What Value Does AI Bring to Music?
A First Look at AI Tech’s Possibilities & Adoption Barriers in the Music Industry.
GM readers 👋,
Happy May!
For this month’s newsletter, we’re going to dive into a well-trodden topic these days – artificial intelligence (“AI”). I’ve wanted to write about AI’s potential impact on the music industry for the past few months. The conventional wisdom is that AI will change the way people work, learn, communicate, and more. My gut reaction from using AI tools like ChatGPT and Midjourney is similar – it feels like we may be in the midst of a once-in-a-generation technological shift. So, what does the future of music look like in an AI world?
Like many of you, I’m trying my best to make sense of everything in real-time. I’m a finance and strategy nerd, not a technical person. I’m working to understand what’s happening as best I can. And then I’m trying to figure out how to translate that understanding in a way that is easy to digest. It requires a mix of summarizing what’s going on, guessing where we’re headed, and looking back at history for guidance. That said, given the rapid pace of change, this is no easy feat and much of what we’re covering may become outdated quickly. As one final caveat, there are people with a great deal more expertise than me on all of this. I recommend that you read and listen to their work, much of which I link to in this piece.
In this piece, we’ll explore what value AI brings to music, what barriers are hindering further adoption of this technology, and speculate on a few potential implications for the music industry.
Finally, I’m very excited to announce Leveling Up’s first sponsor – Royalty Exchange. Royalty Exchange is a leading platform that matches entertainment IP creators seeking funding with investors seeking to add low correlated assets to their portfolios. My primary intention for this newsletter has been to learn in public and meet new people in a shared space. I’ve never really had a plan to monetize. But when the opportunity came up, it made sense, both in keeping with our music IP focus and as a fun experiment. You can learn more about Royalty Exchange below.
I’m grateful that each of you takes time out of your busy days to read my thoughts. And I’m excited to see where the Leveling Up journey goes from here!
And with that, on to the disclaimers…
Note: I write this newsletter to learn in public. This piece is for informational purposes only. None of this is financial or legal advice. Do your own research!
Thanks again for reading. If you have any feedback or ideas for future posts, please leave a comment or shoot me an email.
Now, let’s get after it!
Jimmy
PS If you like our work, please consider hitting the heart ❤️ emoji and sharing this article, because it helps more readers discover our writing and supports our growth! And if you’re not already a subscriber to Leveling Up and don’t want to miss out on future newsletters, feel free to enter your email below and you’ll receive new posts directly in your inbox.
Today’s Leveling Up is brought to you by… Royalty Exchange
Back in 2013, I stumbled across Royalty Exchange’s website while researching a project in business school. I was struck by the idea of acquiring cash-flowing royalties associated with some of my favorite songs. And I was intrigued by the relatively high implied cash flow yields. It started me down the rabbit hole of learning how music IP worked. So it feels pretty surreal to have Royalty Exchange become the first sponsor of the Leveling Up newsletter roughly a decade later.
For those unfamiliar, Royalty Exchange is a leading platform that matches entertainment IP creators seeking funding with investors seeking to add cash-flowing, low correlated assets to their portfolios. To date, rights holders have raised over $135M+ across 1,600+ deals on the platform.
Using Royalty Exchange, investors have purchased royalties tied to songs by artists like Jay-Z, Rihanna, The Doobie Brothers, Beyonce, and the Shrek film franchise. Click here or the link below to see what assets are available for purchase now.
What Value Does AI Bring to Music?
“It frankly doesn’t really matter which trends, or design patterns, you incorporate into your product. If the product is compelling to the market, it will succeed. If the product is not compelling to the market, it will fail. It’s not much more complicated than that.” - Marc Andreesen, Co-founder and General Partner at Andreesen Horowitz.
