Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | RSS
Produced by Joseph Cottrell, Wayne Hall, Ken Fuller and Jeffrey Crecelius
This week we have a bumper, epic episode where Simon Barrow, Mark and I discuss AI and its impact, in the context of Yes music, of course. My daughter, Charlotte, also gives us her thoughts on the subject so many thanks to her. This topic was suggested by the appearance on YouTube of an album entitled ‘YEP – On The Waterline (70’s Prog Rock / Progressive Rock)’ as spotted by Charlie Nolan.
- Should AI be used in music?
- What impact could it have?
- What about the future for Yes and all other music artists?
See if you agree with us and let us know by leaving a comment below.
Let us know if you agree with us!
https://youtu.be/7uECawjO2iM?si=j7QoQVkiXIfrPwPD
Yes – The Tormato Story
Available now!
YMP Patrons:
Producers:
- Joseph Cottrell
- Wayne Hall
- Ken Fuller
- Jeffrey Crecelius
Patrons:
| Aaron Steelman | Lind |
| Al Dell’Angelo | Lobate Scarp |
| Barry Gorsky | Mark Baggs |
| Bill Whittaker | Mark James Lang |
| Bob Martilotta | Mark Slater |
| Brian Harris | Martin Kjellberg |
| Brian Sullivan | Michael Handerhan |
| Chris Bandini | Michael O’Connor |
| Craig Estenes | Miguel Falcão |
| Dave Owen | Paul Hailes |
| David | Paul Tomei |
| David Heyden | Rachel Hadaway |
| David Pannell | Robert Nasir |
| David Watkinson | Robert Vandiver |
| Declan Logue | Ronnie Neeley |
| Dem | Scott Colombo |
| Doug Curran | Simon Barrow |
| Fergus Cubbage | Stephen Lambe |
| Fred Barringer | Steve Dill |
| Gary Betts | Steve Luzietti |
| Geoff Bailie | Steve Perry |
| Geoffrey Mason | Steve Rode |
| Guy DeRome | Steve Scott |
| Henrik Antonsson | Steven Roehr |
| Hogne Bø Pettersen | Terence Sadler |
| Todd Dudley | |
| John Cowan | John Thomson |
| John Holden | John Viola |
| Jamie McQuinn | Tim Stannard |
30 replies on “AI killed the prog star? – 676”
I haven’t listened yet (will do so on my drive tonight) but I predict this subject will prompt the busiest comments section to date. And it will be a great discussion so long as it doesn’t deteriorate into a slanging match between AI haters and embracers (but the five percenters wouldn’t do that, would they?).
Thanks for including my Game answers.
FYI, Trevor Rabin contributed to “Merry Axemas, Vol. 2 (More Guitars For Christmas),” which gets plenty of plays in my house during the Christmas season.
If nothing else, it’s a great title!
Hi Kevin and Mark,
as a small (and only semi-pro) musician I feel I need to react to Mark’s comments on “put down the game controller and learn something”. As a father of two children on the spectrum it has become virtually impossible to create meaningful music since my last album “Alter Ego’ in 2019. Yes, I’ve chosen to be a father next to my day time job and my musical creativity, but I never realized it would be this hard. Only things I managed since 2019 are some remixes of old stuff and some dabs into Vaporwave.
I do think some people don’t have the choice to become musicians or to release records, not because of laziness but because of other things in life.
Does this seduce me into trying out AI to create some music? So far, no. I’m very sceptical towards AI and I’m not a fan of AI meddling in the creative arts. But I do often find myself wondering if I could use AI to make creating the ideas I do have in my head easier. I would want to keep full control over all lyrics I write, melody lines and chord progressions, but maybe AI could demo some material for me to see if what I had in my head sounds cool. Then, if I feel a song is good enough to be released I’d probably want to sing and play some things myself and other things like guitars and drums by friends, cause I’m not good enough to do these things by myself. AI would have been involved then, but lyrics, melody lines and chords would be mine and there would be no AI audible on the finished product. Would that still count as a creative work?
