AESTHETIC NEGATIVITY VS. ECONOMIC NATURALISM: A CLOSE READING OF AN INDIAN AI-ART PRACTICE

 

Anoop Daniel Ponnachan ¹, Anupriya Rajawat ² , Jaynat Kumar Gupta ³, Dr. Megha Attray Purohit ⁵

 

¹ Ph.D. Researcher, Department of Visual Arts, Banasthali Vidyapith, Banasthali, Rajasthan 304022, India

² Ph.D. Researcher, Department of Visual Arts, Banasthali Vidyapith, Banasthali, Rajasthan 304022, India

³ Ph.D. Researcher, Department of Visual Arts, Banasthali Vidyapith, Banasthali, Rajasthan 304022, India

⁴ Assistant Professor, Department of Visual Arts, Banasthali Vidyapith, Banasthali, Rajasthan 304022, India

 

 

 

 

 

1. INTRODUCTION

AI-based art is frequently considered to possess inherent critical significance, either by virtue of the sophistication of its tools or the pressing relevance of the subjects and processes it engages. A concern that usually goes unaddressed is that, once a work is made to move through circuits that reward novelty and technological distinction, aesthetic critique does not simply remain intact within it. Critique itself can remain present at the level of concept while the visual encounter becomes entirely consumable, and in that sense fully compatible with the very logic it appears to resist Han (2017), Gawronski (2021). Harshit Agrawal’s work becomes useful to think with at precisely this point. The 2021 exhibition ‘EXO-Stential: AI Musings on the Posthuman’ entered the market through a language of institutional firsts. Even if this would have been partly a matter of gallery promotion, such framing risks shifting attention away from what is actually at stake in the work well before the work itself is encountered Emami Art (2021), Baas (2024). The problem first appears at the level of circulation, where platforms and speculative markets sort for novelty first and visibility becomes the condition through which the work is first received. The work is also entangled in infrastructures of training data whose politics and biases shape what the system finds legible in the first place Crawford and Paglen (2021). Alongside this sits the persistent difficulty that digital practices still face in contemporary art discourse, where technological mediation is routinely absorbed into production while remaining marginal at the level of sustained critical attention Bishop (2012).

Agrawal’s use of Indian visual archives, including Kathakali masks and Theyyam ritual forms, can be read as an attempt to intervene within this convergence Agrawal (2019). Whether it fully succeeds is harder to say since the assertion of locality does not stabilise the work so much as generate another friction, because the global platforms through which the work must circulate are not especially troubled either by critical intent or by cultural specificity.

Four theoretical positions shape the reading here: Byung-Chul Han’s critique of digital smoothness Han (2017), Adorno’s negative dialectics Adorno (1973), Alexander Gawronski’s account of economic naturalism Gawronski (2021), and Baudrillard’s understanding of the hyperreal Baudrillard (1994). Han names the smooth, affirmative surface logic through which digital works now circulate and become agreeable to consumption. Adorno matters because he keeps the question of critique tied to form than leaving it at the level of concept or declared intention. Gawronski helps make visible the structural problem by which even gestures of resistance can be absorbed back into circulation and value. Baudrillard sharpens a related issue, the condition of AI imagery itself, where the image no longer returns to a stable real so much as intensifies a hyperreal one.

One recurring difficulty in AI-based, and more generally digital-based, practice is that a work changes meaning once it passes from embodied encounter into documentation. In its physical experience, spatial negotiation can cultivate opacity and delay. Online, the same work is compressed into a shareable artifact where the caption often ends up carrying the argument while the image circulates as a pictorial placeholder. The intervention of AI upon an image or text, for that matter, operates on similar grounds. The discussion that follows keeps returning to two linked questions: what does the work demand in installation, and what does it become once flattened into platform media? Han (2017), Gawronski (2021). This, in extension, asks what happens when a dataset of images gets processed, categorised, distilled, and absorbed by an AI model.

