We are not the robots - What does AI mean for electronic music journalism?
Wednesday, February 7, 2024
If you could predict what is going to happen with regards to AI in the next five years then you are either a liar or have managed to build a fully functioning time machine (probably using AI). We do not know what lies ahead with regards to the myriad of opportunities and threats it will bring with it but we can be certain that AI will cause a paradigm shift not last seen since the Web took hold in the late 1990s. The shift is already happening, and you have probably already heard various stories of what kind of impacts it will have on various professions. A friend of mine - who is a fashion designer - told me almost a year ago that his designs were being created and enhanced with AI and that such designs meant there was much less need for models and photographers as the generated images included non-human models. The creative industries are already taking the brunt of the AI iceberg that is heading towards us. Many working in the creative arts as designers, photographers, musicians and writers are already attuned to the threats that come with this technology and where possible the opportunities. Certain areas of ‘journalism’ are ripe for AI to churn out even more vanilla, click bait articles to flood the internet - with little care for facts or copyright.
We wanted to explore what the impact of these new generative AI tools were for the UK music scene, in particular those writing about electronic music. The journalists, like the venues, record shops and festivals are all part and parcel of a rich music ecosystem that helps shape countless national and international scenes beyond just a collection of digital sounds. We spoke to three experienced writers in this area, all have written extensively about black and electronic music. Our guests are Matt Anniss who has written for the likes of Resident Advisor, Mixmag and published the authoritative book on bleep techno, titled Join the Future. Marcus Barnes is a freelance journalist and writing coach who has written for The Guardian, Beatport and authored the book Around the World in 80 Record Stores. Jim Ottewell has also written for various publications and published Out Of Space - How UK Cities Shaped Rave Culture in 2022. He is about to launch an updated version. All three guests were keen to highlight they were not experts in AI and it should go without saying, but AI contributed nothing to this piece.
What are the key requirements to be a good music journalist in 2023?
Matt: This is a very good question, and one that I don’t have an easy answer to. I would say that some of the key requirements that were there when I started in the late ‘90s are still relevant – a voracious appetite for new music, an interest in people and stories, a desire to look beyond hype and well-funded PR campaigns; a strong and unique personal writing style and strong research skills.
Marcus: You need to have strong knowledge of passion for new music, music history, social and political context, and also be able to use balance and nuance. It’s important to be able to tap into the zeitgeist and pitch relevant ideas consistently. To think, and write, critically about music. Also have an acute understanding of the broad and complex nature of music and creativity, and how communities, movements and genres develop and emerge over time. You also need to be able to connect with artists on a human level and tell their stories in a sensitive and meaningful way. Finally, to write with passion, vigour, authenticity and sincere human emotion.
Matt: You also need to be hardened to rejection – by that I mean when editors and commissioners either ignore pitches and turn them down – and be able to use your skills on other projects. It is getting increasingly difficult to earn a reasonable living just writing about music, but it is possible if you also take on other projects. That includes writing sleeve notes for releases, research projects, writing artist bios or developing projects in other media, such as podcasts, radio series, YouTube channels and so on. Keeping your profile high and building a following – readers, listeners and so on – is vital these days in a way that it wasn’t at some points in the past. Understanding the way different outlets operate, the kinds of ideas they respond to, and how they like to engage their audience is also important.
Some areas of journalism are under threat of being replaced by AI, particularly mainstream and local media. Yet a credible dance music journalist has to totally immerse themselves in the music, they are often historians, curators and activists, they’re not passive actors. How important is that proactiveness?
Marcus: It’s essential to what we do, by being out there in the clubs, festivals and music venues. Soaking it up, meeting fans, conversing and interacting with everyday people as well as other ‘industry heads’, to gain knowledge and insight; and the lived experience of witnessing an artist or band in action. Not to mention the power of networking and how face to face connections are still so precious. You have to invest a lot of yourself into this craft, otherwise it’s not worth doing and the lack of attentiveness will show.
Matt: Certainly, for music journalists without staff jobs at online or physical publications – which is most of us – proactiveness is essential; as is carving your own particular niche. That could be as a specialist in specific kinds of stories and features, as someone renowned for making regular listening recommendations that readers trust. As a campaigner around social and economic issues within music culture, as a thoughtful analyst, or as the oracle on a particular style or collection of styles of music, be it heavy metal, drum & bass, folk or experimental electronica. My specialisms are hidden histories, alternative musical histories, and the dynamics of music communities old and new. This is something I was moving towards for a while, but really began to take off following the publication of my book Join The Future: Bleep Techno and the Birth of British Bass Music in 2019.