Thoughts?
Hi Joost. I can sympathise with your home situation. I also have two autistic children (and one who is not) and I didn’t do any composing for 30 years. I now do use AI to help me hear what my compositions will sound like, as mentioned by Charlotte in the episode. However, I would only ever use it to create demos of my music, as you have said.
Two of my children now live away and the third is at university so I could finally spent last year writing 52 choir anthems using score writing software (Musescore) because I cannot play the piano. I also spent many hours trying to find software that would let me hear what the vocal parts would sound like with lyrics. None of the software I found could do that but very recently a new AI tool called ACE Studio appeared that allows you to convert midi files into vocals with the lyrics synthesised by the AI voices. It is a tool and nothing more. It is not ‘generative’ AI and I think that’s a distinction we didn’t mention in the episode. I have no need for AI to help me come up with ideas – I only use it to help me hear what a choir would sound like singing my music, as you mention above.
The AI I do have problems with is the generative software that lets you input written prompts and then writes the music for you. I would see that as dishonest – quite apart from the fact that generative AI generally uses other people’s music (often without permission) to ‘learn’ how to write music. This type of AI presents the existential threat to the creative industries, for me.
Hi fellows ! Cheers from sunny ( and smokey ) Montreal. I Listened to your conversation about AI created music, and it made me think of something adjacent. What about AI created “musicians” ? Say you are an artist, and you write a song. You’ve layed down guitar, drum and vocal tracks, all real instruments and voices, but you really crave a bass guitar track in the style of Chris Squire ? Now sadly, Chris has departed this earth, but what if you ask AI to generate a bass line with the sound and style of Chris for your project ? Interesting idea, right ? But how ethical is it artistically ? You’ve not hired anybody, but you’ve got something you like, and the song is still your original creation, yet in a way you’ve “cheated”. Take this one step farther. Say the record company that owns the rights ( and master tapes ) for the ABWH album want to re-release it with a remix where they replace all of Tony Levin’s bass parts with AI created ”Chris Squire” ones ? They then want to brand it as Yes to stimulate new sales and interest. This could be an interesting discussion.
Thanks for the comment, Steve. Personally, I couldn’t include anything generated by AI in my own music and what you suggest above would make me avoid the resulting music. Of course, this kind of thing will be done – and probably already is being done – without any need for the producers to label it as such – I think there needs to be legislation to prevent this – particularly as the ‘fake’ lines will probably have been made by generative AI software using ‘real’, copyright music as its model.
“I think there needs to be legislation to prevent this – particularly as the ‘fake’ lines will probably have been made by generative AI software using ‘real’, copyright music as its model”
But Kevin, using ‘real’ copyright as well as public domain music as a model is exactly what you, Mark, Paul McCartney and every other composer has used. If it’s OK for you to draw on the rich heritage of Western Music from gregorian chant, through Bach, Beethoven, Wagner, the Beatles, Yes, Taylor Swift and the rest, why shouldn’t AI? If it’s a question of scale, where do we draw the line? Only people who’ve been able to experience, say 10,000 tunes should be allowed to compose? 10 tunes, 10 million tunes?
Great episode. You covered a lot and I won’t rehash it. I just wanted to give my opinion on backing tracks. By this I mean using pre-recorded tracks during a live performance. In general, I’m opposed to it. I say, you should play all the parts live. Even if what you end up with is not like the studio version. I love it when musicians deviate from the studio versions. If I wanted to hear an exact copy of the studio version, I’d stay home and save the money. To me, that’s where the creativity comes in; making something new and different that is unique every night.
I will grant some exemptions. Using some some effects like the nature sounds at the beginning of Close to the Edge: acceptable. If a band is really 2 ppl recording all the parts then when they play live they are going to have to either add exta musicians (like Steven Wilson does) or use pre-recorded tracks. This is OK. What is completely unacceptable is lip-synching; giving the impression that someone is singing or playing the instrument when they’re not. I don’t think many ppl would disagree with me on that.