In the business world, 2023 has been marked by the acceleration of an artificial intelligence hype cycle. The current conventional wisdom is that AI will have a significant impact on the way industries operate and how people go about their day-to-day lives. Consulting firm PwC estimates that AI will contribute $15+ trillion to the global economy by 2030. Meanwhile, Microsoft founder Bill Gates encapsulated this sentiment in a recent post entitled “The Age of AI has begun,” writing that “the development of AI is as fundamental as the creation of the microprocessor, the personal computer, the Internet, and the mobile phone.”
AI development isn’t new, with the first piece of AI music reportedly developed in the 1950s by British scientist Alan Turing (yes, that Alan Turing). However, the public release of several AI tools in the past twelve months – like OpenAI’s ChatGPT (text) and Midjourney (images) – have demonstrated the remarkable progress and potential of large language models (“LLMs”).
These tools are witnessing rapid user adoption. ChatGPT, OpenAI’s text-based chatbot, reached 1 million users 5 days after launching to the public in November 2022. It then reportedly reached 100 million monthly active users just three months later, which is quicker than any other consumer application in history. Meanwhile, as of March 2023, Midjourney reportedly had 13+ million registered users after launching its open beta in July 2022.
In response, established companies are repositioning to incorporate AI tools into their existing product suites. Alphabet is converting itself into an AI company. Microsoft has invested $10 billion in OpenAI and its CEO Satya Nadella recently stated “Every product of Microsoft will have some of the same AI capabilities to completely transform the product.” Adobe, Canva, Discord, and others have also made recent announcements that they are adding AI functionality to their products. For better or worse, AI is already impacting how industry leading companies design their products.
Meanwhile, private market funding and valuations for AI start-ups have increased significantly. According to Pitchbook, AI start-up investments soared to $5.9 billion in 2022 from $1.5 billion in 2020, with VCs crossing all twenty toes that they can find the next OpenAI. Unsurprisingly, the median pre-money valuation for generative AI start-ups has increased more than 4x over that same period.
And the amount of AI research being done is growing, which will likely lead to continued improvements in the technology. The number of academic papers published on Machine Learning or Artificial Intelligence in the arXiv archive each month (depicted below) is increasing exponentially. Notably, this figure only reflects published papers, as a significant amount of additional research is conducted but ultimately never published.
Given the AI’s momentum, it’s worth exploring how entrepreneurs and companies are using AI-driven tools in the music industry. At the same time, the reaction to the generative AI hype train has drawn its fair share of industry critics and skeptics. In this piece, we’ll take a closer look at the potential value AI brings to different music industry participants, the current barriers to further adoption, and the potential implications of greater AI enablement on the music industry value chain.
Some Basics and the AI Tech Stack
Before analyzing what value AI brings to music, let’s try to define what we mean by AI. This will help us better understand the technology’s potential value, risks, and implications for the music industry.
Artificial intelligence is the practice of getting machines to mimic human intelligence to perform tasks. There are several examples of day-to-day interactions with AI currently. For example, customer service chatbots that help you navigate websites and voice assistants – like Siri and Alexa – are founded on AI technology. Meanwhile, machine learning (“ML”) is a subset of AI that enables a machine or system to learn and improve from experience without human supervision or intervention. It uses algorithms to analyze large swaths of data, learn from the insights, and then make informed decisions. ML algorithms improve over time as they are exposed to more data (“trained”). As a result, data quality and quantity is extremely important for building a better model. Although AI and ML are often used interchangeably, they are actually not the same thing.
Finally, generative AI falls under the category of machine learning. It involves training a model to generate new content in response to prompts from a user. Popular Generative AI tools include ChatGPT, Midjourney, and DALL-E. Despite their rapid adoption, using Generative AI models also poses inherent risks (some of which we’ll discuss later in this piece). For example, generative AI model outputs can sometimes be plain wrong and other times can be biased. Again, this is due to the fact that the outputs are only as good as the data on which the models are built.
