As AI drum machines embrace humanising imperfections, what does this mean for ‘real’ drummers and the soul of music?

There’s a moment five minutes into ‘Funky Drummer’ (1970), an instrumental jam by James Brown, when the clouds part and Clyde Stubblefield is left alone. We can hear on the recording Brown instructing his band to ‘give the drummer some’. He tells Stubblefield not to solo, but to ‘just keep what you got’. Even if you’ve never heard the original, you will have heard Stubblefield’s drum break. The looped sample has been used on more than a thousand other tunes. His right hand is playing semiquavers on the hi-hats throughout, with his left foot opening the cymbals to produce an occasional offbeat whisper. His right foot on the bass drum and left hand on the snare are in a conversation. The backbeats on the second and fourth beat of each bar are decorated with what drummers call ‘ghost notes’ on the snare drum, more felt than heard.

In principle, it would be perfectly possible to take each semiquaver, transcribe it, pull the notes from the stave, use readily available software to program them into a grid and fully automate the funky drummer. The beat is repetitive. Drumming is all about patterns, and computers are very, very good with patterns. And yet there is something ineffably human about this performance. The dance of his limbs, the bounce of his sticks and the movement of the air inside his drums combine to produce something undeniably musical. I think a drum machine couldn’t get close. But maybe not everyone cares as much as I do about the nuances of percussion.

I am a professor of science and technology studies. I am interested in social attitudes to new technologies, particularly those involving artificial intelligence. I am also a part-time mediocre drummer, relieved not to have to rely on my musical talents to pay the bills. We drummers tend to be ambivalent about technology. Like most musicians, ours is a craft that is technologically mediated. The affordances of sticks, pedals and things to hit with them enable our sound. We are used to the jokes that suggest we lack the intelligence of our fellow musicians. (What’s the difference between a drummer and a drum machine? You only have to punch the information into a drum machine once.)

We worry that our bandmates, presented with technological alternatives, might look on us as a problem to be solved. We are loud; we take up space; our instruments are heavy and slow to assemble; our sounds are harsh and inconsistent, and sometimes we speed up or slow down when we play. Faced with a drum machine that keeps metronomic time, plays no more or less than is asked of it and, once purchased, costs nothing, we can’t help but feel judged: is that all you think of us? Is that thing all it takes to make a drummer redundant? Practitioners with hard-won skills will, according to the sociologist Howard Becker, ‘resist the new both because [they] find it aesthetically repellent and thus morally outrageous and because [they] stand to lose if it replaces the old’. But before we rage against the drum machine, we should seek to understand its origins and its potential. I think that, in studying the evolution of this technology, we can learn something important about automation, the future of artificial intelligence, and what it means to be human in the machine age.

The ‘machinery question’ dates back at least to the Industrial Revolution. Following the Luddite uprisings in early 19th-century England, the economist David Ricardo wondered whether workers, who were concerned that the growing automation of weaving and other trades would leave them jobless, might have a point. Since then, economists have reassured themselves that, by increasing productivity, new technologies on average create more jobs than they destroy. However, for the weavers, laundresses or lamplighters whose skills were made redundant and who had little prospect of retraining, the averages offered little comfort.

More recently, the rise of artificial intelligence has led some to conclude that the robots are also coming for the jobs of the middle classes. A paper by Carl Benedikt Frey, an economist, and Michael Osborne, a computer scientist, garnered headlines for its claim that 47 per cent of US jobs were at high risk of automation. These jobs didn’t include just receptionists and drivers, but also paralegals and accountants. Their assessment is simplistic. It swallows the hype about AI. For a technology to rapidly transform the world, the world needs to change too, and this change is often slower than either technological optimists or Jeremiahs would imagine. Frey and Osborne’s analysis also buys into a familiar narrative of AI, of a competition between humans and robots. We enjoy the spectacle of computers beating humans at chess and Go, and wonder whether computers might soon win games like the Turing test. The reality is that AI works very differently from human intelligence, and technology need not be very good for human bosses to make use of it. Robots don’t take people’s jobs; people do, often employing what Brian Merchant, in the Gizmodo article ‘Why Self-Checkout Is and Has Always Been the Worst’ (2019), calls ‘shitty automation’. The experience with supermarket self-checkouts is not one of automated superiority; it’s of customers slowly learning how to work around and compensate for the limited intelligence of a robot.

