AVA Festival 2024

‘Surviving the AI apocalypse’: here’s what we learned at AVA Festival’s AI panel

Artificial Intelligence in music was a key topic at this year's London festival and conference.

Daniel Cave
  • By Daniel Cave
  • 5 Mar 2024
  • min read

As we advance further into 2024, discussion and debate surrounding the use of Artificial Intelligence (AI) in music continues to permeate the industry. PRS for Music announced its principles on AI back in January, in which it emphasised ‘its commitment to preserving and championing the inherent value of human creativity in the face of technological evolution’.

The publication of these principles followed a 2023 survey of PRS members on issues pertaining to AI, with 74% of respondents expressing concern about AI-generated music competing with human-made compositions. AI can already create ‘deepfake’ songs without any artist involvement, understandably irking artists of all statures — take Bad Bunny’s irate response to having his, Justin Bieber and Daddy Yankee’s vocals faked on a viral TikTok song back in November.

As AI continues to exert a growing influence on the creative industries, guaranteed pay remains at the forefront of many creators’ minds: 93% of PRS members believe creators deserve to be compensated if their music is used in any AI-generated content. Such debates over fair remuneration, as well as issues of legality and a backlash against AI, are already taking place in creative circles.

Last week, M attended AVA Festival 2024’s panel on AI, Surviving the AI Apocalypse, in London. Moderated by Declan McGlynn (Director of Communications at Voice-Swap), a quartet of panellists — Lex Dromgoole (CEO at Bronze.ai), David Boyle (director at Audience Strategies), Tom Kiehl (Interim CEO at UK Music) and Ruth Royall (Bristol artist) — spoke about how to cut through the hype around AI, explaining both the drawbacks and opportunities of this new technology and how it might shape the future of music creation.

Here’s a round-up of the insights we gained from the panel.

The importance of defining AI in music

The term “artificial intelligence” is usually bandied around in the music industry in a catch-all manner, explained Declan, whose multi-decade career has seen him cover the intersection of technology and music for outlets including Resident Advisor and DJ Mag. This vagueness can drive fear and confusion, he added, heightening an already polarised debate. The subtext: the industry has got to properly define what AI means when it comes to music creation.

This is particularly important, the panellists laid out, because the technology is used in everything from lyric generation to voice modulation, as well as helping write contracts and engage with record labels. ‘AI is such a broad term and it can mean so many different things,’ Declan told the AVA audience. ‘[But that means] we have to define AI and understand why there might be misconceptions.’

Bronze.ai’s Lex, who has worked with Disclosure to create infinitely-iterating AI versions of their songs, believes some applications of AI would likely not irk artists, such as using DALL-E to better inform designers of their cover art ideas, or improve plug-ins for all to use. He believes the debate should focus on technology which can create musical content, often based on an artist’s previous output, at huge scale and speed.

It's these usages that artists feel threatened by,’ Lex added. ‘As such, we need to make a distinction on what we mean by AI when we’re having these conversations.’

Fear and Legislation with AI

Ruth Royall laid out why many within the creative community are fearful of this new technology. ‘There are fears linked to what happened with [the rise of] streaming, and fears someone else will earn from our art,’ she explained.

Even with the touted benefits of AI — ranging from helping speed up business tasks to its provision of better interfacing technology — the AVA panel largely focused on the financial impact it could have on artists, as well as how it could affect intellectual property rights and quality control. As many AI tools are effectively free to use, Audience Strategies’ David warned, the music industry needs to start planning for a huge wave of low-quality computer-generated music coming through. '"Apocalypse" is the right word, and this outcome is possible, so we should be thinking about it,’ he cautioned.

UK Music’s Tom, however, noted that some optimism should be retained. Though there is currently no specific legislation regarding the output of AI, its mechanisms or the discographies it relies on to inform said output — an alarming fact for rights holders, who would rightfully want to get paid for the usage of their work in any context — Tom said the industry is widely united on the issue. Furthermore, the government has already backtracked on plans which would’ve allowed AI to train itself on any previous music for commercial benefit. ‘The sector doesn’t always work well together, but there is uniformity on this issue,’ he added.

While the upcoming general election might put paid to any legislative leap forward for now, Tom said that a framework of requirements for AI, or at least clarity around key issues, could be worked on. This should consider transparency around AI processes, dictate who owns the copyright on AI’s output, answer questions on whether AI content should be labelled and speak to the growing issue of personality rights (something which has become more prominent as deepfakes proliferate).

‘[Any way forward] will be complex and have ramifications for society,’ Tom clarified. ‘But [the government] has seen the importance of creators and the music industry.’

The human side of AI

It’s not all bad news and unanswered questions when it comes to AI and music. Having uploaded her voice to the catalogue of one AI platform, a bigger pool of creators (including those who write music in different languages) can access Ruth’s vocals for their creative work — ensuring royalties for both parties. Plus, she still gets autonomy over its use: ‘I found that cool — to hear my voice in a different language and with different scales.’

Technology, Ruth continued, has a storied recent history in music-making, which means some uses of AI aren’t so different. ChatGPT is now helping inspire lyrics where rhyming websites did previously, and she doesn’t believe this dehumanises the process. AI, as David observed, can even boost creativity by giving time back to an artist when the technology is used to help complete time-consuming administrative and business tasks. This in turn can help tackle burnout, especially for upcoming artists who might be trying to balance marketing, admin and content creation alongside creating music. However, he does warn that any decent AI-driven output will still need human oversight.

Despite the panel sharing concerns that AI could dangerously lower the bar when it comes to music creation, Lex said the advent of this technology could both improve music technology interfaces (making them more human and intuitive) and force humans to reflect on whether music needs to be technically perfect or communicate deeply with listeners.

‘I believe we can use AI to create better interfaces that give [musicians] better agency in making musical work,’ he said. ‘We can rethink [music-making] to just following your ears and being in the moment.’

What does an AI musical future look like?

‘It might do some wonderful things,’ Lex replied when asked this question. But, he added, there is currently a deluge of low-quality, derivative and uninteresting music being made with AI that isn’t emotionally interesting — even though it may be technically proficient. ‘I’m becoming less and less engaged the better the output gets,’ he said.

For moderator Declan, the ability to create technically perfect work at the click of a button also threatens the existence of ‘happy accidents’ which have always pushed music forward. ‘When you’re experimenting, happy accidents come out of that… but you’re going to have less of those moments with automation,’ he said.

Ruth agreed: ‘It’s important to look at the past [with technology]: look at sampling, or at T-Pain and auto-tuned sound. It’s about using this tool to be creative. Creators should use them in ways we’re not meant to, as artists love doing that.’

As it stands, it’s not fully clear where AI might lead the music industry. With fears that AI could flood the ecosystem with bad quality music if left unchecked, the panellists spoke of the need for framework and legislation as well as questioning what creativity truly is and where it comes from. Indeed, Ruth questioned if replicating other artists' sounds or lyric-writing without human oversight diminishes artistry. 'Is that creativity, or does it [only] come from a human? AI speeds up the process, but that’s a tool… these tools never give me the complete idea.'

For Bronze.ai’s Lex, the sheer computational power of AI means big questions need answering — and fast. ‘What happens when we have this infinite scale and speed of music creation? I think we must ask serious questions about how we navigate this vast ocean of content.’