With these definitions in mind, it can be helpful to think about AI tools across two dimensions: 1) how much humans interact with the AI; and 2) how the AI within a system can sense its environment, learn, and take actions in response to changes in its environment. Point being that many AI use cases still require a human’s involvement and this level of involvement exists across a spectrum. If you plot these two dimensions on a table, you have four distinct ways that AI is being used and has been used in the past.
Automated intelligence: Automation (i.e., no human involved) of manual/cognitive and routine/non-routine tasks. Examples include automating the extraction of information from documents.
Assisted intelligence: Helping people to perform tasks better and faster, but allowing the human to make the ultimate decision. Examples include speech recognition software and predictive text algorithms.
Augmented intelligence: Helping people to make better decisions with AI becoming a more active contributor to the decision-making process. Examples include machine learning algorithms and chatbots.
Autonomous intelligence: Automating decision-making processes without human intervention. Examples include an autonomous car that can drive anywhere without human assistance.
Finally, it’s worth considering the AI technology stack, which is depicted below. There are four layers consisting of: 1) a Data Layer where data is collected to train AI models; 2) an Infrastructure Layer where model data is stored and secured; 3) a Model Layer where the machine learning model generates an output (e.g., new content) either autonomously or in response to prompts from a user; 4) an Application Layer where the user interacts with the computer system and AI technology to create new content.
Based on my research, there is innovation within the music industry primarily occurring across three of these layers – Data, AI Models, and AI Apps – with most start-up activity occurring at the Apps Layer. For the Data Layer, organizations like Tribe of Noise and Harmon.ai are working to build open-source and/or fully-cleared data sets to train AI models. Meanwhile, Believe is developing technology that analyzes AI track data to properly attribute and compensate rights holders of the training data. For the AI Models Layer, over ten new models have launched this year alone, including MusicLM (Google), SingSong (Google), Noise2Music (Google), AudioLDM, and Make-An-Audio. This wave of new music AI models could result in music experiencing its own Midjourney or ChatGPT moment. Finally, for the Apps Layer, companies like BandLab, LANDR, Spotify, and Boomy are developing their own AI-powered music applications / tools for users.
With this background information in mind, let’s explore the potential value AI brings to music.
What Value Does AI Bring to the Table for Music?
Armed with a basic understanding of what AI means, let’s look at why there is so much hype around this trend. As mentioned above, venture capitalists, technology, and entertainment companies are investing in the music AI trend. Music AI start-ups have also seen an increase in funding with ~$100 million raised in 2022 and the median AI start-up valuation increasing 2x+ over the past 18 months. Companies – like Google, Sony, Spotify, ByteDance, Meta, and others – have internal research teams developing music AI tools and products. Meanwhile, major labels – like Universal Music Group and Warner Music Group – are making venture bets on music AI start-ups. And data from Media Cloud (depicted below) suggests that ChatGPT is already seeing similar airtime to that given to cryptocurrencies in 2021, when Bitcoin prices peaked.
That said, it’s easy to get caught up in a technology’s Hype Cycle. Many of these same investors and companies have made bets on other hot investment trends – like the metaverse and Web3 – that have yet to result in widespread user adoption. Simply put, investing in a new technology trend does not imply that product market fit will automatically (or ever) be realized. And prioritizing a new technology over the customer can often lead to building a solution in search of a problem.
Along these lines, Amazon founder Jeff Bezos gave a great interview in 1999 where he underscored this point. To Bezos, it didn’t matter that a new technology (the internet) was used to enable Amazon’s business model: “It doesn’t matter to me whether we’re a pure internet play. What matters to me is do we provide the best customer service. Internet shminternet, you know, that doesn’t matter.” The internet helped Amazon to provide a better customer experience by eliminating the cost of distributing information. Using this technology, Amazon sought to provide its customers with a more convenient buying experience, lower prices, greater selection, and more information to make purchasing decisions. In short, Amazon was focused on providing the best customer experience above all else.
So the big question mark is what value music AI tools bring to the table for users / customers? From my perspective, tech and media analyst Ben Thompson has written the clearest and most concise answer to this question: “AI is zero marginal generation of information (well, nearly zero, relative to humans).”
