Mere hours after the arrival of “Heart on My Sleeve,” the AI-generated “Drake” song that went viral last spring, the doomsday projections began pouring in. Most centered on the relationship between artificial intelligence and music, including reflections on copyright, creative license, and the definition of original art.
The main reason the song generated so much buzz is that Drake is one of the world’s most popular musicians. But part of what gave us pause is that hip hop—which celebrates its 50th birthday this week—is driven by a spontaneity that feels as authentically human as anything humans have ever come up with. That is, rap is a unique form of human language, and if AI can mimic that, maybe nothing is safe.
If the future is already here, then the impacts of generative AI will be even greater in the next decades, especially when it comes to hip hop. In fact, AI will change the way all art is practiced, sold, and appreciated. But technology might also provide novel opportunities for exploration, expansion, and even inclusion. Here we use fictional scenarios to communicate our visions for what hip hop might look like in August 2073.
To most of the world, 20-year old Candice Z. Benoit is known by her stage name: Dice Benoit.
The name carries a double meaning: “Dice” is short for her first name (“Candice”) and is a reference to her place of birth—Las Vegas, Nevada, once a thriving casino city that was home to over a million people. Things changed during the summer of 2054 (just a year after Candice was born), when a dangerous heat wave struck the mountain region. Temperatures averaged 122 degrees Fahrenheit between March and August from 2054 to 2059. Thousands of people lost their lives. Millions left, in one of the first massive climate refugee migrations in US history.
Benoit’s family relocated to the climate-safe haven of Duluth, Minnesota, now a major city in the midwestern United States.
Dice Benoit takes a sip of tea and then picks up two very small devices, placing one gently in each of her ears. Connected to them is a glass visor that rests on the bridge of her nose and covers three-quarters of her face.
It is a headset that gives instant access to an augmented reality-encoded, fully equipped recording studio. On this day, Dice Benoit is not recording a song, but freestyle rhyming in response to prompts from a program that generates images at random. The rhymes to these prompts will serve as training data for an algorithm—called a “Jewel” after the Julia coding language used to create the original iteration 30 years prior. Jewel will build an algorithmic battle rapper, an AI-generated version of Dice Benoit’s rhymes, ready to participate in a live rap battle with other Jewel-generated MCs.
The key to the battle circuit is the the rapper’s training data set. The better the training set, the better the Jewel-inspired rhymesayer.
Battles take place in virtual space AI has designed to resemble the urban cyphers of the mid-1990s, a time that only those pushing 100 really lived through: the raucous crowds, the DJ, the artists pacing back and forth like prizefighters, the classic zingers, the verbal sparring, the raw energy of a (then) relatively new art form. AI can recreate these settings with ease.
The rappers in this virtual setting are humanoid avatars, dressed however the author wants. Dice’s avatar is genderless and wears a mask (resembling the one adorned by MF Doom, one of her grandfather’s favorites), with an AfroFuturist bodysuit similar in style to the one worn by the fictional Killmonger in the 2038 reboot of Marvel-DC Studios’ Black Panther.
An hour later, Dice’s avatar arrives on stage. Her opponent in the first-round battle? A rapper named Mephisto from southern Florida (now mostly submerged after the storms of 2044).
Just as the MCs are preparing for their introductions, an error message appears:
BYTER DETECTED
The virtual stage disappears. The ’90s battle setting vanishes. The screen turns black.
Dice, at home in Duluth, sinks into her chair. She hasn’t seen one of these in a long time. Frustrated, she pulls off her headset, awaiting the verdict. Is Mephisto guilty of byte-ing?
In 2073, byte-ing is a serious crime, not only in the AI rap battle space, but in all of art. It happens when a human artist portrays a purely AI-created relic as having authentically human elements (often training data). Severe infractions can yield significant jail time, depending on the scope.
It turns out Mephisto had trained his Jewel rapper on AI lyrics, rather than on his own freestyles. Mephisto is a byter.
