Post Truth Machine Intelligence

The world is hurtling towards a Copernican crisis,¹ in which human intelligence is being rapidly de-centred. Regardless of our view of non-human intelligences, algorithmic entities are having incredible impact on society. Machine intelligences have passed² the Turing test³, prompting goalpost shifting responses.⁴ In these tests, GPT4.5 tested as human significantly more frequently than actual humans.⁵ In this volatile and uncertain environment, artists’ reactions range from eager adoption of powerful tools as part of an expanded practice, through to anger, fear and widespread claims AI art is not “real art”⁶. As we approach this crisis, we face unavoidable dual questions: are machines capable of conscious behaviour, and conversely, are humans biological machines with emergent behaviour, long accepted as “conscious”? In the context of visual arts we can ask “is consciousness a prerequisite for art-making?”

Simultaneously in modern society there is a problem of discerning truth.⁷ Vested political and commercial interests use surveillance, big data and vast computing resources to manipulate attention and perception in order to establish their preferred truth.

This is the intellectual and generative context to which I have responded, manufacturing a series of oil paintings on canvas. Repetitive artworks utilising optical illusions emphasise the relative and unstable nature (or non-existence) of truth, the unreliability of perception, semantic degradation and re-emergence. These illusions quantitatively reveal a mechanistic basis for human visual perception, explained by modern colour science and physiological understanding of how our eyes and visual processing circuits work. Hacking the viewer’s brain, considering optical illusions as a system vulnerability, I confront the viewer with their own machine-like nature.

Perception is dependent on our cultural and social background, like the colour shift phenomenon. Redacted and obfuscated obscenity is deployed as a stand-in for control of speech, where each side of the political spectrum demands free speech while simultaneously denying the same to the other side.

The remainder of this paper discusses and contextualises my paintings through three lenses: formal qualities (how physical objecthood and optical illusions frame the viewer-as-machine), process (how rules and optimisations frame the artist-as-machine) and explicit content (how redaction, obfuscation and repetition of obscenity model post-truth). Representative examples of the paintings are provided in Figures 1–4. Captions are descriptive and non-titular.

Formal Qualities

We begin with the formal qualities of my paintings, showing how physical objecthood and optical illusions configure the viewer-as-machine.⁸

Zylinska considers how generative art shifts emphasis from material presence to algorithmic process,⁹ while Steyerl extends objecthood to the digital domain.¹⁰ Like Fontana, my works are necessarily physical,¹¹ while, as argued by Merleau-Ponty, the physicality of the painting cannot be separated from the viewer’s embodied perception (I will expand on this when considering process).¹² Extension of colour fields around the sides of the canvas confirms the spatial extent of the canvas as object.

These paintings are hard-edged abstractions which, according to Greenberg, exemplifies self-critical modernism, denying representation and focusing on material concerns such as flatness, shape and the properties of paint.¹³ Judd goes further, collapsing the distinction between sculpture and painting object¹⁴, while Alloway, introducing Systematic Painting, focuses on repetition, unity, and clarity, where “the recurrent image is subject to continuous transformation, destruction and reconstruction”.¹⁵ In contrast, Fried criticised such theatricality, demanding work to be more instantaneous.¹⁶ Relevant to my work are artists such as Ellsworth Kelly, Ad Reinhardt, Frank Stella and Victor Vasarely, or more contemporaneously, Odili Donald Odita, Sarah Morris, Felipe Pantone, and Gabriele Evertz. Evertz carries forward Op Art practices¹⁷ established by artists such as Bridget Riley.¹⁸ In my work, this lineage is followed not so much as stylistic repetition, but as functional tools to establish the viewer-as-machine and artist-as-machine.

Focusing on relative and mechanistic aspects of perception, my paintings rely on the colour shift phenomenon¹⁹ pioneered by Josef Albers²⁰ and are colour field paintings in the spirit of his Homage to the Square series. I predominately use secondary colours to activate multiple receptors in the viewer’s eye.²¹

Executed in 2025, my paintings exist within these art-historical contexts; however, hard edges (emphasising artist-as-machine), optical effects and perceptual instability are adopted, not for their own sake²², but in dialogue with AI-generated art and the unreliability of modern public discourse where everything is fake.²³ My use of optical effects relies on an understanding of the human visual processing system, confronting the viewer with their own machine-like qualities (viewer-as-machine).

