?
AI is may work for news bulletins linking stories but not for spoken word content including interviews, documenatries and talkback.
Thank you,
Anthony of let's not get overboard with novelty, Belfield, in the land of the Wangal and Darug Peoples of the Eora Nation.
The threatened invasion of AI into entertainment spaces simply means we humans have to lift our standards.
And after a long time of throw-away song breaks, client written commercials, recycled 1960s radio competitions - a lift in human standards on air is a good thing.
Don't mean to generalise but .... a lot of American presenters (particularly those with low pitched voices and slow speech) kind of sound a little like AI anyway.
In my view, the quality of, and engagement created by, Aussie, Kiwi and Brit presenters is considerably higher than that in the US.
I have played with quite a few AI speech products and none of the sensibly priced one are acceptable for our markets.