Also, thanks to Charlotte for her passionate remarks. It’s nice to hear a young person’s opinion here.
Thanks Guy. The backing tapes issue always makes me think of Madonna’s recent tour where she sang live but all the instrumental music was on backing tapes. I wouldn’t pay (a huge amount of) money to go and see a karaoke concert. She didn’t try to hide the situation but her justification that the backing tracks were the original mixes of the records didn’t make any difference to me.
Hi – a fascinating discussion in this episode. I have 2 points:
1. Whilst I have a LOT of sympathy for Mark Anthony’s position i kept thinking “but Jon Anderson in the early 70’s was not musically adept in terms of expressing himself. ”
He was full of ideas which he tried to communicate to other members of Yes but his lack of technical facility meant the other shad to take their best guess at what the phrase, chord, melody or rhythm actually sounded like. Bill Bruford has spoken about this in the past. IT seems to me that, had Jon been able to harness AI to shape his ideas into what was in his head we would have possibly some very different Yes tunes albeit with titles we recognise. And I think they would STILL be as amazing, beautiful, rich as the songs we have now. I’m not saying I’mn right but it feels as though Jon is a perfect example of someone brimming with musicality and absolutely bereft of the technical skills to experess them. Had he decided to banish himself to the bedroom to spend 3 hours a day working on his guitar / piano technique he would never have met Chris – instead he got out there and sung, hummed and mouthed his ideas to players able to articulate his ideas (at least partially).
2. I think where AI would SO benefit the fans is in remixing bootlegs. As I understand it Peter Jackson[s company owns his own AI software called MAL. His staff fed every Beatles tune into MAL so it could “learn” the band’s music. Then they were able to remix the early recordnings to clean them up AND part of that was demonstrated in detail on “Now and then” where MAL was able to pull John ‘s vocals out, fremove the hum that had caused George to call an end to the session and make the vocals pristine. Imagine if (when?) MAL is commercially available to studios, just what could be done with bootlegs. If a version were made available for individuals…well, we have a motherlode of boots at a certain Yes-related website..imagine being able to clean up thise TFTO and Relayer shows!
Ok must stop there before my inagination starts making me type silly stuff!
Thanks Paul. Yes, I imagine young Jon + AI tools could have been interesting…
The thing about the Peter Jackson approach for me is that he is using it to enhance what is there, not to generate new elements. That’s fine in my opinion.
Faciniating discussion! I tend to agree with Simon (and Charlotte), in that there could be some benefits to AI as a tool, but NOT to replace the creative process or to generate music for publication.
As a (fun?) exercise, I used a free program (Udio) to create two wholly-generated AI pieces (unlike ‘Yep’ which had human intervention), using the following prompt: “Progressive rock with complex time signatures, intricate harmonies, and ethereal lyrics in the style of Yes”. I am attaching the results (two 1 minute clips). Interestingly, one could argue that the ‘word salad’ lyrics aren’t all that different from some of Jon Anderson’s creations, lol. I’ll let you be the judge as to the ‘quality’ This is the first of two clips (it only lets me do one attachement at a time).
Clip #2 attached (see my other comment)
Thanks Bob – fascinating to hear these two examples. For me, they are perfect examples of the horror of generative AI. How did the AI ‘learn’ to make this pseudo-prog music? It used real, human-created music to mimic a kind of ‘average’. How many of the creators of the material used to train the AI had given their permission for its use – and how many have been financially compensated for the use of their intellectual property? Even if, technically, the artists had given their permission via some carefully-hidden clause in a streaming contract, how ethical can this be?
When it comes to setlists my long expressed desire to see a Yesterdays / Relayer tour come to pass is probably only surpassed by an Heaven & Earth / Keystudio coupling for lack of box office sizzle!
Would love this line up to tackle the first two albums or at very least the debut. Won’t happen I don’t suppose.
As for the AI thing, it’s probably going to be much more about producers of music (and mainly non pros at that) than consumers.
