Viewpoint: Sammy Andrews on TikTok and AI

Viewpoint: Sammy Andrews on TikTok and AI

In her latest digital column for Music Week, Deviate Digital CEO Sammy Andrews looks at how the world of TikTok, AI-generated music and more might be opening a Pandora’s box for the music industry...

It’s been an interesting few weeks in the world of TikTok. The news that they entered a copyright battle with ICE isn’t surprising given just how many users they have (reported to be in excess of 500 million monthly active) and just how much music is used (let’s not forget they started life as Musical.ly before being bought by Chinese giant Bytedance in 2017).

Musical.ly shot to fame allowing users to record short videos, set to music. Users started lip-syncing to the tracks and dancing to them and a craze was born that stretched far beyond Asia into the USA and Europe.

In 2018, Bytedance folded Musical.ly into its own TikTok app and that’s the app we know now making the headlines. It’s the one embroiled in some fairy hefty licensing arguments (depending on whose statement you read...), the one that has rolled out a Beta version of its advertising platform in countries including the UK and the one that just bought Jukedeck. The latter AI music company has spent years training deep neural networks to understand music composition and production at a granular level.

Though, to be fair, I always presumed YouTube or Facebook would buy them before TikTok, but given that all services that dig deep on UGC face the same issues on many levels, AI-generated music seems like a great addition to users’ content for the service. Yet in many ways it is about to open its own headaches.

I’ve been on a ton of panels and industry roundtables over the years where we’ve discussed the challenges that lay ahead for copyright when we enter the world of artificially composed music. Some of the greatest copyright, AI, music and tech lawyers in the world can’t yet agree where copyright comes into it. But I heavily suspect we’ll find out over the coming months and years.

It will be equally interesting to see what that means for the conventional ‘human’ music industry. Will songwriters be impacted? Can we credit software or does it belong to the human that asked for it to be generated? Will UGC revenue tumble as the world and its wife gets a machine to compose the perfect accompaniment to their crazy cat videos? As a user, I’d actually welcome some good tunes on the billions of YouTube tutorial videos instead of the utterly fucking awful library music most use now, but I do see a few potential changes and challenges ahead for many sides of the industry (and clearly so do some of our industry giants who have been heavily investing in AI music for many years already).

The copyright issues that could follow all of this are interesting ones. The very nature of machine learning means it’s fed something to analyse and learn from. But where do you draw the line? If something is fed only one artist, does an AI creation resulting from that fall under reproducing their work to create derivative works based upon the original material? But then using the same argument, let’s be honest, every band in the history of the world got their influences from somewhere. What would Oasis sound like if they’d never heard The Beatles? Or indeed Neil Innes, who they ended up having to give a songwriting credit to on Whatever after a court case. AI is already transforming daily lives, but you might not be conscious of it for things like content curation, ad serving, translation tools or deep fake videos to name but a few. There was a fun case for AI going wrong lately with the news that a multi-billionaire believed a guy that said he had the ultimate algorithm for trading and went on to lose millions in a day. He’s now struggling to sue the company/software (can you even sue an algorithm?)

So, don’t always believe the hype folks, but be sure that every big tech company you can name is working on AI-generated music and other content including AI scripts and even books (albeit utterly, utterly shit ones) and TikTik owner Bytedance is already well-entrenched in AI with its other companies’ offerings, including news and content curation. For music AI examples, and there are many now, one that sticks out to me if you want to hear the creepiest song you’ll hear this decade, stop what you’re doing and search for (Sony’s) Flow Machines – Daddy’s Car.

I hope the music and tech industries can find ways to adapt and move with all this in a way that does not further devalue music and songwriters. Licensing for any app or site that relies so heavily on music is a total no-brainer (and to be frank, seems almost ludicrously overdue), but as an industry we have to also be ready to embrace the future and move with it, or risk facing obsolescence down the line.

subscribe link free-trial link

follow us...