Gombrich describes the iterative artistic process, with artists extending stereotyped modes of representation developed by their predecessors.²⁴ Similarly, Popper argues that perception itself is an active process of interpretation, involving trial-and-error and learned conventions, where art doesn’t “make sense” until a viewer has been trained into the relevant conceptual paradigm.²⁵

Echoing Merleau-Ponty²⁶ and Popper,²⁷ Gabriele Evertz expresses that “... without the viewer the painting doesn’t exist. The viewer brings the painting to life.”²⁸ While Barthes would indicate that this is always true,²⁹ it especially holds when exploiting optical illusions. Dynamic and unstable visual effects produce an opportunity for Lee Ufan-style phenomenological “encounters”³⁰ produced in the machinery of the viewer’s mind.

According to Kuhn, perceptual instability signals shifting or contested frameworks.³¹ Keyes³² and McIntyre³³ describe our era as post-truth, in which “facts are subordinate to our political point of view”³⁴ Manipulation of public perception occurs through commodified surveillance capitalism, as described by Zuboff.³⁵ According to Steyerl,³⁶ attention is a battlefield in which weaponised data analytics are used by vested media interests. As elaborated in the next section, my paintings exist in a similar contested space, being carefully engineered attacks on the human visual operating system.

Benjamin argued that repetition provides cultural legitimacy.³⁷ Arendt argues that similar mechanisms underpin the power of repeated lies in politics.³⁸ This psychological illusory truth effect has been studied by Hasher et al.³⁹ and is amplified in the post-truth⁴⁰ circulation of digital media.⁴¹ Foster⁴² discusses how avant-garde practices became legible through repetition by their successors, conferring legitimacy after the fact. Neely discusses how repetition legitimises – once an idea is repeated, assumed intentionality erases perception of error.⁴³ I use repetition of simple ideas over successive canvases as a tool to legitimise an otherwise arbitrary set of artistic choices.

Iteration, repetition, rotation, and fragmentation are formal devices employed in my compositions. These transformations are computational elements. Wolfram proposed that the laws of physics are actually those of computation, and that mathematical physics consist of numerical shortcuts for these computations.⁴⁴ Complex physical phenomena emerge from iteration of simple rules (one-dimensional cellular automata).⁴⁵ Walker builds on these ideas to propose Assembly Theory, which defines life as the propagation of information through complex, evolving structures.⁴⁶ My adoption of mechanistic “computational elements” reflects this idea of emergent intelligence.

How I made the Paintings

Turning now to the process, I show how optimisation and engineering configure the artist-as-machine, while establishing human provenance.

I adopted an engineering approach to the design of these paintings, by which I mean a process of iterated experiments and computational optimisations.

Previously working in monochrome, I sought a way to introduce colour that was non-descriptive and non-aesthetic (or even anti-aesthetic). This led me to the idea of functional use of colour, with a view to deploying Albers’ colour shift phenomenon as an engineered attack on the human operating system.

Human perception of colour differences⁴⁷ is not linear⁴⁸ in colour spaces such as RGB, CMYK or HSL.⁴⁹ A perceptually uniform space, pioneered by Munsell⁵⁰ and adopted by the Commission Internationale de l’Éclairage (CIE),⁵¹ is one in which geometric distance between two colour vectors approximates perceived difference, allowing Euclidean distance to quantify perceptual shifts.

The CIECAM02 colour appearance model⁵² accounts for contextual influences such as luminance level, surround adaptation, and background. Colour is represented in terms of perceptual attributes: lightness \(J\) (perceived brightness relative to a reference white), colourfulness \(M\) (perceived intensity or saturation compared to a reference grey), and hue angle \(h\) (describing the type of colour sensation e.g., red, blue, green).

I developed a brute-force computational method for identifying surround colours that cause the greatest perceptual shift of a stimulus colour. We transform CIECAM02 into CAM02-UCS (Uniform Colour Space) and quantify differences in perceptual appearance via the \(\Delta E_{CAM02-UCS}\) metric. Exhaustively comparing perceived colour of the fixed stimulus across candidate surrounds, we identify surrounds that maximise perceptual shift. This offers an automated, quantitative analogue to the manual explorations of colour interaction pioneered by Albers.⁵³ The resulting Python computer code has been made freely available.⁵⁴ Remarkably, some of the colour triads found by Albers using trial-and-error are quite close to optimum.

Using this tool, I digitally explored colour space, quickly iterating candidate designs. From these candidates I selected outcomes for development in oil.

Ad Reinhardt developed self-imposed rules for the production of art.⁵⁵ His (rather Greenbergian) rules: no texture, no brushwork or calligraphy, no colours, no object, no subject, no matter, are evident in his “black” paintings.

Similarly, I adopted my own self-imposed rules for the production of these paintings: colour choice must be engineered for functional rather than aesthetic effect, same colours may not touch, letter fragments should be grouped by colour in opposition to their semantics, areas should be flat, edges should be hard, paintings should be executed in a single session.

