My friends Jakub, taetaehoho, and I collectively minted over 1,000 editions of angelbaby’s new NFT drop called “im fine : ( : [Feat. GrimesAI X WesGhost]”. We wanted to share our vision and thinking around the purchase.
The inspiration initially began in mid-April, when a song titled "Heart on my Sleeve" featuring the AI voices of Drake and The Weeknd became a social media sensation, accumulating 20 million streams. The track's creator is an entirely new TikTok user under the pseudonym "Ghostwriter977”.
Within a week of the viral success of "Heart on my Sleeve", popular musician Grimes made an offer to other music creators, inviting them to use her vocals in their songs and proposing an equal 50/50 split of any resulting commercial profits. Seven days later, Grimes doubled down on her claim by posting a full guide on “how to make music [featuring] GrimesAI”. She included an on-chain address called grimesai.eth where any creator can split NFT royalties with the particular address.
Fast forward seven days, and a crypto-native artist named angelbaby collaborated with WesGhost to produce a song using Grimes' AI-generated voice, which they subsequently released on the well-known music NFT platform Sound. The song sold 5589 copies on May 10th, generating a total revenue figure of 7.045 ETH, and is now the third highest-selling drop on the platform.
This music NFT experiment represents the first time on Ethereum where a music creator has ever utilized AI to generate a song using an established artist's voice, while also sharing the financial proceeds with that artist on-chain, all in a permissionless manner.
We believe that a decade from now, this very NFT will be regarded as an artifact of the soon-to-be AI-dominated music industry; an inaugural on-chain experiment that embodies unrestricted creativity from unconventional sources.
AI-powered songs will shape the future of music. By the year 2030, we think that a significant portion of the songs on the Billboard Top 100 will be released by unconventional creators that leverage AI voices. Artist voices will become open platforms at some point in the future [and this is already happening without the artist’s or label’s consent], allowing talented creators to freely utilize renowned identities, and enabling consumers to access an abundance of creative content.
On the supply side, there has been no shortage of consumer markets that have been propelled by high-quality fan-made derivative projects. The manga Death Note, the anime Dragon Ball, and sports-documentary channels such as Joseph Vincent are all examples of realized network effects around fan-made content. This new AI music genre has the potential to create a new wave of talented creators who can rise to fame without the need for Hollywood connections or a typical artist persona. Evidenced by Ghostwriter977’s quick rise to the spotlight, the AI song creation space is an even playing field; everyone has access to the same beats, voice models, and social media distribution channels. Moreover, platforms such as TikTok thrive by prioritizing content that aligns with user interests, irrespective of the number of followers a creator account has. There are already tons of AI song creators who have collectively uploaded hundreds of tracks to YouTube, Soundcloud, and TikTok, garnering tens of millions of views. In reality, the growth of the AI music industry is also partially dependent on the artists themselves and their decision to ultimately give away their voice rights, a decision that AI pioneer Holly Herndon warns not to take lightly.
On the demand-side, AI-generated Drake songs uploaded on YouTube have already sparked excitement among consumers, with many specifically highlighting the quality of the songs and newfound possibility of frequent and abundant high-quality releases. This is a stark contrast to the usual heavy anticipation and lengthy waiting periods associated with the release of new albums by traditional artists.
The image above clearly indicates that AI music is quite enjoyable to some, and this fanaticism will likely continue to grow in size and strength as the voice models get better, and the creators become more skilled. Despite uncertainty around regulations and continuous video takedowns, consumers will always find a way to access content that they desire; 57 million Americans today access music illegally through streaming and pirate sites. We also theorize that general macro trends around narrower consumer attention spans, shorter-lived fads and trends, and a 100x increase in high-quality music will make it harder for traditionally dominant artists to stand out and remain memorable. Artists who are unable to evolve into platforms will eventually be surpassed by the ones who can leverage open networks to significantly increase their production of high-quality music.
The on-chain world will play an instrumental role in the shaping of AI-generated music creation and distribution. Contrary to the current NFT music models of selling plain songs as NFTs [akin to $0.99 iTunes songs] or splitting royalties to various tokenholders [akin to RoyaltiesExchange], we believe that programmatically splitting proceeds on-chain between multiple independent trustless parties represents the first music use case that cannot exist without the blockchain. Permissionless composability is only possible within interoperable systems and open ledgers. Self-custodial wallets, transparent addresses, and the public ledger all uniquely establish provenance and authenticity of songs, AI music creators, and original artists. This ensures that creators are being properly attributed for their music publishings, prevents deep-fake outputs from becoming inappropriately relevant, and allows for trustless transfer of economic value. In the near future, blockchain address activity tied to both traditional artists and AI music creators will become frequently researched by artists, fans, collectors, and music consumers.
Crypto’s permissionless nature allows for the partial elimination of traditionally value-extractive record labels by empowering artists with standardized and protocolized infrastructure. We expect to see a rise in trustless and co-creation friendly registries and DAO tools that will further simplify the collaboration process between artists themselves, as well as their fans, which will lead to brand new on-chain experiments such as continually adaptive collective decision-making processes. Post-distribution, the monetization of music NFT sales now becomes simple through a smart contract that programmatically splits on-chain proceeds.
The integration of blockchain tech and AI-powered tracks is leading to a transformative change in the music industry, fostering a more transparent and accessible ecosystem that empowers creators and artists to exponentially increase the production of high-quality music.
Side note 1: the proof that angelbaby’s song was the first royalty split to grimesai.eth on Ethereum can be found here.
Side note 2: the first royalty split to grimesai.eth on the blockchain occurred on Polygon and was facilitated through riffapp.xyz.
Another side note: angelbaby's music NFT wasn't supposed to be the first AI-generated song with embedded Ethereum on-chain distributions to the original voice artist. A song called "morphotones" had launched and ended its Sound NFT mint one day prior to angelbaby. However, since only 13 NFTs were purchased, none of them were distributed on-chain, as the minimum threshold for distribution is 25 NFTs. As of May 17, the drop still remains on sale.
Thank you to Jakub Rusiecki, Watcher, and Jamie Wallace for the feedback on this short blurb!
Disclaimer: These views do not represent the ones of my investment firm 1kx nor Jakub’s collective The Symmetrical. This article shall not be construed as financial advice. Please conduct your own due diligence before making an investment decision.