Byter laws were implemented in 2030 after being passed by the Senate and House (with majority support from all four major parties: Democrat, Republican, Techno-Green, and Techno-Libertarian) and approved by the International Committee on AI, Art, and Creative Protection (ICAIACP), a body convened to protect human authenticity in an age of rampant artificial fabrication.
The laws were not passed without controversy. Segments of the population still oppose government intervention in technology, and the debate has raged for several decades. A consensus emerged, however, that even with the advanced capabilities of AI, there was a desire to protect human authenticity, that there was something essential about human expression that warranted protection. Part of the motivation for this was the emerging aesthetic appreciation for human creation. In the 2040s, creations and archives of famous human artists and thinkers escalated in value. The Nasir Jones archives (containing his masters, rights, unreleased music, original notebooks, emails, transcripts of conversations, and others) sold for a rapper record of $4.8 billion in 2051. The age of AI has led wealthy people to appreciate human-created relics to the tune of a quadrillion dollars.
As with the rise of any lucrative industry, cheaters have emerged. Art fabrication and associated crimes have existed for hundreds of years, long before the invention of the personal computer. The difference in 2073 is that technology has transformed fabrication into a science: It is harder than ever to tell the difference between human and machine.
How, then, is byte-ing detected?
There are AI programs designed specifically to identify human signatures in creative relics. In music, art, literature, and other forms, there are subtle ways to identify human touch and AI intervention. Institutions, governments, and corporations all use byter-detection software to protect the authenticity of their human-inspired creations. This dynamic has created an arms race between the byters and the byter detectors.
These dynamics hark back to the late-20th-century race between performance-enhancing drugs and the tools to detect those drugs (the sports saga ended in 2044, when all forms of enhancement were rendered legal, except genomic enhancement).
In 2073, society is caught in the midst of a battle between AI-generated and human-authentic creativity, with no end in sight.
The case against Mephisto was eventually settled, with the artist fined and banned from the battle circuit. Dice Benoit advanced to the semifinals, where she (her AI Jewel) was eliminated by a rapper named Diogenes (from Athens, Greece), who somehow wove Greek philosophy and martial arts into a battle tapestry that was hard to beat.
But while Dice Benoit is well respected on the battle circuit, she is also considered one of the most talented recording artists of her generation. She makes decent money from battles, but her long-term goal is to secure a lucrative deal with a music conglomerate.
So far she has been unable to, so her next option is to participate in a live art auction. As with any auction, there are risks: She could have taken a deal from Zora Records (0.4 cents on the dollar, standard for the day). But all of the major players in AI music will be at this auction. She is betting on herself—and on the corpus that her genius has generated.
In 2073, human musicians are signed to record labels based on their training data. This data, not unlike the Dice lyrics used to build the Jewel battle rapper, are used to generate original music. And training data ends up being the way that artists are properly remunerated for their original creations: They negotiate an advance payment and a per-dollar rate based on how much money is made from songs using their data. Dice Benoit is hoping to secure an unprecedented deal for a rapper: 1.5 cents on every dollar. This is more than five times the compensation of many successful rappers.
The virtual auction is set up in the style of its counterpart from a century prior. Companies (represented by humanoid avatars) sit in an auditorium, each carrying an auction paddle. An auctioneer sits at the front of the room, speaking in the rambling auction chant and pointing to companies holding up paddles when a bid is made.
Before the start of bidding on an artist, their representatives (sometimes the artists themselves) give a five-minute presentation on the value of that training data.
At this auction, the artists range in quality, but all are impressive in their ambitions. In 2073, rap is considered the highest among the fine arts, so the hip hop artists are the last to go to auction.
Many talented artists are represented: a rapper with Trisomy-21 named Jirau (after a pioneering fashion model), who is an especially strong lyricist. In 2073, neurodiverse artists (many formerly known as “disabled”) are valued, as most AI algorithms had largely ignored them in their early training sets.