Being able to easily generate new content with little to no prior knowledge, effort, or experience is powerful. If you’ve used any of the popular AI tools, this should resonate. In ChatGPT’s case, after entering in a simple prompt, users receive detailed, thoughtful answers from a system capable of passing a bar exam. This leads to quicker access to insights than a typical Google search, output that can assist in the creative process, and even emotional support, with 2 million people currently using Replika’s AI chatbot product.
For music, the potential benefits of using AI-driven tools are not different. Let’s consider a few of these across user types.
For musicians, the benefits of AI tools include:
Creative inspiration. There are a variety of tools that help creators brainstorm by suggesting new ideas. For the former, companies – like CoSo by Splice, Soundful, and BandLab SongStarter – that allow creators to experiment with AI to develop new songs without relinquishing full control over the creative process. Meanwhile, others like These Lyrics Do Not Exist, Sudowrite, and SongPad help creators brainstorm and/or generate song lyrics and text.
Easier to create new content. As a creator, the content treadmill can be quite daunting and often leads to burnout. In addition to helping creators brainstorm new ideas, some AI tools are capable of delivering finished music with minimal human involvement. Companies like Boomy, Loudly, AIVA, and Amper Music are capable of producing full songs autonomously, even if it’s debatable whether these are on par with commercial hits.
Opportunity for deeper engagement and monetization with existing IP and NIL. By enabling users to experiment with a creator’s existing IP and/or name & likeness, there are opportunities for deeper engagement and monetization. There have been several early signs of the potential product market fit here. An AI song featuring cloned vocals from Drake and The Weeknd was uploaded to social and music streaming platforms, generating 600K+ streams on Spotify and 15+ million views on TikTok before being removed due to copyright infringement claims. Meanwhile, artist Holly Herndon created a digital twin Holly+ and is encouraging anyone to create art with her voice and image. Check out this interesting TED talk that Herndon gave about the project. Canadian singer songwriter Grimes recently announced a similar AI project where creators can use her voice to develop new songs in exchange for a 50% share of master royalty income generated from the songs. Along these lines, start-ups like Spawning are building tools for creators to manage use of their existing data in AI models.
Access to new audiences. AI tools enable creators to experiment with new design spaces, which in turn can enable music to resonate with new audiences. One recent example is HYBE-signed artist Lee Hyun unveiling a new project called MIDNATT. The project’s first single, “Masquerade”, was released in six languages by using AI technology from the company Supertone (acquired by HYBE in 2021). This opportunity is interesting given that studies have shown a preference for local music across several top European music markets.
For labels and publishers, the benefits of AI tools are similar to those listed above for creators. Indeed, they can help music content companies find new licensing opportunities and drive deeper engagement across their catalogs. There are also interesting (albeit potentially more controversial) potential benefits to using AI technology on labels’ and publishers’ cost structures. These include:
Lower operating costs. Creators currently have more tools to make, distribute, and market their music than ever before. This is reflected in the fact that the fastest growing part of the global recorded music business is the self-releasing artist. As a result, the leverage between artists/songwriters and labels/publishers continues to shift towards the former. However, AI tools potentially provide the latter with the ability to create new musical content autonomously (i.e., without human creators). Larger labels like HYBE are already using their artists’ data and proprietary AI tools to create new content. Meanwhile, start-ups like Authentic Artists’ WarpSound are experimenting with AI-powered virtual artists. Of course, it remains to be seen if a virtual artist can find the same product market fit as a human artist. That said, the millions of people interacting with their Replika chatbots suggest that there could be a market here.
For listeners, the benefits of AI tools include:
More content from more creators. With a significantly lower (arguably non-existent) technical barrier to creating new content, listeners will have access to significantly more music, with more listeners able to become creators themselves. Along these lines, AI music start-up Boomy states that 14.5+ million songs have been created using its platform since it launched. Boomy claims that this is around 14% of the world’s recorded music. Of course, more content doesn’t mean the quality bar for most of this content will be high, which leads to the next potential benefit.