Jim: Many perspectives surrounding AI certainly seem to be coming from a place of fear - and not as this golden opportunity to meet a new tool or potential collaborator that will make the life of the writer any easier. I know some journos in certain fields who I’ve spoken to are eying up new career paths as a result of what they’ve read online or experienced when seeing AI at work. And of course content is more expendable than ever with all the lay-offs and cuts we’ve seen, particularly in music journalism, over the last 12 months. If AI can produce content quicker and at a cheaper rate for businesses, at a time when budgets are tighter than ever, then it’s totally understandable that these issues are being raised.
Still, I’ve had some interesting conversations with different music producers around the use of AI in their sonic creations and whether they think it will ultimately make them redundant. Cuckoo was quite bold and playful about its potential, saying how he saw it as a challenge, something he thinks will force him to be more creative and innovative if he is to beat the machine.
Jim and Matt have both published books that capture important moments in dance music history. They’re important because they capture the voices of those who were there, putting on the nights, making the records and playing the music. Do you believe that these conversations can only be captured between human beings?
Matt: Fundamentally yes, because capturing those histories accurately, in full, and with sensitivity; not only requires skill and plenty of prior research, but also the ability to build trust and relationships with interviewees over a period of time. You could certainly write an AI program and train it to come up with interview questions, but it would be reliant on there being a certain amount of information out there about the person or people you are interviewing. If you’re researching, say, a historic music community, scene or artist who never got much press and very little has previously been documented, it needs a human-to-human conversation, and an ability as a journalist or researcher to sense what question to ask next. You need to be able to identify potentially important points buried within question answers, and to know when to, and when not to, press your interviewee for more information. That said, you could almost certainly train an AI program to write a biography of, say, the Beatles or the Rolling Stones, because there is so much information out there. Whether it would be accurate is debatable, because it would rely on existing sources, some of which may have been discredited. Applying that to dance music for a minute, it would be easy to train an AI to write something – even a book – about the early days of acid house in the UK. That said, it would most certainly result in something inaccurate and based on easily discredited cultural myths. This of course is because so much existing literature on the subject repeats the same inaccurate and over-simplified narratives. Many researchers and writers have published corrective articles or books in recent years, but they are still outnumbered by inaccurate accounts that rely on the same tired, too-good-to-be-true stories based on narrativized articles (and subsequent books) published close to the time (hello ‘Ibiza Origin Myth’).
Marcus: 100%. For now at least, perhaps in a few generations time people will have become accustomed to AI and robots interviewing them. I’m not sure. At present, genuine connection and interaction between human beings, in that ‘journalist-meets-music head’ dynamic, lies at the core of so many crucial bodies of work.
Regardless of what you write about, disco, techno, drum & bass, it is all about trust between the writer and the artist. How important is that trust within music journalism?
Marcus: I, personally, don’t believe there’s anything that can replace that depth of connection when a journalist provides a safe, disarming space for an interviewee to open up and tell their story. Those who do it best are able to create that trusting space, nurture the storytelling and then recount the stories vividly for people to immerse themselves in. It’s an art and one which I highly doubt could be replicated by AI.
Matt: This is another very good question. My teenage self, who was obsessed with weekly and monthly music publications and the people who wrote for them, would say it is vital. As a reader you instinctively ‘trust’ a music writer if you enjoy their work, their tastes are similar to yours, and you dream of being in their position. Once you get into that position yourself, you initially believe in editorial freedom and the ideal of championing artists, releases, scenes and communities you think are deserving of coverage. Then you start to become aware of commercial pressures, the pulling power of particular artists or ‘exclusive’ interviews, and the need to monetise music journalism – not just in order to earn a living yourself, but also to keep media outlets afloat. At that point, you realise that trust is important – between individual journalists, their peers and the audience, and the outlets and their audience – but you may have to make some compromises, whether you like it or not. Finding a balance – earning trust and being trusted because of your authenticity and ethos, but compromising where strictly necessary – is tricky but essential from a commercial standpoint. Of course, if you’re lucky to have no commercial concerns or earn the majority of your income elsewhere, it is much easier to just stick to your guns and ignore those commercial pressures.
