Extract from ABC News
In short:
Screen composers are among the Australian creatives that have told a Senate inquiry they want laws to protect their livelihoods against artificial intelligence.
One expert says AI is being used as a tool to empower more people to create songs, and won't displace musicians.
What's next?
The Australian Guild of Screen Composers is calling for musicians to be compensated for any use of their work by AI companies, and for prior permission to be sought.
As artificial intelligence continues to evolve at lightning pace, humans who've spent decades working on their craft are fighting back against the prospect of being replaced by machines.
Ask screen composer Josh Hogan what he thinks about the rise of AI generated music platforms that can create songs on demand, and the response will probably be scathing.
"The AI just kind of creates this smoothie, grey goo… a thing that sounds like music but it's homogenous," he says.
"It doesn't really have any focus, it doesn't really have much of a hook or a catch to it.
"It doesn't have any of that intention that usually comes from human beings going, 'we want to say this'."
The West Australian composer has been devoted to music for most of his life, performing in bands and orchestras as a drummer and percussionist.
Today, he's known for award-winning soundtracks for film and TV, created alongside collaborator Ned Beckley.
"I'm still a person who cares about music a lot," he says.
"And, that's really why I hope that I'm hired, because people know that I'll put my heart and soul into the act of creating music."
Mr Hogan wants to make clear that he supports new tools that can help more people make music.
But he doesn't buy the pitch by AI music generators that they are democratising music creation by lowering the entry barrier.
"Actually, what it does is it destroys the creative process, it removes the work of making the thing," he says.
"Ultimately, the optimist in me says that humans will only engage with human stories."
"It's a very hard thing for us to get our heads around as a culture, but we're going to have to ask the question, what is the work of making things? What is the work of creativity? And is that something that a human does?"
How does AI create music?
As AI music generators clock up millions of new users, the screen music industry is among the creative sectors desperately trying to get ahead of the curve.
Udio, a New York-based start-up, launched its tool in April.
It has already had close to two million participants, creating songs within seconds by typing text prompts into a box.
Udio's Conor Durkan described it as a "step change in music creation".
"Our goal is to basically empower all sorts of people to expand themselves creatively with music," he says.
Co-founder Andrew Sanchez said the tool was proving popular among people with no musical skills, and professionals making it part of their workflows.
"I was just with some screen composers last weekend and one of the exciting examples was they said 'hey, let's try to generate some music or the mood for when a hand is emerging from a grave in a darkly lit graveyard'," he says.
"And it sent out some great examples. None of those was exactly what the composer wanted, but it gave them a few ideas that they could then take and then build out into the entire thing.
"When we look three years, even two years in the future, we think that people are going to look and say 'this is actually powerful technology that's enabling a whole host of new things, novel creations'."
But in the meantime, the world's biggest record companies are currently suing the developer for Udio, Uncharted Labs, and another startup, Suno, for alleged copyright infringement.
Mr Sanchez said he couldn't go into detail about how Udio's model was trained, due to the on-going litigation, but compared it to human learning.
"Great composers and songwriters develop their ear for music, develop their musical abilities by listening to music," he says.
"And in a similar way, these technologies are able to assimilate abstract ideas about music, the general building blocks you might say, the fundamental pieces of what make up music which can then be created and assembled in these novel and really, really cool ways to make new music."
But that's not an argument that Kingston Anderson, executive director of the Australian Guild of Screen Composers, is about to entertain.
"Without the original music, they can't learn," Mr Anderson says.
"They're breaching copyright because they are using that person's music to generate new music.
"To say 'oh they're just listening', what does that mean?"
"They're not some person sitting in a room listening to a tape."
Push to protect creatives
As the landmark legal case plays out in the US, bodies representing Australia's creative industries have been calling for regulations to protect their original work and livelihoods.
In a strongly-worded submission to a Senate select committee investigating the impact of AI, the Guild of Screen Composers warns there is "a real risk of destroying a whole industry of creatives".
"It's happening in the advertising world already," Mr Anderson says.
"The next stage of that is to generate imagery as well. So, I would suggest to you that in about two or three years' time, most advertising, a lot of advertising will be AI generated."
The guild is calling for composers to be compensated for any use of their work by AI companies and for permission to be sought before it's used.
Alex Jenkins, director of the WA Data Science Innovation Hub at WA's Curtin University agrees that the law needs to catch up.
"We need to discover a way that artists and musicians can be remunerated if their music, if their voice, if their skills are used to train an AI model," Mr Jenkins says.
The challenge will be acting fast enough to make a difference.
"It's moving faster than any technology we've seen since the introduction of the internet," he says.
"So by the time regulation comes forward, by the time regulatory bodies actually sit down and work through these problems, the technology has changed again."
The Senate committee is expected to report back with its findings in September.
Sector grapples with ethics
Even within the fast-growing AI music generating sector itself there are diverse views on the ethics of generative music.
Soundraw, a Tokyo-based company, operates a different model to other platforms in that it's not text-prompt based and the result can't be uploaded to streaming platforms.
Founder Tao Romera says notably its AI has been trained using content created in-house by the company's own team of producers.
Mr Romera believes there is value in an AI tool that makes music production viable for people working on projects with small budgets.
But he believes it's unethical for AI companies to be scraping music that's subject to copyright.
"It feels wrong because you are basically taking all the knowledge and all the effort that has been put in by hundreds of thousands of humans to create that music and then disrupting the industry based off their work."
Future of the industry
As to the extent to which AI will ultimately replace human creatives, no-one seems to know.
Alex Jenkins remains doubtful.
"Really bringing that emotional impact to music, it's such a human thing," he says.
Josh Hogan agrees.
"[AI] is not necessarily going to support the act of creating great music that audiences will connect with, or in our job, creating great music that connects with characters on a screen, for example, or tells a story," he says.
No comments:
Post a Comment