Opportunity for greater personalization. In a growing sea of content, AI-powered music recommendation and discovery tools become even more important to ensure listeners don’t have a worse experience. Spotify has introduced an AI playlist tool to analyze user preferences, listening patterns, and contextual data to suggest personalized playlists and recommendations. Meanwhile, Tencent Music Group is developing a chatbot to help users discover new content.
For media platforms (e.g., Spotify, TikTok, Tencent Music Entertainment, etc.), the potential benefits of AI tools are very interesting. These tools could help accelerate the early signs of some platforms providing creators with an all-in-one offering (i.e., from creation to distribution to monetization) that resembles that of a traditional label / publisher.
Deeper engagement on the platform. As mentioned above, several platforms are developing music AI tools to help drive more engagement on their platforms. TikTok is reportedly working on a music creation app/feature that will leverage AI tools to make song creation easier. Tencent Music Entertainment (“TME”) has already created more than 1,000 songs on its platform using AI technology with one song already garnering 100+ million streams. TME is now working on AI-enabled music creation tools. By making it easier for users to create new musical content, these platforms hope to keep users active longer.
Lower content costs. If you look at Spotify’s P&L, its cost of revenue makes up ~75% of total revenue. This is predominantly driven by royalties paid to rights holders (i.e., labels, publishers, artists, and creators). Social platforms like TikTok are also constantly negotiating with labels and publishers over royalty splits. By providing artists with easy-to-use AI tools to create new music that can then be seamlessly distributed to millions of users, platforms can potentially circumvent labels/publishers - or at least swing leverage more in the platforms’ direction. Of course, platforms that are more reliant on major rights holders’ music content for user engagement (e.g., music streaming services) will likely be less aggressive in this regard. If you come at the king[s], you best not miss.
New revenue streams. If AI tools keep users more active and attract new users, then these media platforms will be able to generate more advertising revenue, all else equal. They will also be able to potentially layer in new revenue streams, such as selling premium tools & services to creators, or even taking a percentage of creators’ revenues generated on the platform.
AI allows anyone to easily generate new information (e.g., musical content) regardless of past experience, knowledge, or skill. Given this technical unlock, it’s probably not surprising that most companies building AI tools for the music industry are focused on the music creation process. Water & Music put together a fantastic market map of companies building in the space (pictured below), which shows just how many companies are focused on AI tools for music creation.
In summary, AI offers a potential paradigm-shifting technology that could lead to immense benefits for artists, labels and publishers, listeners, and media platforms. These AI-driven tools provide value in different ways to each part of the music industry ecosystem.
Listeners have access to more content and tools to navigate more effectively.
Artists and songwriters can brainstorm new ideas, foster deeper engagement with their catalog, create new content more easily, and reach new audiences.
Labels and publishers can help their artists more efficiently and profitably take advantage of these tools while also potentially lowering their operating costs.
Media platforms can leverage these tools to keep users’ attention on their platforms, lower content costs, and generate new revenue streams.
That said, these are still early days. There are numerous barriers to further music AI adoption, so let’s explore some of those.
What are the Adoption Barriers Facing Music AI tools?
While AI tools offer several benefits to the music industry, there are also a number of barriers that need to be overcome to achieve sustainable mass adoption. From my perspective, these include (but are not limited to):
Intellectual property law considerations. Music AI creators will have to navigate intellectual property laws and figure out how to manage copyrights in new ways. Many AI models are trained on large swaths of data, which may be copyrighted and unlicensed. Practitioners training a public, commercialized music AI model or building on top of one of these models need to make sure that the training that underlies the model is properly and explicitly licensed to avoid infringement issues. The National Music Publishers Association has already sent a letter to Congress to examine how AI models train on potentially unlicensed human works. Along these lines, the removal of the recent viral song that used AI generated vocals of Drake and The Weeknd from all major platforms has shown the risks of not obtaining the necessary licenses. Unlicensed AI music can potentially run afoul of intellectual property and name & likeness laws. But these theories have yet to be tested in court. I expect that we’ll continue to see more lawsuits from rights owners who want to enforce unauthorized uses of their copyright protected content. Making sure the right agreements and attribution processes are in place will be key to monitoring AI IP issues and resolving disputes. For more on this topic, it’s worth checking out Monetizing Media’s great podcast on Generative AI and IP law.