For music, Trevor Pinch and Karin Bijsterveld describe new technologies as generating ‘breaches in musical culture’, revealing social norms that we may otherwise have taken for granted. Technologies, which are often designed to solve particular problems, provide new means to human ends. But they also enable new ends. A good music technology becomes a new source of creativity; it can be hacked and bent in ways that its own designers may not have foreseen. Technologies change our collective sense of what’s possible and, in doing so, they change our sense of how things should be. A drum machine does more than automate a beat. In the right hands, it can change our sense of rhythm.

One note on its own is just a note. Put it on top of or next to another and you can start to build melodic or harmonic relationships. For rhythm, you need enough notes arranged in time to establish a pattern. The notes make sense in relation to what came before. Repetition gives sounds purpose. As Elizabeth Hellmuth Margulis has argued in Aeon: ‘Repetition serves as a handprint of human intent. A phrase that might have sounded arbitrary the first time might come to sound purposefully shaped and communicative the second.’

Rhythm is based on expectations. Rather than reacting to beats, we anticipate them. We dance, clap or nod our heads along to well-placed, predictable notes. This entrainment is more than human nature. Scientists have found that cockatoos, parrots, sea lions and rats all dance to repetitive beats (dogs and most other animals apparently don’t). The shared experience of rhythm can, in some circumstances, be downright subversive. In 1994, the UK Criminal Justice and Public Order Act responded to a moral panic about young people attending raves by giving the police new powers to shut down gatherings involving music ‘predominantly characterised by the emission of a succession of repetitive beats’.

The drum kit became an ergonomic instrument, designed around the body of a drummer

Without some dynamics, harmonic interest or psychoactive stimulants, the comfort of repetitive beats quickly becomes boring. A rhythm also needs elements of tension or surprise. We call the placing of a beat where the ear doesn’t expect it ‘syncopation’. (In the late 19th-century United States, it was called ‘ragging’, giving ragtime its name.) Variations within and between bars take listeners and dancers to new places. James Brown was known for rigidly enforcing his band’s parts, but each bar that Clyde Stubblefield plays in ‘Funky Drummer’ is subtly different from the one preceding it.

Good drumming is about more than staying in time. Musicians often talk about a good drummer’s ‘feel’. A drummer’s expertise may be impossible to describe. Like riding a bicycle, the relevant knowledge is tacit, held in the body as much as the brain. A rhythm is enabled and constrained by the interaction of limbs and instruments that give music a feel. In a Brazilian samba bateria (percussion section), the loose jingles of the chocalho naturally fall behind the beat while the tamborim (a small drum played with a small stick) is played ahead of the beat, driving the music forward. A range of samba instruments might all play semiquavers, but the semiquavers are not all evenly spaced. The ‘swing’ of a Brazilian samba gives the rhythm a galloping feel. This elasticity can be thought of as the rhythmic equivalent of a ‘blue note’, which is sung or played in the gap between the strict semitones found on a piano keyboard. It is also heard in a Viennese waltz, in which the second beat of three is traditionally brought forward to give the music a lilt. Classical musicians know this as a form of ‘rubato’ ­– stolen time. Time is taken from one note and given to another, with the music still ‘in time’.

In New Orleans, the unmistakable feel of a second line marching band is produced by multiple drummers. The invention of the drum kit in the early years of the 20th century combined these roles into one labour-saving bricolage. William Ludwig patented a bass drum pedal in 1909. Another pedal, normally played with the left foot, brought together two ‘sock’ cymbals on the floor, but this was quickly superseded by a ‘hi-hat’, which kept the pedal but lifted the cymbals so they could also be played by hand. The addition of a snare drum helped normalise a one-two, bass-snare, stomp-clap rhythm of downbeats and backbeats. Over the course of the 20th century, the tools and techniques of drumming co-evolved. Drumming is physical, and the drum kit became an ergonomic instrument, designed around the body of a drummer, decorated by cymbals and tunable tom-toms. Plastic drumheads were invented that could be tuned much higher than earlier calfskin ones, giving Stubblefield’s backbeat its ‘crack’.