What about passion? We are all into the music we love because it is a bonafide addiction, something we all feel remarkably passionate about. Can you see computers bluffing that kind of unbridled, trainspotting, yet objective enthusiasm?
Marcus: Potentially. Martin Lovegrove at Dig Magazine recently wrote a newsletter post about AI compiling its own instalments of the Street Sounds Hip Hop series from Volume 23 onwards, which, of course, were never actually released. They’ve now been generated by ChatGPT. Sadly for ChatGPT it made a lot of mistakes. Humans can be so deeply, emotionally connected to the things they get nerdy about that they would never allow themselves to make such mistakes. So perhaps in terms of data mining, AI might be better, but accuracy is essential to a music nerd and they’re less likely to make such mistakes. AI can get ultra nerdy, mining all the data out there and regurgitating the information it collects. Could it do that better than me? Maybe, but could it do that with genuine heart and passion, actually caring personally about the information it’s sharing?...
Matt: Yes and no. If you fed an AI program a great volume of work by a specific journalist, in time it would be able to replicate that passion. In written work that passion manifests itself via sentences packed with adjectives, an excitable and enthusiastic tone, and the things that you choose to write about. I’ve no doubt that my writing style could be replicated, though it frightens me that it could. I doubt it could replicate the knowledge that I and many other long-standing music journalists have accrued over the years, or the ability to recall that information and put it into context in a an article about, say, a new artist whose work you see fitting into a particular movement or being influenced by something weird and obscure. Unless those comparisons had been made before and featured in the sources fed into the AI program.
Jim: The proactiveness of a dance music journalist is obviously really crucial when it comes to tracking down new stories and perspectives and getting under the fingernails of a story or producer/DJ or promoter. If you don’t do something about capturing a story, then it can pass you by and I’m unsure of whether AI can do that from scraping and rehashing existing web content… yet. At this point we can talk about unbridled enthusiasm and capturing moments of conversations as something very human too. But AI still feels like it's in its infancy as a technology and there are plenty of unanswered questions surrounding it - do readers want the internet to be stuffed with dry, unengaging content regurgitated from other sources? Can Google change its ranking tools so as to detect and potentially penalise content written by AI?
I really enjoyed Todd L Burn’s Music Insider Journalism newsletter where he talks with ChatGPT about its potential impact on the future of music writing. In the Q&A, he asks whether AI will be able to replace humans - the answer: “Here’s a thought that might stir the pot: AI might eventually produce music journalism that is so good, so accurate, and so engaging that the average reader won’t be able to tell the difference between AI-generated content and human-generated content.”
Ultimately, as ChatGPT says, I guess this is a definite concern. It’s always been an all-out hustle for opportunities in music writing and if we’re going to programme our way to a point where there are less of them, then I guess we’re going to have to keep on evolving and adapting too.
If someone asked you to go to a night where the music was generated by AI and mixed using purely algorithms, would you go?
Marcus: For sure, I pitched the idea of covering an “Algorave” years ago! I also reviewed and played some music generated by AI in my last batch of techno reviews published through my newsletter musicistheanswer.substack.com
Jim: At this point, I feel like it would probably be an interesting experiment to experience and definitely worth writing about. Club culture is moving into different areas of tech and expanding all the time - as it should - since inception, it’s always been about looking to the future. Or at least was before we started living in it.
Would it be enjoyable? The Algorhythm rave in east London sounded fairly weak as a party but the ideas behind it are really intriguing. I went to a VR rave experience - ‘In Pursuit of Repetitive Beats’ - last year and that was totally mind bending.
Matt: Possibly, purely out of curiosity, though it would be tough setting aside my prejudices, years of DJing to audiences, and my fundamental belief that creativity is essential to the human condition and that people should be able to make a living out of their creativity. The fact that so few are able to do that now is another issue, albeit an important one.
If so, would you write about it?
Matt: Yes, I probably would – regardless of any misgivings, which could be outlined in the subsequent feature article or review, it is something we’re unfortunately likely to see more of in future, so it is newsworthy and interesting. There’s a story to tell, even if it’s a potentially dystopian one.