Political and regulatory uncertainty. In the US, the Biden Administration is beginning to study possible regulation of the AI industry. Meanwhile, a Senate Judiciary Committee recently held a meeting with OpenAI’s CEO Sam Altman to explore the risks associated with AI technology and the need for greater regulation and disclosure of AI use. Italy went so far as to ban ChatGPT over privacy concerns (but then reinstated the service after OpenAI introduced several changes to protect user privacy). In the music industry, NMPA President David Israelite and RIAA CEO Mitch Glazier have been outspoken towards regulators and politicians about the need to protect creators’ rights against the risks posed by AI technology. And a coalition of several prominent music organizations, including the NMPA and Recording Academy, have formed the Human Artistry Campaign with the goal to “ensure artificial intelligence technologies are developed and used in ways that support human culture and artistry – and not ways that replace or erode it.” In short, political objectives and regulatory complexities may not align with business objectives and consumer wants, thereby potentially slowing adoption.
Key platform holder risks. It’s currently unclear how major platform holders will treat the distribution of AI music. This creates friction for users trying to find, download, and play this music. If companies like Spotify, Apple, Deezer, Soundcloud, etc. are supportive, it will go a long way to accelerate the industry’s growth. Conversely, if these platform holders reject or severely limit AI music, it will be more difficult for the industry to grow. To date, the platform holders have taken a somewhat mixed stance. For example, Spotify CEO Daniel Ek has made comments that suggest he is concerned about the abuse of AI technology, but he also wants to keep the door open for AI songs on the platform, stating: “[T]he AI pushback from the copyright industry, or labels and media companies, is really [concerned with] issues like ‘name and likeness’, what is an actual copyright, who owns the right to something where you upload something and claim it to be Drake [when] it’s really not, and so on. Those are legitimate concerns. Obviously, those are things [Spotify is] working with our partners on in trying to establish a position where we both allow innovation, but at the same time, protect all of the creators that we have on our platform.” Meanwhile, Believe recently announced it would not allow any AI music to be uploaded via TuneCore to streaming platforms. In short, distribution policies towards AI music remain unclear at best.
Privacy and ethical considerations. Data privacy concerns have dogged big technology companies, with CEOs recently called before the U.S. Senate for questioning. This potential risk for AI models is non-trivial. Massive databases might include user data that is both personal and sensitive. Along these lines, Samsung recently banned the use of ChatGPT after sensitive proprietary data was leaked due to usage of the generative AI service.
Lack of available training data. A generative AI model’s output is only as good as its training data. While text-based AI models like OpenAI’s GPT-3 have been trained on vast datasets, OpenAI’s Jukebox music AI model is trained on 1+ million songs but many of these songs are copyrighted and unlicensed. As a result, creating tools and/or using tools that leverage these unlicensed music AI models poses a risk for companies and users.
Complex user experience. Creating a commercially ready AI song today still requires a certain amount of knowledge and technical skills. In order to grow the Music AI market to mass adoption, it will be necessary to foster a UX that is accessible to everyone, including casual users.
