Interview with Martin Clancy – Part 2
by Killian Laher

Martin Clancy’s book, Artificial Intelligence and Music Ecosystem is out now.

“Artificial Intelligence and Music Ecosystem highlights the opportunities and rewards associated with the application of AI in the creative arts.

Featuring an array of voices, including interviews with Jacques Attali, Holly Herndon and Scott Cohen, this book offers interdisciplinary approaches to pressing ethical and technical questions associated with AI.”

How does all this work from the artist’s point of view, between this and sponsorship and crowdfunding?

Martin Clancy (MC):  There’s no limit. I was thinking about back in the eighties when everybody was broke, the idea of a concert being sponsored, or a band being sponsored, was like, you’ve got to be kidding! That was inconceivable. Now we move along a little bit and some things have changed. The idea of VIP guests, special access and similar stuff. If an artist has a loyal fan base and certain parts of that are themselves financially successful and wish to support the artist, then it’s almost like a form of patronage. Their way of supporting that musical artist is similar to the way that traditionally fine art painters were supported. You’d buy a painting because you knew that’s how the artist was going to pay their rent or mortgage for the next year. I wouldn’t be sniffy about someone spending 350 euros on a VIP ticket, but it’s easy enough to imagine it going beyond that, to a stage where there’s a ‘special guest’ and up pops a VIP audience avatar to play the lead solo with the band! I could see that happening. Who knows, we may have a new business idea there Killian if you’re interested!

But seriously If you heard that that was happening, you might draw a line and go, does Bruce Springsteen really need that extra five grand? But moving down the musical food chain there are a lot of artists that such innovations could make a profound difference for. Again, if it’s the artist making the decision, then that’s their relationship with their audience to figure out. However, if it’s the estate, as with Roy Orbison, you might go well, hang on a second here, Roy is not around to say whether that’s a good idea or a bad idea.

Where does the whole idea of copyright, and ownership come into all of this?

MC: Music has been made on for a quite some time. 40,000 years ago, a bone flute was discovered. That’s the oldest piece of music technology that we know of. Some argue that music predates language. So the behaviours of making music prior to the invention of recorded music have been going on for some time. For 40,000 years or so we were making music in a certain kind of way. Then there’s this little blip and this blip occurs that grows with the growth of copyright. We get to the 20th century and the industrialization of music recording and you end up with four areas that you can make money from and from that what we call the music industry. You can write a song – that’s publishing. You can record the song, that’s the Recording Industry. You can take that song and play it live, that’s Live Performance, and then you can make a T-shirt that somehow relates to the person who’s playing that song and sell it, and that’s merchandise. There are four areas, and they’re all based on music copyright so copyright is intrinsic to the existing business. What do we know about copyright? Well, copyright like all expressions of intellectual property is exclusive to humans. That might sound self-evident, but for instance, there is a famous case where a monkey took a photograph known as ‘the Monkey Selfie’. Here a photographer David Slater who was on holiday in Indonesia and a monkey appeared, touched the camera, and took a selfie. He (Slater) published the photographs and PETA the animal rights organisation sued Slater and the case ended up in the US Supreme Court. The Supreme Court ruled that while a monkey can take a photograph- it cannot create a work of art. Clearly, it can! However, we will not allow copyright to be given to a monkey or anything that’s not human. That’s how AI is legally regarded as being non-human.

But there are lots of reasons, and I talk about some of this, of why certain interests outside of the creative arts can benefit from copyright being removed. And it’s not completely sinister. Some of it is sinister, and some of it’s not. For instance, like in the book, I argue that we should be careful about changes to copyright. The law needs to be adapted, but whatever its flaws are, copyright principles are how everybody in music generally struggles to get by. Scott Cohen, who’s an Elon Musk type figure in the music industry, Chief Innovation Officer at Warner music, a powerful, influential person, makes a strong argument to say the opposite. AI should be able to have the right to make an intellectual property because it drives innovation. Who’s to know? I did five interviews with him and even though we came up with different approaches, our passion and commitment to employment in music was the same. We just find different routes to the same goal.

