While machines have helped us make music for decades does the equation change when artificial intelligence becomes a band member?

---

“All this machinery // making modern music // can still be open-hearted // not so coldly charted, it’s really just a question // of your honesty.”

That’s from Rush’s 1980 hit single, “Spirit of the Radio.” When I first heard those lyrics, I thought they were a critique of synthesizers – the rock resistance to the emerging apostleship of Kraftwerk and Co. 

Many will remember the infamous moment when Dylan went electric, another example of people holding tight to tradition, or perhaps it’s some deep-rooted belief that anything plugged into the wall somehow mitigates the authenticity of our work. But in both cases, resistance was razed and the artists who risked censure were memorialized as our fearless guides.

In actuality, “Spirit of the Radio” is a lyrical lament on radio’s sweeping shift from free-form programming to commercial formats [editor: I had no idea about this]. But in the context of today, this stanza could just as well have been a prescient review of AI’s role in music. 

In a few decades, how will we regard tech like ChatGPT, the now infamous interface OpenAI released alongside its new GPT (generative pre-trained transformer)? As the next great creative instrument, or – as sci-fi has long warned us in sundry tales of robot rule – as another signal that our existential dread was warranted all along. Time will tell, but for the time being, let’s take the latter route and explore how the tool might help musicians better understand web3.

---

In 2020, I interviewed Chasson Gracie, a data scientist and music documentarian whose film The Music Sounds Better with Whom? won Best Documentary at the Toronto Shorts International Film Festival. The film investigates the intersections of music and AI, specifically exploring the implications of three affected areas: the industry (e.g. labels and marketing platforms), artists and fans.

At the time, Grimes had just made headlines for claiming, “I think live music is going to be obsolete soon.” Holly Herndon and Zola Jesus were two of the more notable musicians that pushed back. The former invoked her new AI-infused album, PROTO, to showcase its potential to supplement human art, rather than replace it, and Jesus stamped her own lengthy rebuke with: “you can’t automate enlightenment.”

A lot’s changed in two years. Deep fake technology – one of the more visible advancements – is terrifying, challenging the ownership and representation of our very identities. In mid-2021, Herndon started Holly+, a decentralized autonomous organization (DAO) that imparts ownership of her vocal model to her community – in part to defend her own voice from time. 

“In the event of my death, I feel more comfortable with distributed ownership of the rights to my voice model among a DAO of stewards who are invested in maintaining the value and reputation of my voice than the rights being passed down to someone less familiar with the values and standards associated with my work,” she wrote. 

And ChatGPT has introduced a slew of new concerns to the mainstream, from plagiarizing book reports to red alerts at Google, a company as imperishable as they come and yet critically concerned at the conversational AI’s implications on its ad business (why search for an answer when you can just ask your ad-free robot friend). Conversely, Microsoft just invested $10 billion more into OpenAI, and though the latter is governed by a non-profit entity, it’s hard not to see the move as a power play (one caveat of the investment is that Azure remains the technology’s exclusive cloud partner).

Read more: Taking the Nina Protocol’s Music Publishing Service The Hub for a Spin

The seminal research DAO Water & Music recently held a ChatGPT workshop, asking the program to help generate artists strategies with questions like what to look for when booking a tour, writing venue contracts and risks involved with 360 label deals. They even had ChatGPT write the event invite, iterating through drafts with inputs like “can you sprinkle in a slightly sarcastic existential tone?” Generously, it acquiesced and delivered a quippy invite intro: “Are you ready for the robot uprising?”

---

Water & Music is transitioning its members-only newsletter DAOnload to focus on AI. They also recently built a database of Creative AI tools for artists, where machine learning can facilitate everything from songwriting to mastering to speech synthesis.

When AI fails to distinguish between copyrighted and open data in its modeling, there are legal implications, and already this year there have been two major lawsuits against popular visual AI tools like Stability.ai, DeviantArt and Midjourney. Across the music world, comparisons to piracy are inevitable, and the Recording Industry Association of America (RIAA) has already cracked down on some AI tools.

Water & Music founder Cherie Hu posits that lawyers have more power in the music industry than in any other, and that any attempt to train machines on vast amounts of non-public music data would be quashed in short order. Her Twitter thread is an incisive breakdown of why music won’t go the route of images when it comes to AI.

