Under the Influence (with Jo Piazza) - June 30, 2024


Sunday Nice Things: The Creative and the AI

Topics
Stop Giving All the F*cks AND What to Read Right Now Escape from the Christian Patriarchy

Episode Stats


Length

24 minutes

Words per minute

190.56

Word count

4,681

Sentence count

211

Harmful content

Hate speech

3

sentences flagged


Transcript

Transcript generated with Whisper (turbo).
Hate speech classifications generated with facebook/roberta-hate-speech-dynabench-r4-target .
Topics generated with Qwen2.5-3B-Instruct.
00:00:00.260 I didn't like what you said, Odin pauses, about the future.
00:00:05.900 This is the love story of real hinge couple Odin and Edward,
00:00:09.880 written and read by me, Curtis Garner.
00:00:12.640 Listen to the free audiobook now.
00:00:17.000 Before you knew what a stock was, you traded snacks, cards, turns.
00:00:22.740 You knew what something was worth because you felt it.
00:00:26.160 That instinct to trade didn't disappear.
00:00:28.900 It just grew up.
00:00:30.420 TD Easy Trade taps into that instinct so you can build something real for the future.
00:00:35.400 With no minimums, no monthly fees, and 100 free trades.
00:00:39.200 You already got this.
00:00:40.820 Because you are made to trade.
00:00:43.360 And TD Easy Trade is made to help.
00:00:45.820 Download it now.
00:00:50.800 Hello, all.
00:00:52.120 Welcome to Sunday Nice Things, where we drop an episode of a new podcast or an old podcast that I'm just discovering into your feeds so that you can give it a try.
00:01:03.960 It's hard for people to discover new things that they love, and I want to try to make it easier.
00:01:09.340 I've recently been listening to this show called The Creative and the AI, and I've been listening to it because I live in massive fear of AI taking over all of my writing jobs.
00:01:19.620 which which to be honest is an actual real fear it is i was recently doing a consulting job
00:01:27.560 for for a large large company that had a lot of editorial content and they recently replaced
00:01:34.720 writers with ai and they asked us to edit the things that ai had quote written essentially
00:01:41.920 my job as an editor was now to make the ai things sound less like a robot and i'll tell you i quit
00:01:47.120 on the spot. That was my line in the sand. But I think about this all the time. So this podcast,
00:01:53.520 The Creative and the AI, is essentially a journey to understand how AI is revolutionizing the
00:01:59.440 creative process across industries, from art and literature to design and technology. They uncover
00:02:05.560 groundbreaking innovations that AI can bring to the fields, and how it can enhance but also
00:02:11.920 challenge and transform the work of human creators. Today's episode is about how AI is changing the
00:02:20.000 world of podcasts for podcast producers. I mean, if you're an audio nerd, this episode might just
00:02:25.680 be your dream. And if you're not, and you're just curious about how AI is changing the world of
00:02:30.240 podcasts, you may also be into it. But you never know. You never know what you're going to get on
00:02:34.980 Sunday nice things. You don't. And that's the beauty of it.
00:02:41.920 all of my thoughts, all of my creative work can be somehow augmented by AI. And I do say the word
00:02:49.280 augmented because I know a lot of us in various creative industries are worried, are we going to
00:02:54.480 be replaced? And I think these tools are getting better and better. And there's no doubt about it.
00:03:00.280 Many roles and many functions will be replaced or changed. Imagine you are a podcast producer.
00:03:07.120 you have an increasing number of tools at your disposal.
00:03:10.080 Eleven Labs, Descript, and OpenAI's Whisper are becoming a regular part of your process.
00:03:16.000 You even use ChatGPT to research ideas and concepts faster than ever before.
00:03:21.700 However, many of your colleagues in podcasting, radio, and music
00:03:25.180 are worried that they stand on the front line of AI replacement.
00:03:28.840 So what's the reality?
00:03:30.080 Should audio creators be concerned?
00:03:32.060 My guest today is Mike Russell, Creative Director of Music Radio Creative, a specialized audio
00:03:38.540 branding agency that offers podcast, radio, and music production services.
00:03:43.780 Mike shares his 25-plus years of experience in audio production with his large following
00:03:48.120 on YouTube.
00:03:49.320 He is an AI enthusiast who regularly talks in person and online about the various ways
00:03:53.920 that AI can augment the creative process.
00:03:56.580 Let's jump in with Mike.
00:03:58.360 Audio has been my thing since a very young age.