Then there’s DJ Congolia, who made original beats exclusively from the sounds of living things in the Congo rainforest (near where he was born and raised).
And, of course, Dice Benoit.
Sitting at home in Minnesota, headset on, she exhales in anticipation. In the virtual auction, her avatar rises on stage, prepared to pitch.
Her presentation is compelling. It focuses on how she can rhyme about everything: politics, climate, prison, religion, war, famine, gaming, family, friendships, love. She is a strong candidate for a good training set because a company can take her data and build rap music of almost any kind: summer love ballads, political campaign slogan-songs, flying car commercials.
Dice Benoit isn’t entirely unique in her topical diversity: Long gone are the days when some hip hop artists felt the need to center misogyny, material consumption, or any other singular tropes. The reason? In the late 2020s, AI became so advanced that it was able to generate songs with that sort of content far more convincingly than any human artist could. Replicating old ideas fell out of favor, while novelty was rewarded—fresh data means new music that more people can relate to, which translates to more money. This arithmetic led to the destruction of older categories and a rapid flowering of new rap subgenres around the world.
Some of the most lucrative and popular rap scenes of 2073 include those dedicated to Angolan literature, astrobiology, cyber theft, Harriet Tubman, Jack Kerouac, Maori history, Santeria, sex work, space tourism, and Sufism. And new subgenres are born regularly.
Further, AI made language barriers a thing of the past. In 2073, a rapper can generate a rhyme in one language and have it translated into a hundred others in seconds. The challenges that have plagued translation forever—not all things translate well between languages—remain, but AI can optimize as well as anyone ever has. Rappers from the west side of Chicago have large fan bases all over the world who now listen to the lyrics translated to Bengali, Tagalog, and Ibo (AI ensures that despite the translation, the music remains enjoyable).
This AI ecosystem rewards human artists who are productive generalists, who can generate as much quality training data as possible. Musicians who play multiple instruments well can generate training data for an entire orchestra, which improves their financial gain. In hip hop, artists who can rhyme about a lot of things—like Dice Benoit—are poised to do well.
This perspective is not without controversy, however, as some lament the loss of specialization that had dominated the art world for centuries. But the excellence of AI technology has rendered this perspective less popular. The truth is, no human could ever hope to specialize the way AI can (a lesson the world first learned when IBM’s Deep Blue defeated Kasparov in chess in 1997).
In 2073, humanity’s hope for influence resides in our ability to do the one thing that AI has a little bit of trouble with. The past century has taught us that no one sells surprise like rappers can. And this is what Dice Benoit believes might make her skill set special, and worthy of a large contract.
In these fictional anecdotes, we depict a single universe out of many possible universes, one in which AI profoundly influences everything about the way music is made. But we also observe that AI need not spell the end of the human creative process, but could rather offer new modes of creation and new challenges for how artists are perceived (and compensated).
The musician BT, a leading voice on the topic of the interaction between AI and music, recently spoke on the matter: “I split artificial intelligence into two categories, generative and assisting. There are technologies that will do things for us, or do things autonomously, and then there are things that will assist us in our daily lives.”
In our fictional 2073, we see a world struggling with this dichotomy. But our 2073 society also made the aesthetic decision to appreciate human-made art and relics that could be assisted by AI.
It’s all science fiction, but it communicates the step that we believe is missing in many apocalyptic takes on the future of creation. As with many human inventions—from the wheel to the hydrogen bomb to the personal computer to large language models like ChatGPT—we can decide what we value.
Challenges with AI will persist, especially around labor and economic inequality. Surely, the notion that technology alone will solve humanity’s problems is naive and baseless. But through a fictional examination of AI’s potential impact on hip hop—one of the latest and greatest of human creative endeavors—we consider the possibility that we have more agency than we think, that the war against the robots may not play out as some predict, and that the end may not be as near as many believe.