During the early 20th century, drummers and their contraptions or ‘trap sets’ became vital for holding together dance bands and providing musical decoration. Matt Brennan, the author of a new history of the drum kit, describes how drummers encountered various opportunities and threats. Brennan quotes the in-house journal of the Ludwig company in 1930: ‘A drummer in a dance band can do more toward making, or breaking things than a whole room full of saxes, trumpets, and what not. A drummer has responsibilities.’ The biggest disruption was the introduction of talking pictures into cinemas, which led to the rapid unemployment of more than 18,000 cinema drummers, who had previously provided sound effects and music. This led the American Federation of Musicians to create the Music Defense League against what came to be called ‘canned music’. But it was just one wave of creative destruction wrought on music by innovation.

John Philip Sousa was the most popular US musician of his day. In 1906, he wrote an essay on ‘The Menace of Mechanical Music’ that took aim at the disruptive technology of the time, phonographs, which ‘reduce the expression of music to a mathematical system of megaphones, wheels, cogs, disks, cylinders, and all manner of revolving things’.

Sousa admitted that his opposition to the phonograph was partly economic. His business model was built on his US Marine Band’s relentless tours and on selling the sheet music for his many marches. Recordings of his music, as they were not yet protected by copyright law, would earn him nothing. But Sousa’s principled argument was about the soul of music, which he believed came through its teaching, learning and performance. ‘The nightingale’s song is delightful because the nightingale herself gives it forth,’ he wrote. Sousa likened the phonograph to the invasive English sparrow, which had, decades earlier, forced American songbirds from their homes: ‘what of the national throat? Will it not weaken?’, he wondered, ‘What of the national chest? Will it not shrink?’

Stories of musicians’ reactionary responses to technology go back centuries. Harpsichordists rejected the piano for its weedy attack. The inventor Johann Nepomuk Maelzel, who made ear trumpets for Ludwig van Beethoven to help with the composer’s deafness, also invented the metronome, a device of which Beethoven once said: ‘Whoever has the right feeling, needs none; and whoever lacks it, has no use for one – he will run away with the whole orchestra anyway.’ In the 19th century, flautists who used the pads of their fingers to shape and bend notes were concerned that the invention of a flute with linked mechanical keys would take away the instrument’s nuance. From the start of the 20th century until the arrival of the phonograph, player pianos and reproducing pianos brought virtuosic performances, recorded on paper or metal rolls, into middle-class living rooms.

Stravinsky chose to treat the pianola as a new instrument rather than a caricature of a piano

The pianola was wildly popular, but generated outrage. One European critic argued in 1929 that music required communication ‘from soul to soul, a communication that the machine is, and ever will be incapable of creating’. But while some were horrified that the pianola would devalue music and lead to commoners mimicking the ‘cultivated class’, modernists saw new musical opportunities. The composer Arnold Schoenberg made the case for surfing rather than being drowned by waves of new technology:

[T]he objections touched off by the rather provocative expression ‘mechanisation of music’ collapse when one realises how much mechanisation has taken place in our most important instruments … [T]hink simply of the clarinet’s keys, the horn’s valves, the harp’s pedals, the guitar’s frets, and finally the very scroll of the violin, and then decide whether we can do without the mechanical element in our tools for producing sound and whether it has made music worse. It is sentimental to wail about mechanisation and unthinkingly believe that spirit, so far as it is present, is driven out by mechanism.

The pianola freed avant-garde composers from an important constraint: the reach and number of a pianist’s fingers. Igor Stravinsky was one who chose to treat the pianola as a new instrument rather than a caricature of a piano. Replicating the polyphonic spree of his composition Étude pour pianola (1917) live requires three human pianists.