00:04:00.960 you know when I was in single digits I was doing bedroom radio and pretending to be on my own little
00:04:06.340 radio station so I've always had that interest in radio and then as a result audio but I've
00:04:11.680 also been a like just fascinated by technology and the ability to improve and make our lives better
00:04:18.200 and I think if used correctly and if you learn the ways to do these things it can really make
00:04:25.880 for incredible times ahead. One of the other aspects of AI, I guess related to audio, is the
00:04:33.040 ability to speak to AI and hold a conversation with AI. Now we know that OpenAI can do that
00:04:38.760 with their chat GPT. You can hit the headphones icon and you can have a chat and it's pretty good
00:04:42.800 back and forth. But one of the tools that I played with a while back and I covered it actually on a
00:04:48.060 podcast I did, it's called Pi from Inflection AI. I don't know if you've heard of Pi. Pi is
00:04:53.960 incredible. So I think the two co-founders asked were something to do with LinkedIn and Google
00:05:00.020 DeepMind. So they've come from a very interesting tech background and they've put together this
00:05:05.620 essentially large language model like ChatGPT, but it's so incredibly human. So it's a bit like
00:05:12.040 the movie Her. Have you ever seen the movie Her? Right. So it's a bit like that. And it really is.
00:05:17.960 I'm not kidding when I say it's a bit like that. You can have this conversation and, you know,
00:05:22.980 I've started using Pi a little bit more. And the interesting thing I find when talking to Pi
00:05:28.060 is it remembers things about me. So I can ask it questions and it will repeat back to me.
00:05:33.880 And it's pretty scary the things that you just say passively that it remembers.
00:05:37.780 So it'll be like, I know you like running. I know you've got a dog. I know you go out early.
00:05:42.500 I know you live here. I know you've got two kids. It's like, wow. Okay. Yeah, I have. I have
00:05:47.700 mentioned all those things, you know, in one way or another, it's incredible for brainstorming
00:05:53.660 things. I think there's also the opportunity to go down the therapy line as well. Although
00:05:58.300 obviously you're giving your very personal information potentially to an algorithm and
00:06:03.280 essentially to a computer, right? But some people in a sense, particularly I think those growing up
00:06:09.280 who are growing up now with AI, so not my generation who were around even before the
00:06:14.920 internet right to tell your kids i live before the internet makes you sound like a dinosaur right
00:06:19.840 but so maybe i think the next generation is going to be well this is just normal just to talk to ai
00:06:25.160 of course i'm going to tell it all my problems and spill everything out because this is just this is
00:06:29.540 just what we do and actually it's far better than talking to a real human because they judge you
00:06:34.220 even if you're paying them if they're a real therapist they judge you right but a computer
00:06:39.620 doesn't. So for those outside of podcasting and audio, could you give a brief overview of some of
00:06:47.300 the new AI enabled ways of working that you're seeing today? There's so many cool ways that you
00:06:53.560 can use AI in your workflow right now. I love it because I am never short of a tool at the moment
00:06:59.960 to use to make something creative. I think one of the biggest things we've seen happen in the last
00:07:04.340 year since everything has surfaced like chat gpt and the image generation models and things like
00:07:10.600 that generative audio is getting incredible so the ability to not only use voices that are generated
00:07:18.120 by ai so you can you know use a slider to change the accent change the gender change the age of
00:07:24.440 the voiceover but you can also clone your own voice this is something i'm finding really
00:07:28.840 interesting right now for creating produced pieces and and mixing things and before about a year ago
00:07:35.780 voices still sounded very glitchy they sounded very ai generated but now you can really make
00:07:41.160 stuff another couple of areas that i'm interested in at the moment are generative music i think
00:07:45.780 there are some amazing tools that are creating music much like we're seeing with the new tool
00:07:50.300 from open ai sora for generating movie clips you know you can generate full pieces of music now
00:07:56.140 giving creators an opportunity to say, well, hey, why would I license stock music when I can make
00:08:01.660 my own track and have it sound exactly the way I want? And then let's not even start with AI
00:08:07.060 generated sound effects as well. So the area is vast and that's just lightly touching the surface