To my mind it is just a further extension of what has been the direction of travel since the Mellotron, passing via the Fairlight, the TR808, the Akai S1000 and other innovative gear-oriented staging points along the way. All those were about ultimately making sounds of every shape size and duration available to the creative musician. It’s not a threat to creativity. It’ll be an aide to some and an irrelevance to everyone else. There were after all people who thought the church / pipe organ would put orchestras out of business. Not so much as it turned out.
Unusually (possibly) for a progressive music fan of 50+ years standing I have never got bent out of shape over sampling and don’t much care if people can play well or not if the end result is original and pleasing. Whatever gets the idea into the world is fine with me whether that is “96 Tears” or the Ring Cycle.
After that it is all about the ears of the beholder and, just like food, those who just want to feel full might be happy with a partly AI delivered musical diet, the rest, looking for a truly enriching experience will hold out for the real thing however we define “real”. Listeners might even start to value it even more than they do now and (again like food) might think a bit more deeply about how the music they love is made and marketed, for whose gain and at whose expense.
Thanks Ian – great to see your (partially) different points of view on AI. For me, the key includes the legality, the ethics and the potential further reduction of opportunities for real artists. Just like not crediting session musicians (with which I wholeheartedly disagree) hidden use of generative AI is dangerous, I believe.
I hear you and I tend to flip back and forth. Existing copyright laws protect ideas (and ideas that can be notated or otherwise preserved in a written form) more than they protect performers. This much is true.
That said, legally speaking I don’t think there is any real need to expand on what we have now. If we are talking about AI sampling real musician’s sounds and using those to generate new machine-tooled music then we have laws and processes to protect against that too. it then becomes all about the publishers and record companies having the will to enforce those protections.
When AI can recreate an all day argument between a 5 piece band as to whether the next note should be B or B flat then I might start to worry!
Gallow’s humour aside, it would be a much safer world for creatives if composers and musicians could behave more like screen writers, unionise properly and bring collective power and collective bargaining to the table. If you are serious go on strike and stop giving permission for your work to be used under conditions that you know are not serving you.
Sadly most of the union people I meet still have their thinking rooted more in 2000 rather than 2050. They simply do not understand the market and have a sensibility rooted in the worlds of pit bands and orchestras rather than at the frontiers of music making.
The unionisation also doesn’t happen because artists are in many (I would say most) cases competitive beings and are encouraged to be self-interested, have sharp elbows and often merciless when it comes to the rights of peers and band mates. This often at the behest of advisors who have calculated whose self-interest delivers the biggest bucks at 15% or 20%. I would recommend that musicians spend more time learning the industry and less time worrying about whether the bass player has earned his or her 15% of the writing credits on your latest non-hit.
As for session musicians, they should be the best protected of all and in terms of run-of-the-mill rate card money for work measurable by the hour they are fairly well looked after (better than lets say most therapists or physios but not by any means generously). However when it comes to credit it beggars belief that the MU and (as far as I know the AFM too) do not insist on publically listed credits for everyone who plays on every session. Not easy to enforce but there is plenty else listed in digital music metadata that is of little interest to non professionals.
So listening to the latest podcast, little bit frustrating because I work in AI for travel companies and it has some good points about it so for example just understanding the raft of information you need in today’s society. It really helps us a good assistant and I think there’s nothing wrong with really using AI to get some ideas to inspire ideas for music. punk was fabulous because it enabled people who didn’t have music skills to really express themselves and actually there’s some great punk albums and songs out there and those and those artists who started with very basic punk went on to be very strong artist moving forward take the clash take the pistols moving to pill you know there are a lot of good things come out of being aided and supported by things like AI. You really don’t need to be a top musician to do something creative. Think Eno. He has computer generated music and allows computers to evolve music. I have bought a lot of Eno’s music. You don’t need to be a top author to write something creative. You just need good ideas and very often as you know writers who write biographies have a ghost writer and they do the heavy lifting and ask the right questions and nonetheless the quality of the book that comes out still captures the essence of that person who it’s written about.So in conclusion I think AI is here it’s not gonna go away. I don’t think you should be negative about it. You should look at ways in which AI can support you. I use it every day in my work and we assist travel companies in getting started with AI because if they don’t embrace it today, they might not be here tomorrow just my take but IMHO there is not an issue with AI in music. After all you don’t have to listen to the AI music just as I don’t listen to lift music.