Perhaps the most transformative recent music technology has been Auto-Tune. The human voice, while it is able to produce tones of great beauty and emotion, is an unreliable instrument. For singers who are untrained, unable to hear their own flaws, in possession of weedy vocal cords, or just having an off day, Auto-Tune provides a quick fix. The patent for Auto-Tune is very clear on the invention’s purpose: ‘When voices or instruments are out of tune, the emotional qualities of the performance are lost. Correcting intonation, that is, measuring the actual pitch of a note and changing the measured pitch to a standard, solves this problem and restores the performance.’

The first uses of the technology were not to fix music, but to provide a novelty effect, much like its forebear, the vocoder. Since 2000, however, Auto-Tune has become a ubiquitous studio tweak. Time magazine labelled Auto-Tune one of the world’s 50 worst inventions. Catherine Provenzano describes how Auto-Tune raised concerns that ‘it is not just the musician being deskilled but the listener, too’. Critics worried that its turd-polishing would take even more power from musicians and hand it to producers. Software like Auto-Tune requires elaborate programming to reshape the complex waveforms that emerge from human vocal cords. Compared with this, a drum machine is as easy as 1, 2, 3, 4.

It is possible that the first programmable robot was a drum machine. In his work The Book of Knowledge of Ingenious Mechanical Devices (1206), Ismail Al-Jazari describes a model boat, upon which sits a band. Every 30 minutes, a waterwheel powered by drips from a bucket would bring the band to life by turning an axle with pegs that would activate the musicians’ limbs. The two drummers would play their instruments according to the placement of the pegs.

Since Al-Jazari’s boat, there had been various toys, music boxes and experimental devices that emitted rhythmic noise. Aside from the invention of recorded music, however, no real existential threats for drummers emerged until the 1960s. (In the 1930s, Leon Theremin, inventor of the eponymous device, was commissioned to create a Rhythmicon, which generated polyrhythms too complex for a human percussionist, but this novelty was of little use to the working musician.)

The first recognisable drum machine that an ordinary person could buy was a mahogany cupboard called the Wurlitzer Side Man. Inside, a rotating arm, which could be adjusted according to tempo, dragged metal strips across evenly spaced contacts on a concentric sequencer, activating noises according to preprogrammed foxtrots, tangos or waltzes. Its cream Bakelite buttons had labels like ‘bass drum’, ‘wood block’ and ‘maracas’, but its tones were unmistakably synthetic. The Side Man’s name announced its inventors’ job-stealing intentions. A newspaper advert promised ‘a full rhythm section at your side’. An enterprising organist could become a one-man band. The UK Musicians’ Union, on hearing the Side Man, was outraged on behalf of its drummers, and labelled the machine a ‘stilted and unimaginative performer’.

By the end of the 1960s, drum machines were offering producers psychedelic textures good enough to appear on records. Sly Stone, who was prone to alienating his band members, decided that he didn’t need his drummer, Greg Errico, and could make do with a Maestro Rhythm King, which could reproduce a range of preset ‘traditional’, ‘American’ and ‘Latin’ beats. Stone called it his ‘funk box’. There’s a Riot Goin’ On (1971) was the first major record to feature a drum machine. Stone, Herbie Hancock, Stevie Wonder and others who had impeccable musical taste and access to the best drummers money could buy, were excited more by the novelty of the machines’ sounds than by their consistency.

In 1977, the disco producer Giorgio Moroder, working with Donna Summer, used a Moog synthesiser to create a synthetic drum track for ‘I Feel Love’. The synthesiser was unable to recreate a disco-appropriate bass drum sound, so a drummer contributed a simple four-on-the-floor using his right foot, but the tune, according to David Bowie, made Brian Eno remark: ‘I have heard the sound of the future … This single is going to change the sound of club music for the next 15 years.’ New Order were one of the bands inspired by Donna Summer. Their tune ‘Blue Monday’ (1983), using an Oberheim DMX, became a drum machine standard after their initially sceptical drummer, Stephen Morris, one of many to acquire the nickname the ‘human drum machine’, was persuaded to embrace the technology. The band’s singer, Bernard Sumner, said of ‘Blue Monday’: ‘I don’t really regard it as a song. I kind of regard it as a machine to make people dance.’