 

2. Methodology

This study adopts a qualitative interpretive approach through close reading, and it treats Agrawal’s practice as a focused case through which a broader structural question can be tested. When artificial intelligence becomes both the means of image production and the theme that legitimises the work in public circulation, under what conditions does critique remain active? This paper addresses this question through the following three works, The Anatomy Lesson of Dr. Algorithm (2018), Masked Reality (2019), and Strange Genders (2020), because together they move across generative image making and interactive transformation.

It reads image behaviour, installation demand, interface design, and online circulation together to ask whether critique occurs in form or is carried mainly by caption, curatorial framing, or technological rhetoric. Earlier histories of computational art remain relevant here because they already staged questions of system and authorship that persist in contemporary machine learning practices Cohen (1995), Mohr (n.d.). The theoretical frame is used here to test where the work sustains friction, where it weakens, and where it folds back into the logic it appears to resist.

 

3. Findings and Discussion

3.1. Computational genealogy and critical frame

Computational art’s earlier history already brings into view problems that remain relevant for Agrawal’s practice, especially the shifting relation between artist and system, the institutional fiction of authorship, and the recurring suspicion that algorithmic images may appear formally resolved while remaining affectively distant Cohen (1995), Mohr (n.d.). That older problem does not disappear under machine learning. It returns in altered form. Agrawal’s work becomes publicly visible partly through the rhetoric of technological newness, yet the more consequential issue is whether critique survives once newness is already functioning as value in circulation Baas (2024), Emami Art (2021).

 

3.2. The Anatomy Lesson of Dr. Algorithm

The Anatomy Lesson of Dr. Algorithm (2018) is one of the points where Baudrillard enters most clearly. By invoking Rembrandt’s anatomy lesson and replacing the singular body with a synthetic reconstruction produced from medical datasets, Agrawal shifts the work away from empirical reference and toward a hyperreal register Agrawal (2018). The resulting images hover between figuration and abstraction, at times suggesting bodily elements and at others dissolving into painterly quality. That ambiguity is crucial because it can estrange the medical gaze by withholding anatomical clarity and returning the body to a state of unease. At the same time, the same ambiguity can become visual pleasure. What ought to disturb may instead begin to delight.

 Figure 1

Figure 1 Agrawal, Harshit; “The Anatomy Lesson of Dr. Algorithm” (2018); Installation View

Source: https://naturemorte.com/exhibitions/gradientdescent/selectedinstallationviews/5505/

Here an older problem of computational art returns in changed form: the image can remain visually lush while affectively distant, and that distance becomes most legible precisely when the aesthetic surface coheres too well Han (2017), Mohr (n.d.). The real issue is whether the formal behaviour of the image sustains contradiction or whether it lets that contradiction slip away into painterly attraction. In the physical experience, the work takes time. The viewer has to remain with it long enough for the image to hesitate between medical legibility and abstract seduction. Online, however, it circulates too easily as a series of seductive stills, and here the argument tends to migrate into explanatory text while the image itself settles into detached surficial acceptance Han (2017), Gawronski (2021).

 

3.3. Masked Reality

Masked Reality (2019) is more instructive because cultural archive here does more than ornamental work. The piece uses facial-recognition and transformation systems to overlay viewers’ faces with digital masks drawn from Indian theatrical and ritual traditions, including Kathakali and Theyyam, so that identity appears less as essence than as an assemblage of inherited forms Agrawal (2019). The immediate effect is uncanny but also playful. One sees oneself in live video and then, almost at once, as something composed through other histories and visual codes. In theory that mode of engagement gives the work a sharper edge than a merely decorative use of cultural reference, yet the risk is equally obvious. Interactive systems often generate instant delight, the quick pleasure of recognisable transformation, and that pleasure can cancel duration, which is precisely the condition under which negativity might begin to emerge Han (2017). If the viewer remains in front of the work long enough to feel discomfort at the instability of the self, the piece exceeds the filter logic. If not, it collapses back into novelty.