But let’s give this an optimistic spin – after all, nary one of us still carries an aversion to synthesizers and stratocasters: how can chatGPT be a net positive for music? Set aside the grifters and cheats – what does chatGPT look like when we assume good will?

MIDiA research asked a similar question and proposed independent artists use the technology as a music business assistant – to make a point, the first part of the piece is even surreptitiously “penned” by ChatGPT (nothing has been in this piece, yet…). The industry can be defeatingly complex, after all, so why not form a basis of knowledge – it’s a means of catch-up.

Now add web3 on top of that, another tortuous ecosystem with a tall barrier to entry. That friction is of foremost concern, to the point that education and onboarding have become focal points of nearly every sizable music organization in the space. So why not leverage ChatGPT for support? 

To test it, I prompted: “Create a community building strategy for music artists who want to onboard into web3.”

The results were near-instantaneous and uncanny:

“Build a presence on decentralized social media platforms such as Steemit, Hive, and Minds, where music artists can connect with existing crypto and blockchain communities.

Utilize blockchain-based music streaming platforms like Audius, Choon, and Musicoin to distribute and monetize music.

Host live streams and Q&A sessions on platforms like DLive, SuperRare, and Sorare, where artists can interact with fans and showcase their music in a unique and engaging way.

Offer exclusive content and merchandise for sale on marketplaces like OpenSea, Rarible, and SuperRare, using non-fungible tokens (NFTs) to prove ownership and authenticity.

Collaborate with other web3-based music artists and projects to create a sense of community and shared mission.

Develop a (sic) educational content and resources to help educate artists and other stakeholders on the benefits and how-to of web3 and crypto-based monetization.

Create a community forum or Discord group where artists and fans can connect, share ideas, and collaborate on projects.

Utilize web3-based analytics tools to measure engagement and track the success of the community building efforts.

Overall, the key is to create a sense of community and shared mission around the use of web3 technologies in the music industry, while also providing tangible benefits and resources to help music artists succeed in this new ecosystem.”

The guide may be simplistic and lacking some of the technical nuance and implementation details necessary to truly get on-board, but it’s an incredible foundation.

“This is great, thanks,” I responded, feeling the need to express gratitude. “What's the primary difference you see between creating a web2 music artist strategy versus creating a web3 music artist strategy?”

The response, again, was so fast and thorough that it was as though the machine had prepared a draft well ahead of time:

“The primary difference between creating a web2 and a web3 music artist strategy is the use of decentralized technologies and blockchain.

Web2 music artist strategies typically focus on leveraging centralized platforms such as Spotify, Apple Music, and YouTube to distribute and monetize music. These platforms control the distribution and monetization of the artist's content, and often take a significant percentage of the revenue generated.

On the other hand, web3 music artist strategies utilize decentralized technologies such as blockchain and smart contracts to distribute and monetize music, allowing for a more direct relationship between the artist and the audience. These platforms are often more transparent, and the artist has more control over their content and revenue.

Additionally, web3 music artist strategies also have the potential to use non-fungible tokens (NFTs) to monetize exclusive content and merchandise, providing new revenue streams for the artist.

In summary, web3 music artist strategies are focused on using decentralized technologies and blockchain to create a more direct and transparent relationship between the artist and the audience, while also providing new revenue streams.”

It’s important to note that as a language model, ChatGPT’s knowledge cut-off is 2021, so it won’t (yet) provide bleeding edge insights

Again, it’s a simplistic overview, but its efficacy as a primer to this world is powerful, especially when more pointed follow-up questions yield more nuanced answers. It’s important to note that as a language model, ChatGPT’s knowledge cut-off is 2021, so it won’t (yet) provide bleeding edge insights, and given the pace of web3, it will be important – as (hopefully) always – to speak with people. 

Consider this a silver lining amidst a sea of trepidation. This is an incomplete tale, but let’s hope the context of history looks kindly upon this moment, and it’s more ‘Dylan goes electric’ than ‘robot uprising.’ With the right guides, all this machinery making modern music can still be open-hearted.