Amen to all that. Seems to me the places to really worry about AI is in the body politic and the areas related to criminality and personal privacy not in the arts.
I grew up a fan of Bowie and Roxy who moved into exploring progressive music in the mid 70s but at 15/16 loved what Punk and Post Punk offered in terms of freedom and the licence to be creative before you had a monster skill set on your instrument. All those barriers were instantly erased.
It is significant to me that Fripp in particular could see possibilities that his peers could not. His early solo work is full of Punk and Post Punk’s influence and has an umbilical connection to the best of that scene and also to the 60s and 70s Art Rock that made Post Punk what it was in the first place.
For me in post 70s progressive rock there was an awful lot of hiding behind technical facility and the utilisation of a handful of comforting clichés when it came to arranging and dynamics. There was a new brevity for sure (mainly I think for American radio’s benefit and the commercial advantages therein) but the days of real innovation were all but over. Dial some future-shock or quiet desperation lyrics into a familiar formula and you were golden. That was really just humans doing what AI might do with the same recipe and a lot slower.
Great compositional contributions that you can stand next to the classics of the late 60s and early 70s? Not so much.
My list of must-own Prog albums from the last 45 years will run dry at well under 20 selections. No wonder Prog magazine has over time expanded its remit to include almost anything that isn’t pure pop or from out of a roots genre. Even then Folk Rock often gets invited into the Prog tent.
Those of us of a certain age will remember a lot of musicians in the bigger bands cutting their hair and narrowing their trousers in 77 / 78 out of career necessity but the music became much more conformist. They abandoned all that was original about their earlier work, stopped developing and missed the point of Punk completely as a door to creative freedom.
Amazingly there are people in prog bands today (especially British ones) who were more or less around at the time who the best part of 50 years on still seem deeply bitter about it. Of course they almost to a man confess their love of the likes of Blue Nile and Talk Talk but they can’t quite get their heads around the idea that their multiples of 10,000 hours of dedication have absolutely no value if the ideas are not there to make good use of them.
AI meanwhile will effectively have tens of millions of hours of recorded human creativity to draw on and yet its chances of coming up with something that stands the test of time and truly thrills and moves us really is monkeys-and-typewriters territory.
Of course a lot of the most interesting Post Punk artists were progressive fans who took the best lessons from their predecessors in terms of making adventurous choices and left all that was inauthentic or irrelevant on the cutting room floor. The Police, Magazine and Japan are just three, very different but significant bands that fell into that category. That’s the lesson here I think. Be the thing creatively that AI can never be. Yourself.
Yes, as you say, Jon, “using AI to get some ideas to inspire ideas for music” – I couldn’t agree more. AI tools for helping human creativity are great – but not for replacing human creativity.
Hi Kevin and Mark
What an excellent thought provoking episode! I really enjoyed reading the spectrum of opinion and thoughts of fellows listeners in response to your discussion.
I personally am deeply suspicious of AI. So I sit on Marks side of the fence. Simon argued persuasively that AI ‘democracizes’ music so that people without the ability to learn or perform music on a musical instrument would be able to produce some.
Is the ability to produce music a natural human right, or is it a talent and the result of endeavour, training and self discipline? Why should everybody have the ability to produce music?
Should everybody be allowed to drive a car (unless they earn a licence)? Should everybody be allowed to enter the profession of their choice without first gaining the necessary qualifications?
We all have our own innate strengths and talents and creativity. AI is no substitute for that.
Out of curiosity after the episode I listened to a track from that online recording ‘Yep’. It was really bad, almost elevator music.
Mark, do not hang up your guitar just yet.
