Skip to main content

We are open 10am-5pm Tuesday-Saturday and 12-5pm on Sundays. We are not open on Mondays.

Authorship in the Age of Artificial Intelligence: An Art of Disappearance


Platform Graduate Award artists Bevan Hill & Will Griffiths create artworks generated entirely by AI technology. Their exhibition Real Art for Real People brings together painting, sculpture and written work, conceptualised and designed by computers. By removing themselves from the ‘creative process’ they explore the importance of individual expression and agency in an increasingly automated world.

This essay is drawn from their book. Posing as fake art critics, Hill and Griffiths’ written work gives more insight into their artistic intentions and the issues their work addresses.


Authorship in the Age of Artificial Intelligence: An Art of Disappearance

Saul Grimberg

A cursory glance at history shows us that technological advances have routinely revolutionised prevailing means of contemporary perception. No less affected by the radical changes of the last century is our mental impression of the artist and the location of their authorship. We saw the advent of photography liberate the hand of the most important artistic functions, and the flatbeds of Pop Art shift the analogue of visual experience away from the insulated formalism of Abstract Expressionism. Opening the door to the rationale of consumer capitalism, the processes of mechanical reproduction and the division of labour have led us to the current situation, in which an artist’s role is redefined as a managerial one, and where efficiency overrides the merit of expression through direct engagement with materials, so that manufacture can justifiably be offloaded to the production line.

With Real Art for Real People, Hill and Griffiths launch a mocking critique of this current state of affairs by reversing its terms. With the aid of artificial intelligence, the duo lampoons the detachment of the artist-manager model by offloading the entire creative process, becoming the apathetic manufacturers of a potentially indefinite number of generated works under an automated, unfeeling superior. Interposing a generative conga of automatic procedures between themselves and the creative act, the absent collaborators paint us a menacing picture of a world where gallery objects are nothing more than an empty art of disappearance; a world where the extent of an artist’s creative input is reduced to the *click* of a button in a final act of vanishing. For all the massive leaps towards a technology of immense capacity and penetration, it seems like the relationship between perpetual innovation and human expression is as brittle as ever. Through my contribution to this short catalogue, we will attempt to establish a theoretical framework to understand the piece in relation to the idea of authorship or lack thereof.

“A world where the extent of an artist’s creative input is reduced to the *click* of a button…”

These ideas – the artist vanishing, the prospect of disintegration and the space therein – cause to recall the very literal representation of the disappearing artist, pictured cogently by Keith Arnatt in his series of nine photographs titled Self-Burial. As a fitting starting point to contemplate the piece in question, this useful representation will help us to map the trajectory of the disappearing author in recent art history. Central to the work is the disappearance of the art object. Transient and withdrawn from the permanence of a gallery setting, the images of Arnatt’s disappearance into the ground were initially reproduced on German television screens over a period of eight nights, entering each viewer’s home. What this demonstrated so vividly was Berger’s important summation that, because of the camera and the reproduction of images on television screens, the art object now travels to the viewer as opposed to the viewer to it, and “in its travels, its meaning is diversified” (Berger, 1972, pp. 19). Crucially here, there is no unique art object – it is a work of art designed to be reproduced. The photographs we look at retroactively serve only as documentation; the very brief ‘location’ of the work was in the millions of homes of German families who happened to be watching TV at 8:15pm on a cold October night in 1969. Additionally, there is no ‘authentic’ print of Self-Burial, because any number of prints could be drawn out of the original photographic negatives. A work of art designed for reproducibility, unfixed in its location and entirely ephemeral – it typified the ideological severance from the unique and permanent art object that was so revered in the earlier half of the twentieth century.

If we agree with Walter Benjamin’s assertion that “the presence of the original is the prerequisite to authenticity,” the widespread broadcast medium of Self-Burial is in itself a symbolic burial of the age-old dogma that art should be inextricable from a permanent, authentic form (Benjamin, 1935, pp. 3). It epitomised the zeitgeist of the sixties, an era which saw mechanical reproduction emancipate the work of art from the teleological doctrines of l’art pour l’art (art for art’s sake) and its “parasitical dependence on ritual,” which had, up to that point, defined preceding art movements and their rigid centering on authenticity and introspective genius (Benjamin, 1936, pp. 6). As Arnatt later wrote about the premise of Self-Burial, “The continual reference to the disappearance of the art object suggested to me the eventual disappearance of the artist himself” (Arnatt, 2015). If Arnatt’s vision was a metaphorical premonition of the disappearing artist in response to the goings on of late sixties postmodernism, Hill and Griffiths’ Real Art for Real People is evidence of the artist disappeared.