00:08:13.540 with audio. Of course, there are so many other cool things. So what creative possibilities have
00:08:18.040 you found with these new tools or are there limitations that you would highlight?
00:08:22.240 creative possibilities are huge look at audio that's a big lane that i set in really in my
00:08:28.960 day-to-day so i i used to work in radio before i got into producing online through our company
00:08:34.640 and then obviously now using ai and i would connect with a voiceover artist on another side
00:08:40.700 of the world and you'd prompt the voiceover artist and you say right can we have three
00:08:44.420 different takes of this and no could you say that word differently now we've got to the stage with
00:08:48.820 AI where you can type a little script in and it will give it back to you in seconds, not minutes
00:08:54.300 or hours. It's instant. And what is more with technology such as AI speech to speech, you can
00:09:00.960 actually speak the way you'd like your voice to say something and it will repeat it back exactly
00:09:06.340 the same. It's pretty incredible. So I could pick a really good AI female voice and I could say my
00:09:12.460 liner exactly the way I'd like it delivered and it will repeat back to me exactly the same. It's
00:09:16.900 speech to speech. That is a creative way. I'm loving using AI because as a producer, an audio
00:09:22.580 producer, I like things to sound a certain way. And I can now get that instantly with AI. Then
00:09:28.220 there's the generative music and also the ability to create sound effects and things like that we
00:09:33.180 cannot see into the future. Like I alluded to earlier, it's like in the past, you would just
00:09:38.540 get a job for life and it would be fine. And you do that job and hopefully you'd love it and then
00:09:42.960 you die but now we're at a turning point where you know even I question that so not to get really
00:09:49.180 you know existential on on the show but it's like you know I learned audio all through my life
00:09:55.220 and now I'm realizing probably a lot of the knowledge I have won't be needed anymore
00:09:59.640 things like how to set an audio compressor how to EQ a voice how to remove background noise from
00:10:06.000 your recordings people won't be interested in general as to how that's done they'll just click
00:10:12.720 a button and the AI will do it and to have that realization and I'm sure many of us in our
00:10:17.060 industries are having that realization is incredible you think I spent years dialing
00:10:23.280 that in and working on that and working with so many different projects and that information is
00:10:27.800 now invalid right it's mind-blowing in a way but like you say it's about leveraging right and like
00:10:33.280 and to some degree and this is how I think about some of the projects I'm working right now it's
00:10:37.040 like which are kind of AI adjacent is like, well, you just free your mind up for the more creative
00:10:44.920 processes. So like that was just like a redundant thing that like, you know, it was taking time,
00:10:49.880 but it wasn't like really like needing the creativity of a human where there are so many
00:10:54.660 things that do. So it just maybe, you know, levels you up in a way because it takes away that sort of
00:10:59.440 baseline or those table stakes things that you were going to have to do. And now you don't have
00:11:03.200 to like waste your time or spend your time or use your mental energy. And then you can just kind
00:11:06.980 of like I say hopefully bring yourself to a whole nother like plane because you're not doing those
00:11:12.180 sort of mundane repetitive behaviors I fully agree with that what you said there about doing the
00:11:19.040 mundane behaviors I think even if we're doing what we like creatively there is so much of what we need
00:11:26.640 to do to get that final result that is like you say mundane and boring monotonous repetitive and
00:11:33.440 you know, I'm all for that. Even before the big AI boom that we've had, you know, computers have
00:11:40.760 been able to aid us in incredible ways. But this is a whole new thing. And it's exciting.
00:11:46.720 Wow. Yeah, so deep. The whole other world.
00:11:50.520 I didn't like what you said, Odin pauses, about the future. This is the love story of real hinge
00:11:57.580 couple Odin and Edward. Written and read by me, Curtis Garner. Listen to the free audiobook now.
00:12:06.680 Before you knew what a stock was, you traded snacks, cards, turns. You knew what something
00:12:13.440 was worth because you felt it. That instinct to trade didn't disappear. It just grew up.
00:12:20.100 TD Easy Trade taps into that instinct so you can build something real for the future.