The sonic grandchild of the Side Man was the Roland TR-808, the machine that made the ultra-modern sounds of computerised percussion affordable. First sold in 1980, but more popular later, the 808 introduced a new vocabulary to late 20th-century music. Most usefully, its distinctive booming bass drum gave producers of electronic music the power they needed to move dance floors. Early adopters of the 808 included Marvin Gaye and Phil Collins, both of whom began their musical careers as drummers.

When the pianist Fats Waller was asked what swing was, he replied: ‘If you gotta ask, you ain’t got it’

The arrival of the Linn LM-1, also in 1980, gave musicians an authentic-sounding, programmable drum machine, albeit at a cost. Roger Linn, its inventor, was a guitarist and songwriter, looking for an easy way to record demos and make them sound as real as possible. His machine, and the upgrades that followed, became popular with producers. Linn emphasises in interviews that his intention was always to bring out human creativity rather than make drummers redundant. He was frustrated that the first big LM-1 hit was the Human League’s robotic-sounding ‘Don’t You Want Me?’ (1981).

The advent of digital recording meant that Linn could record the best drums, using the best studio, and save the sounds onto microchips. Bass drums and snare drums were easy; cymbals were more complicated because their long delay times used up more memory. The high price of computer memory forced early drum machine designers to divide beats into equal sections, giving few options for the placement of sounds. This limitation was sold to customers as a feature. Linn started by allocating one byte to each semiquaver, which meant that when he input beats to his machine, they were automatically shifted to the nearest beat. In the manual for the LM-1, Linn labelled this serendipitous effect ‘auto-correct’. It would later come to be called ‘quantising’, with sounds being automatically shifted onto the nearest line of a ‘grid’ to standardise their placing.

By the time of its release, the LM-1’s high-spec RAM gave it a resolution of 24 parts per quaver, giving Linn a way to breathe life into his robot. The manual described how users could tweak their beats ‘on to time slots that make the part “feel” more human’. Rather than a quaver split evenly in half, which would give strict straight time, Linn’s ‘swing’ function could delay alternate notes to give the rhythm a more interesting gait. The amount of swing could be set as a percentage, with 50 per cent being straight time and 66 per cent shifting each alternate note so it was two thirds of the way to the next note, sometimes called a shuffle feel. (Think about the difference between ‘Billie Jean’ and ‘The Way You Make Me Feel’, both by Michael Jackson and both recorded at about 115 beats per minute. The former is in straight time; the latter is a shuffle.)

A jazz purist would consider ‘swing’ a quality that was hard to describe, and impossible to express in terms of maths. The story goes that, when the pianist Fats Waller was asked what swing was, he replied: ‘If you gotta ask, you ain’t got it.’ Swing comes from music when it is played rather than when it is written down. In the history of jazz drumming, Jo Jones, Max Roach, Elvin Jones, Tony Williams and other greats have each had a distinctive swing to their playing, and they typically swing more than the soloists they are accompanying. On old Count Basie recordings, it’s often hard to hear Jo Jones’s drums, but you can hear the effect he’s having on how the band swings.

Scientists have sought to understand how the microrhythmic displacement of beats can give music a particular feel. ‘Funky Drummer’ has, according to one study, a swing ratio of 1.1:1 (in Linn’s terms, this would be 55 per cent swing), meaning that the semiquavers are not quite isochronous. Odd numbered semiquavers are fractionally longer than the even ones. The drum part also has a slightly delayed backbeat (beats two and four of the bar). The jazz pianist and scholar Vijay Iyer has argued that the ‘microdelay’ of a backbeat relative to beats one and three gives this sort of music a relaxed feel, sometimes called playing ‘in the pocket’. The ethnomusicologist Charles Keil goes further, arguing that, for music to have social meaning, it must be imperfect: in some sense, ‘“out of time” and “out of tune”’.