Computing resources necessary to train and maintain AI models are currently expensive at scale. The infrastructure companies that provide computing resources to AI companies are seeing tremendous growth in revenue (as depicted below). For example, NVIDIA’s share price rocketed 25%+ or $180+ billion, when it announced earnings far above expectations driven by generative AI demand. Demand for compute could eventually outstrip supply, with Oracle Chairman Larry Ellison saying during an earnings call in March that "there’s actually more demand for AI processing than there is available capacity." That said, the costs to train AI models reportedly continue to improve over time. Even still, these costs and access to these resources pose a potential barrier to mass adoption.
In summary, there are currently several barriers that must be overcome to enable mass market adoption of AI music tools. That said, I’m optimistic that many of these will be overcome eventually, and the probability of the technology having a meaningful impact on the industry is high.
Some Potential Implications & Closing Thoughts
This first essay on AI is already super long and we haven’t even speculated on many of the potential implications of this technology for the music industry. I intend to dig into this more in the future. But let’s touch upon a few potential implications / predictions before signing off.
AI models will be commoditized. Maybe this is a hot take, but I think that we’ll eventually see AI models differentiate based on the quality of their output. And this will be determined by the quality of their inputs. More on that in the next bullet.
Proprietary data will be very important to differentiating AI models, meaning rights holders will accrue significant value. Like we saw with the internet, I expect that creators and rights holders will have their intellectual property rights protected by the law. In that world, licensing copyrighted songs and name, image & likeness (“NIL”) rights will be crucial to producing the best AI outputs. So whereas many Wall Street analysts see AI as an existential threat to major labels, I’m currently of the opposite view. I think that it will eventually become a meaningful new revenue source for rights holders, especially those owning NIL rights.
A deluge of new AI content will catalyze rights holders and streaming platforms to restructure digital monetization. AI technology is helping to democratize the music creation process leading to more and more content on streaming services. In a previous newsletter, we explored how UMG is working with streaming services to improve its artists’ payouts despite this flood of new content. As a result, AI music is helping to accelerate a long overdue change of how digital music IP is valued. Over the next 12 to 24 months, I expect that we’ll start seeing a premium placed on fandom via greater digital scarcity. This may include different royalty rates paid out to and/or pricing tiers for accessing career artists’ music vs. background / mood music. This dynamic reminds me of one of my dad’s favorite sayings: “sometimes your bad luck is your good luck.”
AI music will dominate the background music market. This prediction doesn’t feel too controversial, but I still agree. It seems likely that functional music will be overrun by AI generated music in the next five years. In light of this, Universal Music’s recent partnership with AI music company Endel suggests that the major label has a similar viewpoint.
Most AI music start-ups will fail. Going back to the Jeff Bezos interview cited at the beginning of this piece, Bezos acknowledged that at the time Amazon’s destiny was far from certain and that many internet companies were likely to fail. In my view, we will likely see lasting, important music companies born of the AI era, but Bezos’ prediction is also likely true – many AI music start-ups are going to fail. Just because AI technology may make a major impact on the industry, doesn’t automatically imply most start-ups will succeed. Turning hype into sustainable financial results is extremely hard.
Human artists aren’t going to be replaced. I don’t think AI is going to replace career musicians. I expect that AI tools will play a bigger role in the creative process for more artists going forward. And I wouldn’t be surprised if there is a market for AI-generated songs by virtual artists or if novel AI songs top the charts at some point. At the same time, I expect that we’ll see many professional musicians reject and protest the use of AI tools, like they have in other industries. In an interview with the BBC, the artist Sting warned of the impending culture war between AI and human musicians. Even still, I believe that the vast majority of music listeners are looking for inspiration and connection. For me, this requires a degree of authenticity and storytelling, which I believe will require a human artist being involved in the creative process.
I admittedly may be wrong on many of these predictions. I’m still learning and developing my opinions on the topic. This is definitely an area for more exploration. If nothing else, I hope this piece piqued your interest and made you want to explore the topic more too. That’s why I’m writing this newsletter in the first place – to find a community and learn together.
Thanks to Hannah, and Adam for the feedback, input, and editing!
If Alderbrook can provide you with consulting services or if you are raising capital in this space, please reach out!