00:12:24.580 with no minimums, no monthly fees, and 100 free trades.
00:12:28.900 You already got this because you are made to trade.
00:12:33.060 And TD Easy Trade is made to help.
00:12:35.540 Download it now.
00:12:36.740 So AI-enabled tools are disrupting and changing the creative process in your work.
00:12:41.500 And how are you seeing this happen or some of the examples that you've been witnessing?
00:12:45.700 So I can certainly say that we're on an upward curve with AI.
00:12:51.640 I think everyone is agreeing on that.
00:12:53.920 We don't know where it's going to go. We don't know if it's a bell curve. And if we're in the sort of middle to start of that curve, we don't know if it's a hockey stick. We don't know if it's a wave that's going to go up and then down like the, you know, the dot com bubble and all of that. We just don't know. You know, my hunch is it's going to keep going up and up and up.
00:13:11.600 But, you know, saying that, we have definitely seen over the last year, 2023, there was a
00:13:20.360 certain point where interest in regular voiceovers went very quiet for a while.
00:13:27.060 Thankfully, it did pick up again.
00:13:29.040 That was when I think it was the peak of the zeitgeist for AI-generated voices.
00:13:33.680 So everyone was trying it out.
00:13:35.320 And suddenly, well, why would I buy a real voice?
00:13:37.400 when I, and then I think we got to the stage where, well, it can go so far, but maybe we still
00:13:41.660 need humans a little bit. So there's a nice mix there, but I've no doubt that that is going to
00:13:46.080 increase and increase. Now, what we have seen, because obviously I like to follow the trend
00:13:50.320 and I'm very keen to integrate AI into our business. And I think any creative listening
00:13:56.160 to this show should be thinking about, you know, at the top of their mind, what is my AI policy?
00:14:02.220 how am I going to disclose how AI is being used in my business activities? If it's just a simple
00:14:10.020 reply to someone's email or making their voice available as an AI clone, how is that going to
00:14:15.920 be disclosed? So already we're thinking about processes in our business and also the fact that
00:14:21.140 we have AI voices available alongside human voices. And it's interesting to see how we've
00:14:27.500 suddenly hit a point where AI voices are starting to get really popular. So this is a trend that I'm
00:14:34.040 seeing that is not slowing down. So it's like, you know, get on the train or you're going to get left
00:14:38.840 behind. Embrace AI or you will get left behind. And what I'd say, it's crazy to think that in the
00:14:45.720 voiceover and the radio and the audio industry and television and media in general, you know,
00:14:50.600 a lot of things are licensed on usage, right? So it's like, where is this going? Is this going on
00:14:55.800 a podcast okay well you get charged you know x that's quite a low rate because you know we don't
00:15:00.740 know how many downloads it will get you know if you get to the size of joe rogan we might charge
00:15:04.480 you a bit more but if it's going on radio well where is it going is it going locally is it going
00:15:08.760 nationally if you're creating an ad in the us you know is it is it going on a like regional tv station
00:15:14.140 or is it going on during the super bowl and you know based on that that's how much the creative
00:15:19.120 would charge right but now with ai you just look at the terms and conditions on the ai voice website
00:15:25.000 and it says, this is available for your own commercial use, unlimited. Great. So I can
00:15:29.500 generate an AI voice for 30 pence and have it on a Super Bowl ad. Excellent. It's nuts. So the
00:15:37.060 economy dynamics of this are interesting. So I would love for you to delve a little deeper
00:15:43.380 into the AI capabilities around voice acting and voice cloning. What else are you seeing today that
00:15:49.020 you find really interesting? Yeah, sure. So it's an incredible space that's developing at light
00:15:54.080 speed just over a year ago now you were able to accurately clone a voice with a three minute
00:15:58.700 recording which was pretty incredible to be able to do that before 2023 you'd have to literally
00:16:05.420 upload hours and hours of work and it would have to accurately match scripts that you were provided
00:16:11.000 with but now we've seen that it's cut down to 30 seconds of audio to clone a voice and I know that
00:16:19.040 we'll probably touch on this later on in the show but the ethics of that are insane because we've
00:16:23.900 already heard you know the the clones of joe rogan you know hosting his show and talking to the late