While jazz musicians may have been sceptical of the machine swing of the LM-1, some like Herbie Hancock embraced the machine for their 1980s fusion experiments. By the end of the 1980s, drum machines were ubiquitous, being used in rehearsals, recordings and on stage. Each device came with an implicit idea of what a drummer is, what a drummer does and what drums could become. Though the machines had musical limitations, they gave producers a new set of possibilities. Producers were no longer limited by what their drummer could do. They could rethink the role that rhythm played in a tune. Musicians and producers came to know, and occasionally to anthropomorphise, their drum machines. Some insist that the circuitry of even strict grid-based drum machines produces a distinctive feel. In researching this piece, a friend who is a music producer sent me a WhatsApp message: ‘I have never heard a drummer that sounds like they’re having as much fun as my 606 [a cousin of the 808]. That little guy loves to drum and you really feel it.’ For real drummers, the rise of the drum machine has presented some interesting challenges.

As drum machines were becoming more human during the 1970s and ’80s, human drummers were becoming more robotic. Electronic drums expanded drummers’ palette of possible sounds. In the recording studio, producers’ desire for control led to greater use of ‘click tracks’ that musicians would be expected to follow, which could then be deleted. Session drummers were celebrated for their ability to play on top of or behind the click, giving the music a human feel while still being metronomic. They were prized for their consistency in the studio. Multitrack studio recording meant that band members no longer had to play at the same time as each other, and even that the constituent parts of the drum kit could each be recorded separately.

This musical Fordism ­– a regimented, synchronised division of labour – elicited a backlash from those who supported a lo-fi alternative. Bands like Nirvana made much of their rejection of all things automated. Death Cab for Cutie started a campaign against Auto-Tune to ‘bring back the blue note’. Dave Grohl, accepting a Grammy in 2012, gave a shout-out to team human: ‘It’s not about being perfect, it’s not about sounding absolutely correct, it’s not about what goes on in a computer.’ One man, calling his campaign the ‘Society for the Rehumanization of American Music’ began selling ‘Drum machines have no soul’ bumper stickers. But, as Sousa and countless others discovered, authenticity is slippery. Arguments about what constitutes real music normally hide other concerns about who has power, who defines cultural taste, and who is able to access music-making. Those who learnt with their friends on cheap guitars may worry that kids with drum machines will have different, more solitary musical apprenticeships. The technologies of concern change, but the debates remain the same.

Just as the pianola had done a century earlier, drum machines freed producers from the limits of human bodies and premodern percussion. Bass drums could be tuned to produce bass lines, tempos could increase, rhythms could become more complex. Once new sounds such as the 808 bass and skittering ‘trap hi-hats’ became ubiquitous in 21st-century music, some drummers saw a new challenge in replicating these software creations. Live humans, augmented by electronics, opened up new styles.

Drummers have sought to emulate Dilla’s off-grid, machined imperfections

According to the drummer, producer and filmmaker Ahmir ‘Questlove’ Thompson, the world’s greatest drummer at the start of the 21st century was not a drummer at all, but a drum-machine beatmaker. J Dilla is the subject of a new book by Dan Charnas that credits him with the invention of a whole new feel – Dilla time. Dilla’s instrument was the Akai MPC, a sampler designed by Linn that allowed not just for the programming of beats but also for the scavenging and reuse of any available sound. Linn wrote in the manual for the MPC 3000 that samplers had ‘evolved to a point at which they are now a legitimate instrument on their own … the piano or violin of our time … [A]s an MPC3000ist … your MPC virtuosity may be identified by your particular swing settings, your creative use of Note Repeat’, and so on. Dilla used his MPC to experiment with error. He slowed down samples from his record collection so that their rhythmic imperfections would stand out; he used the MPC’s ability to swing some parts of beats while leaving others straight; he rejected the machine’s ‘timing correct’ function, which had been imported from the earlier Linn LM-1. He moved backbeats so that, rather than being ‘in the pocket’, they were jarringly early.