Marcus: I have reviewed and played some music generated by AI in my last batch of techno reviews published through my newsletter musicistheanswer.substack.com
Do you feel we could get to a point where we would have AI writing electronic music reviews for AI produced music?
Matt: For sure, particularly for online music stores, streaming services and the like, where the accompanying write-ups are specifically aimed at shifting units, selling downloads and gathering streams. At present there are a number of large outlets of this kind that employ reviewers to create copy which looks like an authentic review. These are written by a human, with context, descriptive text about what’s on offer, and value judgements – but which is uncredited and solely exists to try and sell music. I’m pretty sure that within a few years, AI programs will provide these marketing style ‘reviews’ and definitely the press releases emailed out to writers with promotional copies of releases and nobody, bar the writers now out of work, would know. AI produced music is already out there and in some cases scarily accurate, so computers ‘reviewing’ computer-generated music is not far off. I don’t personally think this is a good thing, but as a reviewer as well as a researcher and feature writer I would say that.
Jim: I’ve had some interesting conversations with different music producers around the use of AI in their sonic creations and whether they think it will ultimately make them redundant. Cuckoo was quite bold and playful about its potential, saying how he saw it as a challenge, something he thinks will force him to be more creative and innovative if he is to beat the machine.
Marcus: I don’t see why not. Though, whether anyone but AI would be interested in reading the reviews is another question…
Could AI help with reviews, especially album reviews? Given many reviews use the label’s press release as a basis for the review, could an AI tool like ChatGPT or Google Bard be used to at least generate some of the review with the journalist adding how they feel about the music on a human level?
Matt: In principle it could, yes, but personally I would not take that approach. I would much rather spend the time researching an artist and release, thinking about context and narrative, and writing the entirety of the review. The process, and the thought that goes into it, is part of the job – and a rewarding part of it too.
Marcus: Maybe, if you’re feeling lazy, or want to save a bit of time. I wouldn’t myself, but I have my own way of working and AI is not integrated into that process right now.
Have you considered using AI to generate ideas for articles and would you consider using generative AI to expand on ideas you might have for articles, or even a book?
Jim: Like I’ve said above, I think AI is a useful tool at this point and believe it’s certainly effective in enhancing certain aspects of the creative process. But it’s not something I’ve utilised when writing anything for a book or a feature - these are elements of my freelance work that I really love putting together and give me a certain sense of satisfaction that I don’t think would be achievable if I’d asked ChatGPT to give me a leg up.
For other areas of my freelance work - copywriting for example - I’ve definitely explored AI in greater depth and run content through SEO tools to help enhance Google ranking and searchability. This is where it comes in - so it can aid and abet some of the work I do rather than muscling me out of the way…
Matt: I have never considered using AI to generate ideas or expand on them, in part because it feels like cheating but also because I have no interest in this. Maybe my attitude would change a little if someone demonstrated the benefits of such an approach, or how AI could be used ethically as a work tool that improves the ideas you generate, or provides alternative takes that maybe had not occurred to you. I have friends who aren’t professional writers, but who run record labels or regular club events, who use ChatGPT to tweak text or suggest ways to write marketing material and social media posts. They say that they do it to see whether it comes up with better or alternative phrasing, or different words that might be superior to their initial basic text. I did once ask ChatGPT to write a Tweet in the style of my social media posts, and it came up with this: “Check out this great release I’ve found!” I think I’ll keep writing my own tweets!
Marcus: No. I know people who’ve used it for similar purposes and found it useful. Good for them. It’s just so far removed from what I care about when it comes to my work and creativity that I can’t see myself leaning on it anytime soon. Perhaps that will be to my detriment, I don’t know.
What role does radio and podcasting have for modern journalists these days? You all DJ and have engaged in Q&A sessions, it seems like an effective additional way to share knowledge and insights?
Matt: Radio and podcasting have an important role for music journalists, as both are platforms to convert casual readers into ‘fans’ and ‘followers’, while also increasing your profile and solidifying your position as an expert in your field. It’s another direct form of communication, especially if the radio shows or podcasts you are hosting, creating or producing closely connect to your role as a music journalist, historian or critic. For many years I did radio shows that were purely music-based and often mixed, often under DJ aliases, but came to the conclusion that they weren’t the best way to connect with the audience I’d built up through my writing and research. That’s why I changed the format and style of the monthly Join The Future shows. They air online and on DAB in Bristol via Noods Radio, but can also be accessed through the Join The Future website, Mixcloud and Soundcloud – and they get far more listens that way. While music does feature, the majority of shows have a theme or a featured interviewee. Some expand on aspects of my research-based work, others are extended audio versions of previously published written features, and some are mini-documentaries – for example the two-part exploration of the history of bassline I did a couple of years ago.