Reinhardt worked hard to eliminate gesture. He is known to have methodically brushed out all brush marks, using only a one-inch brush.⁵⁶ However, I regard this as laboriously imposing will over the natural tendency of the medium, resulting in a painting that speaks very loudly to a human gesture, denying the will of the paint itself.

In my work, the obvious way to avoid brushstrokes and to have clean, straight edges is via digital print. However, this erases human presence and would in my view be uninteresting. Unlike Reinhardt, I am interested in the quiet human gesture detectable in the attempt, but unavoidable failure to achieve a straight line or a flat area of colour. It is precisely these “imperfections” that convey human origin and provide physical evidence of human labour in manufacture.⁵⁷ Risk is central to contemporary art practice. Rosenberg describes the canvas as “an arena in which to act,”⁵⁸ admitting the possibility of contingency and failure, while Moholy-Nagy advocates adventurism and experimentation.⁵⁹ More recently, Bois frames painting as a series of wagers.⁶⁰ Embracing risk, I attempted (but sometimes failed) to execute each painting alla prima.

When emphasising human provenance, one might question the use of digital tools and computational optimisation. However, all art production involves arbitrary choice of allowable technologies. Huizinga frames art as a game played according to rules,⁶¹ while Kubler describes artists arbitrarily deciding what frameworks to continue or abandon.⁶² Flusser emphasises the role of artistic tools in determining the outcome.⁶³ Arbitrary restrictions in literature are ubiquitously exemplified in prosody, and famously by Perec’s avoidance of the letter ‘e’.⁶⁴ In digital works, Manovich argues that apparent freedom is actually constrained by the underlying data structures,⁶⁵ while Steyerl shows how technological constraints shape the image.⁶⁶

When approaching without prejudice the mechanistic basis for machine and human intelligence, it was natural for me to adopt a hybrid process. Canvas, brushes, staples, stretcher bars, oil, pigment are all technologies. To this list I add digital computation.

Humans value embodied human labour, whether as Adorno’s “sedimentations or imprintings”⁶⁷ or in Sennett’s account of craft representing “…the special human condition of being engaged.”⁶⁸ Conversely, Singerman notes recent preference for theory over craft in art higher education.⁶⁹

According to Benjamin, mechanical reproduction transforms our perception of authenticity.⁷⁰ In an age of mechanical generation of art, perceived provenance is significant. Horton et al. conducted experiments in which AI- and human- generated art were presented to study participants, randomly labelled as AI or human.⁷¹ In their study, participants were asked to score the works by creativity, value, worth and skill. Their results show that the labels mattered more than the image, with human-attributed images scoring higher (independent of actual source). Humans care about art made by humans, not because of what it looks like, but simply the connection formed by the fact it was made by a human. This further motivates physicality and human trace in my paintings. Despite these biases, Demmer et al. show that machine- generated art can produce emotional responses in humans.⁷²

With colour selection and design performed digitally⁷³ within a narrow range of self-imposed rules, execution of the paintings becomes a LeWitt-style⁷⁴ mechanistic manufacturing⁷⁵ process. In contrast to the dematerialisation of Lippard and Chandler,⁷⁶ I take the position of Jones,⁷⁷ where the paintings are a witness to and product of a (private) performative process. In this anti-anti-form⁷⁸ stance, process and form are equally valid.

Presenting the first instances of a potentially infinite sequence, my paintings imply a space available for anyone to explore. Additional paintings could be executed by anyone, similar to Sol LeWitt Wall Drawings. In this respect I adopt Reinhardt’s position that “this painting is my painting if I paint it... this painting is your painting if you paint it.”⁷⁹ Tehching Hsieh, Time Clock Piece, On Kawara, Date Paintings and Roman Opalka, One to Infinity are also relevant.

By working in a mechanistic way, producing paintings that highlight the machine-like qualities of the viewer, I attempt to deal with machine and human intelligence without prejudice – pushing back against an unspoken “othering” of AI⁸⁰. Human society needs to prepare for the moral and ethical questions⁸¹ raised by the imminent possibility of machine consciousness. As argued by Shevlin, this can probably only be resolved by shifts in public attitudes and close relationships between humans and AI.⁸²

Explicit Content

Finally, I consider how redaction, obfuscation and repetition of obscenity model perceptual instability, extending our mechanistic framework to language. As discussed earlier, repetition also operates here as a legitimisation tool for testing cultural thresholds of legibility and offence.