00:16:29.600 steve jobs and having it come out like an absolutely accurate podcast we've heard the likes of npr you
00:16:35.360 know creating a podcast 100 using chat gpt and ai voice generators you know we also see tremendous
00:16:42.360 good so this is something i haven't touched on with voice cloning but it's not only the fact you
00:16:46.900 can clone a voice you can have that cloned voice speak the way you'd like it to but you can have
00:16:51.840 that cloned voice speak up to 27 different languages accurately and in the same voice as
00:16:57.820 the native speaker which is incredible so we'll have a global audience no matter what language
00:17:02.780 we speak the only caveat i have with that that concerns me a little bit is i i do worry and i
00:17:09.980 suppose you know you could say well i speak one of the most popular languages in the world so why
00:17:13.440 should i worry about it but i do worry that some dialects and languages will get lost when the big
00:17:18.720 AI companies look at, well, we're going to put these into our models and train them, you know,
00:17:22.960 but what about the African dialects? What about the different dialects of Arabic? How is that
00:17:27.460 going to play out? And are we going to have our cultures flattened by AI? Again, slightly touching
00:17:32.420 on ethics as well. There's so much to think about. Yeah, so interesting. I'd love for you to expand
00:17:37.240 on some of the ethical concerns that you mentioned earlier. What should we be paying attention to in
00:17:41.960 audio? Yeah, it's huge. Obviously, we've seen this kind of consideration with the image generators
00:17:46.780 and music generators and artists and creatives, you know, worrying that their copyright is being
00:17:52.640 infringed or being used to train the model. We've seen the big case with the New York Times and
00:17:57.280 OpenAI slash Microsoft and things like that. In the audio space, particularly, I see how easy it
00:18:03.500 is to clone any voice without restriction. And even if the big companies like OpenAI,
00:18:10.420 Eleven Labs, even if they come in and say, we're going to put in things in place, and they are
00:18:15.340 saying that you know 11 labs are saying you know we're going to put more and more things in place
00:18:19.380 if you want to do a professional voice clone on their platform you have to record quite a lot of
00:18:24.100 audio and then you have to do what's known as a voice capture so similar to when you're clicking
00:18:29.040 traffic lights on when you're trying to log in somewhere but it's with your voice so that's a
00:18:33.300 step in a good direction companies like resemble for instance have really good deep fake audio
00:18:38.600 detectors where they're already trying to put technology i know 11 labs are also doing this
00:18:43.460 that will detect, you know, was this audio created by AI? The only issue with that, as far as I'm
00:18:50.740 aware so far, is, you know, you look at something like Eleven Labs, they will detect whether it was
00:18:55.140 created using their platform. So it could be generated using something else and it wouldn't
00:19:00.200 be picked up by that. But then we look over at the other side of the open source side of things
00:19:04.140 where, you know, someone can just download some code from GitHub and run it on their own server,
00:19:10.380 providing they've got enough processing power, and clone their own voice of their own accord.
00:19:15.300 So there's no stopping that if someone knows what they're doing. And this has happened. You know,
00:19:20.980 I've seen this certainly here in the UK. And I know it's happened in the US with obviously 2024
00:19:25.600 being an election year. You know, we've had even before 2024 deep fakes, particularly of the
00:19:31.880 opposition party leader here in the UK. He has been quite seriously deep faked on a couple of
00:19:36.720 occasions one of which actually i was asked to investigate and listen to the audio and try and
00:19:41.240 figure it out how it was faked and it's incredible because it got to the stage where it was very very
00:19:48.140 hard to tell and you know i listened to it and eventually there were a couple of giveaway telltale
00:19:54.080 signs like the way the mixing was done in particular with the deep faked voice and the
00:19:58.900 background sound effects but had the creator have done a couple of other steps you know they could
00:20:04.760 have easily fooled most people you know even you know me with an ear for audio we are definitely
00:20:10.260 getting to the stage where unless those big companies are putting watermarks or some kind
00:20:15.040 of way of detecting it i think we're already at a stage where you look at chat gpt where we just