Marcus: Yeah, radio and podcasting can be great for sure. If you have the confidence and ability to articulate yourself verbally, then it’s a great outlet for expanding one’s repertoire and reach. Also, nice for readers to put a voice to the words they read.
Matt: I do think that podcasting and audio is one way that AI could be used positively by writers. AI programs exist that can create audio versions of written texts in the “voice” of a specific person, such as a writer. They seem to work by analysing a small number of inputted audio recordings of a person’s voice. So, for example, I could record myself reading an article or a chapter from one of my books. From that, the program could, in theory, create audio versions – that sound like a recording of me reading them – of anything I’ve written. The ability to quickly create audio versions of written pieces – albeit AI-generated ones that merely sound like the writer – could be very useful, as you could create podcasts of varying lengths out of everything you write. I’ve not tried this yet and I’m still not sure how I feel about it – can these programs impart emphasis, tone and style as a human could – but I’d be prepared to try it.
We’re probably going to see a lot more AI-generated vanilla music journalism in the next few years, but could it also result in a greater appreciation for the architects of good quality journalism which you have all gained a reputation for?
Matt: This is the hope, but when AI programs become more advanced and any issues are ironed out, will readers be able to tell the difference? You’d hope so, but plenty of people who currently read music journalism don’t check the bylines or tag the writer in when sharing articles on social media. Perhaps we need AI-generated content on websites and in printed publications to be clearly labelled as such, with more prominent bylines and credits for those written by actual humans.
Marcus: I don’t think it will be as simple as that, nothing ever is. There are so many little layers and social bubbles in western society. People engage with writing and music journalism from all manner of different approaches and the written word alone is no longer the only avenue for music journalism to be shared with the world. I don’t think AI is necessary for anything I currently do, maybe I’m making a rod for my own back but I would rather not lean on it for my practice. I’m also not sure very many people from Gen Z onwards are really that fussed about the architects of good quality journalism. We’re not that important… But the narratives and histories are.
Ultimately do you worry AI will be monumentally bad for creativity? Or is it too soon to say?
Marcus: It’ll be bad for those who adopt it wholesale and allow it to be their only outlet for expression. Yes, the human still has to instruct it to generate whatever it is they want it to, but that process of actually creating something with your mind-to-hand connection is lost. AI might do it perfectly, but there is something beautiful in the imperfection of human creativity. That soulful element is lost when we only produce what we deem to be “perfect”. Creativity also needs to be a process bound by time. What I mean by that is, if you’re just using AI to generate articles, books, headlines or whatever, in mere seconds, you’ve lost an important part of the creative process. Unfortunately, a lot of humans seem to value convenience over soulful authenticity, an easy life over making an effort, so there will be more and more people utilising AI to create for them.
Matt: I do worry that AI will be bad for creativity, but then I do know people who have used AI positively as a tool as part of their creative process while delivering work that is undoubtedly their own. Fundamentally I do not know enough about it and need to learn more. I know that’s a surprising thing for a music journalist to say – we tend to be over-sure of our opinions, regardless of whether we have any actual knowledge of the subject or not – but it’s true. Within a few years we’ll know, as the pace of change is quickening – for better or worse.
Where do you discover new music? Are you still reliant on personal recommendations or have AI algorithms started to influence your tastes?
Marcus: I search around myself, and take personal recommendations. In terms of algorithms, YouTube occasionally serves up mixes and tracks I haven’t heard before. I rarely use Spotify and, if I do, it’s to look up a specific song or album, not to take any recommendations.
Matt: I discover new music through a combination of personal recommendations, digging in physical and online record stores, discussions with artists and label owners, and pieces in music media. I get very irritated when Apple Music and other similar streaming services randomly play things after something I’ve chosen to listen to – just because I like X artist that uses synthesisers (as an example), doesn’t mean I’ll like Y artist that also uses synthesisers. Leave me alone to select music myself!
Marcus Barnes Discover new digital dance music with Trackhunter