Real Art for Real People is evidence of the artist disappeared.”

Save for thematic similarity, and the trivial fact that both Hill and Griffiths live in Oxford like Arnatt before them, there are some obvious differences here inasmuch as the duo take this subjugation of authenticity and push it to an unprecedented extreme. As much as Self-Burial visualises the artist’s gradual submersion to a point of vanishment, the author is still abundantly present, for it was he who devised the concept and staged the images. However transitory, Arnett’s allegory still reflects a visuality arranged by the artist himself. It is still, then, an overt expression of his own philosophical ideas about art. As Brennavan Sritharan puts it, “by denying his presence he paradoxically announces himself, heightening his role” (Sritharan, 2015). We can compare this to Real Art for Real People, where – in the offload of all creative involvement – any creditable intent from the duo is completely impenetrable and absent from the art objects themselves. The artists have completely screened themselves off from the ‘work’ by placing AI processes between themselves and the end products.

This is made even more evident when we realise that the only active role the young rascals play is that of ‘assistants’ to their invisible AI superior, faithfully reproducing machinemade designs so as for them to exist in a physical form. For Hill and Griffiths, the process of bringing these objects into existence is reduced to nothing more than a process of manufacture – a necessary hurdle between the digital and the physical. When we look at all four of the paintings, we can see the duo’s lack of intent; they are restricted to the same square forms of the JPEGs from which they derive, a picture plane neither portrait nor landscape and identical in dimension, negating any kind of perceived world space we might be familiar with. Crucially, though, we must also remember that the body of work we see are the products of a chain of processes that further removes the collaborators from view. In a way that digitally simulates the early steps an artist would take in the creative process – drawing upon their prerequisite knowledge to inform their own work – they use an AI to displace the cognitive origin as well. All of the objects we see derive from an algorithmic text-generation process that sucks up descriptions of works from the canon of Western art and spits out new ones. Owing to these combinatory processes of automation, surely we can no longer call these objects the brainchildren of Hill and Griffiths? Submitting all intent for the efficacy of technology, they successfully deliver the final blow to the idea of the auteur as a creative genius expressing an inner vision. In the space created by their absence, the only place left for the objects to be made sense of or ascribed meaning is in the mind of the viewer. Here we can return to Roland Barthes’ conception of The Death of the Author: “We know that to restore writing its future, we must reverse its myth: the birth of the reader must be ransomed by the death of the author” (Barthes, 1987, p. 146). By refusing the ‘author,’ the duo refuses to assign any ultimate meaning to the work and thus refuse to fix its meaning. Unlike Arnatt, whose allegorical disappearance served only as a prognosis of the author’s death, Hill and Griffiths withdraw without a trace, in an ironic, suicidal embrace of artificial intelligence.

“By refusing the ‘author,’ the duo refuses to assign any ultimate meaning to the work and thus refuse to fix its meaning.”