00:20:20.680 don't know what's real or not anymore so we had this cutoff date it was somewhere in late 2022
00:20:26.640 where we knew most things on the internet were written by humans certainly in you know forums
00:20:33.280 and on Twitter or now X.
00:20:35.940 But now we're at a stage
00:20:37.200 where we don't know what's real and what's not.
00:20:39.080 It's so weird.
00:20:40.860 Well, thank you so much for joining me today, Mike.
00:20:42.720 Do you have any final thoughts that you'd like to share?
00:20:45.140 Yeah, so I essentially end by saying that,
00:20:49.420 you know, the future is unknown,
00:20:51.620 but you can do something about it now.
00:20:53.560 And if you are scared about where we're going
00:20:56.120 and what AI is going to do
00:20:58.020 and what it's going to mean for you
00:20:59.660 listening to this show right now as a creative,
00:21:02.660 I would just encourage you to learn as much as you can about AI.
00:21:06.400 Hopefully, as you said earlier, Corinne, it's going to take away all the boring bits of
00:21:11.120 what we do so we can just have more fun because there is a phrase, and it's been echoing through
00:21:16.840 my mind recently, when you do what you love, you never work a day in your life.
00:21:21.560 And I think the ultimate aim of us, I think this is why we're here in the creative industry,
00:21:27.500 because we love to create things, to do things.
00:21:30.220 we have a spark for life and generating ideas and and inspiring other people i mean that's a big part
00:21:35.360 of my why i love to see people inspired and i don't think that's ever going to go away because
00:21:41.060 while ai can definitely give you some great thoughts great ideas you can brainstorm with it
00:21:45.300 i hope that the one thing that will remain if ai does all of our jobs so literally we wake up and
00:21:52.220 say shall i go to the beach today shall i do this shall i skydive today should i do that you know
00:21:56.580 because it's everything sorted and we've got a universal basic income i would love to think that
00:22:01.540 if we are a creative and we're you know to use the word an influencer so people listen to us
00:22:07.400 hopefully that human element will remain of you know well that ai has given me a great idea
00:22:13.540 but i found it really interesting listening to mike because i know he's a human and i know he
00:22:18.420 shares my my journey in this universe in this space that we live right now as another fellow
00:22:24.080 human and as much as code can really get right into our brains and give us the answers we may
00:22:28.880 need you know humans i i think maybe i should just end the word connection it's important right
00:22:36.240 mike's vision for a world enhanced by ai is utopian it's a place where we can play create 0.87
00:22:43.720 and make our dreams reality it's also somewhere that we've figured out how to eradicate the kafka 1.00
00:22:49.280 side of modern life. This dream of comfort and free time was one that was coveted by many during 0.99
00:22:55.860 and after the Industrial Revolution. The editors of Time magazine wrote in 1966 that, quote,
00:23:02.300 in the year 2000, according to one estimate, only 10% of the population will be working,
00:23:07.840 and the rest will in effect have to be paid to be idle, end quote. Obviously, mass idleness,
00:23:13.040 as they put it, did not come to pass. As much as it changed our lives dramatically,
00:23:18.060 automation didn't deliver the leisure utopia that was promised. It's anyone's guess what
00:23:23.600 will happen in this revolution. Is it too much to ask to be once again hopeful about this new
00:23:29.040 type of automation, despite what has come before? Thank you so much for joining me today. I've been
00:23:35.000 your host, Corinne Leinstari. It would mean so much if you would follow the show on the podcast
00:23:40.180 app of your choice. It would be extra fantastic if you could leave us a review. If you're on
00:23:45.580 Spotify, we're going to end each show with a question for you. You'll find it in the Q&A
00:23:49.980 section on the app, just under the episode. This week's question is, do you think AI will make our
00:23:55.800 lives more leisurely? If so, what would be the most important thing it could automate? See you
00:24:01.440 next time. Bye for now. Saying how you feel out loud is terrifying, but that's what Hinge's new
00:24:11.780 free audiobook, No Ordinary Love, is all about. It's a collection of real love stories about five
00:24:19.760 couples that met on Hinge. The stories are written and read by Tembi Dinton Hurst, Nicola Dinan,
00:24:25.440 Curtis Garner, Raven Smith, and Rufy Thorpe. I, Hunter Harris, wrote and read the foreword.
00:24:31.780 The audiobook's out now. Give it a listen.