I deploy the word fuck⁸³ as an index for censorship and control of speech. This is not aligned with any ideology, rather an observation that all sides of politics simultaneously demand “free speech” while aggressively denying it to their adversaries. Lukianoff and Schlott⁸⁴ describe the differing frameworks used by the left⁸⁵ and right⁸⁶ in cancelling their opponents, while Atkins and Mintcheva’s edited volume documents the multiplicity of modes of censorship in modern culture.⁸⁷

Empirically optimised redaction⁸⁸ has been used to transform the word into fragments at the boundary of legibility. The word is simultaneously present and absent, simultaneously text and shapes. I iteratively progressed from redaction alone (Figure 1), to semantically orthogonal colour assignments to the fragments (Figure 2), and finally to semantically disruptive colour fields, similar to but distinct from the Stroop effect⁸⁹ (Figure 3). This use of colour can be regarded as camouflage, operating on similar principles as early work of camoufleurs such as Andre Maré. Perception-altering colour fields echo Kuhn, who showed that perception is affected by the paradigm within which one is trained.⁹⁰

Redaction has been used in text-based works such as Jenny Holzer, Redacted (Top Secret), while Ull Hohn, Sex Painting, 1987 and Jon Campbell, Fuck Yeah, 2016 used obfuscated obscenity as compositional elements.

McWhorter describes the historical shift of taboo words from religion, and disease to sex and slurs against identity groups.⁹¹ Douglas shows that these share an underlying logic of pollution and boundary violation.⁹² Pinker describes the cultural feedback loops that both reinforce and transform what counts as unsayable in society.⁹³

The long-assumed intellectual deficiency of swearing has been thoroughly debunked in experiments by Reiman and Earlywine⁹⁴ and by Jay and Jay.⁹⁵ Pinker argues that swearing is not inherently immoral; instead it is an evolved form of emotional language.⁹⁶ De Vries goes further, arguing that swearing is morally innocent,⁹⁷ while Jay argues that harm rather than immorality is the real issue, noting that offensive words can cause genuine damage, but not invariably so.⁹⁸

Repetition of obscenity, within a single canvas, and across the series, mirrors desensitisation and cultural evolution of taboo words. One could even consider my paintings as a perceptual experiment regarding cultural thresholds for legibility and offence. Through repetition I aim to achieve a semantic drain where the word loses all meaning. For Barthes, repetition is a mechanism whereby language is emptied and mythologised,⁹⁹ while for Derrida,¹⁰⁰ repetition destabilises meaning via an endless unresolved chain of différance.¹⁰¹

Going further, I discard the letter structures and use the fragments as autonomous shapes in works (e.g. Figure 4) that present the viewer with an emergent language where repetition has broken down existing meaning. This is a literal Derridean scattering or dissemination,¹⁰² where, following Deleuze,¹⁰³ difference becomes generative, extending to Baudrillard’s simulacra,¹⁰⁴ or Wolfram’s automata¹⁰⁵ where simple rules generate meaning without external reference.

According to Bergen, swearing reveals much about the mechanism of the brain, occurring both autonomously (viewer-as-machine) and consciously.¹⁰⁶ Through an extensive range of studies, Jay identifies neurological, physiological and cultural factors that modulate swearing prevalence.¹⁰⁷ Viewer-as-machine is further activated by exploiting text-completion mechanisms in the brain.¹⁰⁸ These perceptual mechanisms are culturally dependent,¹⁰⁹ affecting how we recognise and remember information. Similar processes underpin all of modern digital computing and communications.¹¹⁰

Redacted and obfuscated obscenity functions as an engineered system requiring completion in the mind of the viewer-as-machine, while simultaneously modelling the malleability of meaning in a post-truth era.

Conclusion

I have responded to the unsettling de-centring of human intelligence and attendant othering of machine intelligence through hard-edged oil paintings that emphasise the artist-as-machine (through arbitrary rules, optimisations, and mechanistic processes), and viewer-as-machine (via perceptual instabilities, optical illusions, and semantic dissipation that destabilise what the brain and eyes think they “know”).

Perceptual instabilities along with repeated, redacted and obfuscated obscenity further reflect on the challenges of human discourse in a post-truth society. My paintings participate in the broader cultural negotiation of what it means to be human. Truth and perception have always been unstable, and it is not art’s role to offer certainty. Rather, it reveals this fragility, and the possibilities that emerge when modes of knowing inevitably collapse.

The formal qualities, process, and explicit content of my paintings form a device that positions the artist-as-machine and viewer-as-machine, simultaneously modelling the perceptual instabilities of the post-truth era.

My paintings embody the AI Copernican crisis: denying the viewer their accustomed position of perceptual dominance and reminding us that we are all machines.