Questlove recounts how, the first time he heard a Dilla production, ‘It was the most life-changing moment I ever had … It sounded like the kick drum was played by a drunk 3-year-old.’ Charnas argues that Dilla’s rulebreaking caused a generational shift in rhythmic sensibility, a ‘new, pleasurable, disorienting rhythmic friction’. His imprint has transcended hip-hop. Pop records now feature his distinctive feel (try listening to the songs ‘Stay Next to Me’ by Quinn XCII and Chelsea Cutler or ‘Sincerity Is Scary’ by The 1975). Where once we would nod our heads in recognition of a well-located quaver, with Dilla time we now enjoy the effort of having to search for the beat.

As well as changing how producers use drum machines, Dilla time has also changed how drummers play. Questlove told Charnas how, in his quest for live hip-hop, he had wanted to sound more and more like a machine. Following Dilla, he began a process of microrhythmic exploration. Dilla time has led to a new wave of live innovation. Drummers have sought to emulate Dilla’s off-grid, machined imperfections. In doing so, they have had to unlearn their instinctive beat placements so that they could search out new ones. If Charnas is right, the use of a drum machine in expert hands has not just introduced a new musical aesthetic; it has changed our collective sense of rhythm. So can we use the history of drum machines to anticipate what the next wave of computer innovation might mean for music?

Much of the hardware of mechanical music has now, as with so many other 21st-century innovations, become software. GarageBand and other digital audio workstations have handed the means of record production to anyone with a laptop. These programs don’t just package all of the accumulated sounds and features of earlier drum machines into their software. They now offer something more – an AI drummer that can listen to your tracks and compose an appropriate part to accompany them. The GarageBand user guide says: ‘You can choose drummers from different genres. Each drummer comes with its own drum kit and a distinct playing style.’

A drummer might view this as yet another attack on their expertise. Drummers pride themselves on not just being able to play, but also on knowing what to play. Stubblefield was not just the player of the ‘Funky Drummer’ break; he also composed it, drawing on his years of accumulated knowledge. Drum machines have shown us that drumming is not just hitting things. It is a form of embodied, tacit, creative skill that is impossible to completely transcribe or put into code. It would be understandable for us drummers to become defensive. We could assert that an AI drummer would fail a Turing test if it were asked to improvise in a jazz quartet, listening, anticipating and responding in real time to its bandmates. But that discussion would be a sideshow, like a Muppet Show drum battle between Animal and Buddy Rich.

All innovation in art has generated new answers to the question of what counts as art

Kate Crawford argues that we should think about AI not in terms of a battle between human and computer intelligence, but as a ‘gold rush … enclosing different fields of human knowing, feeling, and action – every type of available data – all caught in an expansionist logic of never-ending collection … a pillaging of public space’. When technologies encroach on new aspects of our lives, we are forced to reflect on what we value. Over the past year, artists might well have felt some anxiety as generative AI systems such as OpenAI’s DALL-E enabled the instant, free creation of new images. Musicians may have been more intrigued than threatened by Jukebox, an AI from the same company, and Google’s MusicLM. The point is not that, as has happened with chess, these computers will surpass the best composers or players, but that a modern version of what used to be called ‘muzak’ can now be generated without the involvement of a producer or the payment of any royalties. With the encroachment of AI into music, there are important questions about what happens to creativity, imperfection and the livelihoods of music producers, but the bigger questions are about who gets the rewards and who gets to define culture.

Musicians are all too familiar with the gig economy, and the exploitation that comes with it. Despite James Brown’s exhortation to ‘give the drummer some’, Clyde Stubblefield was paid a session fee but saw hardly any royalties from his wildly successful creation. Drum machines took creative power from drummers and handed it to producers. Generative music seems to displace not just one or two musicians, but all of them, and their producers. Companies are already autogenerating ‘functional music’, made to provide aural wallpaper rather than artistic novelty. Their machines feed on historic music data, but unlike with Stubblefield’s sample, we won’t know where they get their inspiration from. There need be no credit, just profit. Musicians do not just fear their own redundancy. They also fear the devaluing of their art. All innovation in art has generated new answers to the question of what counts as art. Drummers may be the canaries in the coal mine. We have seen with the development of the drum machine that there is an opportunity to use AI to release a new wave of musical innovation. The risk is that the technology will concentrate power in ways that devalue and homogenise music.