Now that’s very different to what potentially is occurring with many of the open source products that are coming out. With these new open software tools, what’s happening is – if it’s open-source and there are no new copyrights attached to the new creations that are despite the fact that AI data sets are, trained on existing copyrights. They can argue, “we’re not commercially selling these creations. It’s free.” So what does that mean? It’s similar to the legal notion of fair use. Positioning themselves similarly to an educational institution. “We’re just using it for fair use, as you would in university research” And what about your share prices? How come your share prices are going up, and you just received a $100 million investment based on the popularity of your model? So the point is that you have to lift the veil, as they say in law, and consider what is going on behind the scenes. So you, the user, may think, that’s great that you’ve given us all these tools that everybody can now be a musician or a designer. That’s fantastic. But you, the corporation might become a billionaire off the back of that, and that’s not so cool.

I don’t think we’re ready to replace artists with AI or anything like that, but at the same time, if you’d asked yourself, 30, 40 years ago, would people go and watch a DJ?

MC: When I was growing up one of my favourites Kraftwerk had already peaked. There were those electronic two-piece bands, and you had all the rage at Top of the Pops of people saying ‘these aren’t real musicians’. Then you had purists like Queen saying there are no synthesisers on this record. So what’s different about this (AI) then? Kraftwerk had a vision of the future that was about as real as David Bowie’s Ziggy Stardust. They were just different possibilities. We’re now looking at technologies that, because of the speed of change, you just have to be kind of careful with. When you talk about DJ culture, it was a very long time before scratching was invented in the early seventies. The gramophone was invented, 100 years beforehand and it took 100 years for someone to go and do scratching. For it to become something that was mainstream took that long!

I’m talking to you about technologies that no one saw coming five months ago, and research papers are rapidly coming out. I’m a full-time research fellow, my gig is to kind of be on top of this, and I’m looking and people in the field and we cannot keep up with what is going on. It’s wild. It may settle down again, but right now it’s like literally stuff is coming out every week. I’ll give you an example. Chat GTP, within the first 5 or 6 days it was downloaded a million times. This is scale, I went to get it, to use it yesterday, and I’m on a waiting list for it. So there are a million people who got there before me. It is the smartest conversational chatbot that’s ever been created, capable of very impressive conversations, and writing. If you have a conversation with the chatbot, it probably makes a lot more sense than a lot of what I’ve told you! Seriously, it’s wild.

It’s clear it’s not just going to affect music. I mean, potentially, obviously, you were talking about the text-to-image, which will affect the visual arts.

MC: Already has. Greg Rutkowski does these beautiful designs, featuring dragons from sword and sorcery/Lord of the Rings type design,, very detailed paintings. Rutkowski became the most (text to image generator) requested ‘in the style’ of artist. 93,000 times his name was used as a prompt to generate artwork. And he’s living, a Polish painter.

So the most successful artist means most successfully used artist. If you think about it, it means that the success of his work makes him unemployed. If someone goes to number one in music, they’re the best-selling artist. If you’re Drake, the number one artist, or Bad Bunny for the last couple of years, you’re making money from streaming. Not very many others are. Only about half a percent of recording artists make money, but they make a lot of money from streaming. But imagine that in this case, this artist, Rutkowski. He’s the most successful text-to-image generation artist. Currently, in the last month, 93,000 people have used his name as an AI prompt and made works of art that look similar to his. Except for one caveat. Unlike Drake or Bad Bunny, he’s not getting any money from this activity. He’s just famous. Does that make sense?

There’s just so much to think about. It’s limitless.

MC: One more thing, Open AI. It was founded by Elon Musk, though he’s no longer involved in it. If you go into Jukebox on SoundCloud, Open AI Jukebox, you’ll hear recordings they did two years ago where you have Frank Sinatra singing Christmas songs. Or U2. Or David Bowie. Not samples, just the machine without humans recreating songs in that style. If you listen to them they sound a bit dodgy and a bit rough.

The music is from two years ago, but OpenAI are the people who made DALLE-2. They’re the people who made ChatGPT. One company making all these different products and they’re all free to use. Make of it what you will.

After we spoke to Martin the first music-to-text company appeared, Riffusion https://www.riffusion.com.

Martin Clancy’s book, Artificial Intelligence and Music Ecosystem is out now.

Martin Clancy’s book, Artificial Intelligence and Music Ecosystem is out now.