One way to assess Real Art for Real People is to chart the implications of automating the creative process to a point where all that is left is a machine way of seeing. As we touched on earlier, the advent of mechanical reproduction techniques saw the artist and their work pried from a well-established ideological shell. No longer intrinsic to the value of art was a militant individualism or autotelic notion that it be dislocated from any didactic, social or political function. This seismic shift was demonstrated perhaps most lucidly by Warhol, whose assembly lines and out-sourcing of [re]production set the precedent for the artist post-pop: the executive business-artist archetype that we see so frequently today with the likes of Jeff Koons, Anish Kapoor and many more. We can trace the executive mentality that permeates today right back to his ‘Factory’ studio, where his employed division of labour downplayed the importance of creativity to the point of his literal absence: “The whole time I was in the hospital, the [Factory] ‘staff’ kept on doing things, so I realised I really did have a kinetic business, because it was going on without me. I liked realising that, because I had by that time decided that ‘business’ was the best art” (Warhol in Jones, 1996, pp. 203). The Pop artists of the sixties successfully shifted the analogue of visual experience away from the subject of a perceived world, bringing the processes of such reproduction and their diffusion of information to the fore. As Leo Steinberg vividly laid out, the changes brought about by mechanical reproduction and their legitimation as a valid artistic process can also be understood symbolically as the “tilt in the picture plane;” the ninety-degree turn from the established vertical model of picture-as-window to a horizontal model of picture-as-data, illustrative of the radical shift from the ‘natural’ to the ‘cultural’ in image-making (Steinberg, 1972, pp. 82-92). Liberated from the confines of sanctity and ritual that were previously necessitated, art’s relation with mechanical processes and automation reconnected it to a socio-political function.

With this in mind, part of the legacy of Pop Art is a mitigation of any requirement for the artist to work with materials; a partial offload of the creative process. However, it is necessary to stress that the work created still hinges on the selection process of the artist. The subjects of Warhol’s prints, or Koons’ choice of what trinket to inflate next, for example, are still contingent on the artists’ inclinations. Radical as it may seem when compared to a Renaissance painter for example – whose work was defined by a solitary struggle with the canvas – up to now, technology and the division of labour only serve to ease the load on the conceptual artist. Mechanical reproduction processes in the previous century were limited by their mimetic capacity and had to be manned by assistants, whether it was to push and pull a squeegee or click a shutter release of an analog camera. The designated artisanal, telematic and corporate production processes available to the contemporary artist are also mediated by the same requirement of human involvement, thus still distinguishable as ‘human’ ways of seeing. However, when we consider the recent innovations of artificial intelligence and their application in Real Art for People this is no longer the case. As the authors of the text Ways of Machine Seeing recently noted, “new ways of algorithmic seeing allow the constant production of new visuals while simultaneously increasing the gap between seeing and knowing” (Azar, Cox & Impett, 2021). Hill and Griffiths’ latest work exemplifies this sentiment; the objects we see are the manifestations of machine learning processes, which are outside of the human scope of perception and reason. When we look at each of the hanged paintings or the sculptures fixed atop their plinths, our mind tries to perceive a subject but we are denied entry – we become alienated. This troubling new visuality emphasises the point of Azar, Cox and Impett, that “the current distribution of agency across complex networks of non-human agents allows simultaneously more visibility […] and less knowledge about the very processes behind the way in which these new visualities are rendered visible” (2021). In other words, the trouble we have in deciphering this mode of seeing is that the processes that make it visible to us are beyond the confines of our own faculties and frameworks of knowledge. It offers us a vision previously unattainable, whilst simultaneously – to borrow Bernard Stiegler’s words – “short-circuiting the deliberative functions of the mind” (2019, pp. 26). Perhaps even more disconcerting than our fractured encounter with Real Art for Real People is that machine learning can equally, almost seamlessly, pretend to be factual. Recently, for example, we saw with the new ING project The Next Rembrandt, that an AI has been used to convincingly simulate the painting style of Rembrandt using an image dataset of his original artworks.

“The objects we see are the manifestations of machine learning processes, which are outside of the human scope of perception and reason.”

It is important to note that, despite taking credit for the work, the processes of automated generation are just as foreign and indecipherable to Hill and Griffiths as they are to us. If we find ourselves in a technological zero-sum game where no artist can stand behind the work with any semblance of authority, what happens to the idea of authorship? As ambivalent as they seem on the surface, it is my inference that this is a question they are posing to us. Even as the question of authorship has been perpetually pushed to the periphery of artistic consciousness, here it is again – asked more vociferously than ever. It seems to me that Hill and Griffiths provide us with an eerie reminder of the hypothesis Baudrillard so eloquently outlined in The Perfect Crime: Out of compulsive longing, we are building ourselves a virtual world so as to be able to opt out of the real one without a trace. Through technology, we try to regain liberty by relieving ourselves of the burdens of our own will and imperfections. In Baudrillard’s words, “The artist too, is always close to committing the perfect crime: saying nothing. But he turns away from it, and his work is the trace of that criminal imperfection. The artist is the one who, with all his might, resists the fundamental drive not to leave traces (Baudrillard, 1996, pp. 1). If we are to qualify algorithmically generated art as a legitimate practice in itself and accept the inherent fracture between what we see and what we know, we draw ever closer to Baudrillard’s summation.