Kevin’s daughter spoke really well. Congratulations to her and wishing her well in her studies…
Thank you Alan. I think we can all agree that a system of licensing for the right to create music is an avenue we should explore 😉
I agree 100% with both Mark and Charlotte.
I’m actually using AI right now and stopped to write this comment—I’m using it to split apart a song into its stems so that I can listen more intently to the guitar part to learn it. Then, I can do the opposite and REMOVE the guitar from the song and jam along myself for fun, like I’m a part of the band. I totally get the concern about generative AI, but using it as a learning tool in this way is great!
Hi Ronnie. What software are you using to split the stems? I have an account with one of the AI engines but it gives mixed results
Hi Ronnie,
AI used in this way….I support 100%.
I remember back when I was younger..pulling my hair out trying to figure out the guitar parts for Metallica’s – Master of Puppets…or various Rush songs.
This would have been great to have back them.
Interesting episode especially enjoyed the AI discussion. Charlotte is a broadcaster in the making for sure, very good input from her. Regarding the Fragile tour, as is often been stated, the reason it’s called the music business is because it is a business. Wrapping the tour around a well known album (maybe more significant in America as this was the album that broke through for them) potentially puts bums on seats! Seeking a tour on the back of Mirror to the Sky whilst would please the 5% it would potentially leave 95% of the seats empty. I agree that of all albums it’s an odd one given the ‘solo’ tracks but it’s only 40 minutes of what we hope is a two hour show.
Whilst I can understand Mark’s anger at having invested years learning to play instruments – Tough! Technology has frequently made the craftsman redundant. The successful craftsmen are those who adapt to work with it rather than stamp their feet screaming “It can’t do what I spent years learning” (it can).
The same goes for the arguments about learning music theory and composition. They are just a set of rules which, like most, if not all, rules, are built to retro-fit an already established concensus. in the 16th century no one sat down and wrote the rules of counterpoint and then wrote music to fit those rules. Various contributors analysed the combinations of notes that produced certain effects and noted common features from which they could derive rules. As time progressed more rules were added and eventually it became so restrictive that forms of music strated to break away from it (and form their own rules).
Generally, where rules come first, the music is intellectually interesting but emotionally bereft (eg twelve tone music).
However, if AI is composing “pop”, this will be based on “rules” developed over the last century or so, exactly the same as Mark is.
To that extent the result is the same – a solution to a quadractic equation is the same whether produced by hand by a mathematician or a scientific calculator. I’m sure no-one would argue that the mathem,atician’s answer is “better”.
The difference is the human element – what Mark decides to go with and what he decides to leave out. Only the listener will be able to decide which is better, I suspect (but don’t know) most would prefer the one with the human decision.
But in case this all sounds too bleak to those unwilling to embrace AI composition (and my gut says that I may be one of them) I do forsee one positive outcome. As technology pushes all the arts towards some sort of homogonised result (all pop songs sounding very similar, all action films looking the same) I predict a resurgence in the popularity of live performance – only here can we experience a true “one off”, never to be repeated series of moments. The only thing holding this back is the comparatively high cost of putting on any sort of production when compared to producing an album of music or an AI generated film on a PC.
I find all these comments fascinating. Thanks to everyone who has left a comment. Almost 30 comments! I had a feeling this topic would generate some conversations…but this much…wow!!
So ..I have no problem with technology…in fact a lot of stuff that producers..mixing and mastering engineers use are computer based technology. While there are studios that use hardware stuff (compressors..eq’s..tape delay)..lots of these studio guys have realized that it’s much easier and convenient to use plug ins of these units. I’m one of those people. But again ..the AI thing just doesn’t appeal to me because it takes away the most creative and rewarding part…which is performing the music…on guitar…bass…drums.. etc. That’s what I love about being a musician/songwriter. Its nice hearing people saying “wow Mark that album is so good…the guitar playing is great.” I can’t imagine the same feeling with AI….”Wow…Mark that’s some great suggesting ideas to a program”. Anyways…we can debate this for hours…but the truth is…only time will tell what happens…people saw doom when Napster came into our lives …we survived that…we will survive AI.