“We are building ourselves a virtual world so as to be able to opt out of the real one without a trace.”

We have seen how the work can be read as a warning against the implications of an automated creative process. However, what has been deliberately left unmentioned thus far in this particular survey – making clear the various angles of the piece – is another dimension to Real Art for Real People. This dimension, as expanded on further by Wayne Gadbrooks in relation to the commodity status of the work, hinges on the duo being the ones to physically materialise the algorithmically generated images as one-off oil on canvas paintings. Antithetical to Koons, Murokami and Kapoor to name a few, whose outsourced labour implies that the concept of a work is where authorship lies – with material realisation a negligible obstacle – Hill and Griffiths’ involvement lies solely in material reproduction. Reading the piece this way, we can see their offload of any cognitive and visual input to the indifference of AI as a thumb in the eye of the art world: an exaggerated reversal of the artist-manager set up which calls for a reevaluation of engagement with materials and the physical processes required to realise an artwork. It is crucial that we acknowledge this interrogative angle because it redefines the position of Hill and Griffiths’ authorship. Rather than ambivalent, disappeared artists merely transforming the gallery space into a showroom for the work of a mechanised other, they become subversive tricksters intent on disrupting the status quo of the art world.

Although the duo embraces neither prospect wholeheartedly, their role in the process of artisanal reproduction is certainly suggestive. Unlike the artist-manager set up discussed, where outsourcing implies a disregard for hands-on material realisation, the sheer amount of labour Hill and Griffiths burden themselves with is in itself indicative of its value as part of their practice. This seemingly futile drudgery of tirelessly reproducing JPEG images with oil on canvas can be seen as devotion to craft. It shows an ideological conviction that even an algorithmically generated design can yield something in the way of expression when reproduced by a human hand. Perhaps the unrectified discrepancies between the generative results and the actual results act as a residual homage to the importance of material exploration. Whichever interpretation we as viewers prefer to afford Real Art for Real People – central to the work is the idea of unstable authorship, and how fragile it becomes when elements of the creative process are offloaded elsewhere.


Arnatt, K. (2015). In K Arnatt: the conceptual photographer who influenced a generation [online] Available at: who-influenced-a-generation/ [Accessed 03.04.2021].

Azar, M., Cox, G. & Impett, L. (2021). Introduction: ways of machine seeing. AI & Soc: Knowledge, Culture and Communication [online], pp.1-12. Available at: [Accessed 10.04.2021].

Barthes, R. (1987). Image, Music, Text. London: Fontana Press, pp.142-149.

Baudrillard, J. (1996). The Perfect Crime. Translated by C. Turner. London: Verso pp. 1-15.

Benjamin, W. (1969). The Work of Art in the Age of Mechanical Reproduction. In B. Walter, H.

Arendt (ed). & H. Zohn (trans)., Illuminations. [PDF] New York: Schocken Books, pp. 1-26. Available at: [Accessed 07.04.2021].

Berger, J. (2008). Ways of Seeing. St Ives: British Broadcasting Corporation and Penguin Books, pp. 1-64.

Jones, C. (1996). Machine in the Studio: Constructing the Postwar American Artist. Chicago, London: The University of Chicago Press, pp. 189-267

Sritharan, B. (2015). Keith Arnatt: the conceptual photographer who influenced a generation. [online] 1854 Photography. Available at: conceptual-photographer-who-influenced-a-generation [Accessed 03.04. 2021].

Steinberg, L. (1972). Other Criteria: Confrontations with Twentieth-Century Art. New York: Oxford University Press, pp.82-92.

Stiegler, B. (2019). For a Neganthropology of Automatic Society. In: T. Pringle., G. Koch and B. Stiegler. (eds) Machine. Lüneburg: Meson Press, pp. 26.

Visit Real Art for Real People until 17 October. Click here to find out more and plan your visit.