 

Figure 2

Figure 2 Agrawal, Harshit; “Masked Reality” (2019); Installation View

Source: https://harshitagrawal.com/masked-reality-interactive-work/

 

Its critical force lies in how long the viewer stays with that unstable mirroring. In the physical experience of the work, the shifting mask can produce discomfort because the viewer is held within a system that mirrors them back as archive. Online, the same operation is easily reduced to a short clip or a before-and-after transformation. The mask becomes a filter, the politics becomes caption, and the work is tempted to vanish into shareability Han (2017), Gawronski (2021).

 

3.4. Strange Genders

Strange Genders (2020), developed with 64/1, is probably where aesthetic negativity appears most sharply in Agrawal’s practice, because classificatory violence is made visible in the form of the work rather than left hidden inside its infrastructure NVIDIA (n.d.). A generative model trained on naïve drawings labelled “male” and “female” produces images that remain visually ambiguous, yet a classifier still assigns them probability scores. A form may refuse easy gender legibility while the system still insists on decision. The work does not resolve that conflict. It keeps the mismatch in view. That is where much of its force comes from. The issue is not gender as a theme in any broad or reassuring sense. It is classification as pressure.

That same concern shapes the spatial logic of the piece too. The drawings are arranged in a loosely concentric formation that reads less as neutral display than as a kind of diagram, a map of a latent space in which ambiguity is still being organised under a kind of taxonomy. The work stages contradiction rather than moving beyond it. It does not offer a consoling message about fluidity, nor does it celebrate indeterminacy as if it were some already secured freedom. The visual ambiguity of the images and the numerical insistence of the scores remain in conflict. Negativity emerges here not from what the work says but from what it refuses to let the viewer do, which is to reconcile that opposition into harmony Adorno (1973). Formally, the work is post-conceptual and procedural but it offers little by way of pictorial seduction. What works for it is that its cognitive demand is not smooth or reassuring. The awkwardness of percentages attached to bodies that do not fit their categories is where much of the work’s pressure begins.

 

 

3.5. Position within contemporary artificial-intelligence art

Within the broader field of artificial-intelligence-based art, Agrawal’s work does not primarily seek immersive spectacle. It turns instead toward archival identity, simulation, classification, and the pressure exerted by computational systems on perception and selfhood. Even so, that does not place it outside the same economy of visibility. In that sense, artificial-intelligence-based art is repeatedly packaged as a cutting-edge commodity, and visibility itself can become one of the conditions through which critique is neutralised Gawronski (2021).

The work turns toward how identities are recomposed through algorithmic systems, but just as much toward how classification begins to harden into force, and how simulation, once no longer only a theme, starts to act more like an environment Agrawal (2018), Agrawal (2019), Baudrillard (1994). Yet the attempt remains uneven. That unevenness is important. It is where the practice becomes most readable, not as a secure resolution but as a field of unresolved pressure.

 

4. Conclusion

These works do not settle the question, but at best they sharpen it. At certain points, the conditions of artificial-intelligence production become aesthetic pressure: distributed agency becomes visible, classification is felt as violence, and archival identity becomes unstable form rather than reassuring cultural content. But that pressure does not hold consistently. It weakens when conceptual framing, and curatorial rhetoric carry the work so completely that the encounter itself becomes smooth and quickly consumable Han (2017).

The central question is therefore structural rather than moral. Under what conditions can artificial-intelligence-based work remain incompatible with frictionless circulation?

Agrawal’s practice is better read not as a resolution of that question, but as a place where the contradiction becomes especially visible. It brings machine learning into contact with local visual archives and, in doing so, opens a space where global artificial-intelligence infrastructures are made to pass through culturally specific material Agrawal (2019). But the work enters circulation within the same atmosphere that rewards polished interfaces, promotional “firsts,” and platform visibility Emami Art (2021), Han (2017). Its critical force lies less in overcoming these contradictions than in failing to fully resolve them. That incompletion is not something to be corrected too quickly. It may be one of the few conditions under which critique can still survive once novelty and market legibility have already become part of the work’s public life.

 

CONFLICT OF INTERESTS

None. 

 

ACKNOWLEDGMENTS

None.

 

REFERENCES

Adorno, T. W. (1973). Negative Dialectics. Seabury Press.

Agrawal, H. (2018). The Anatomy Lesson of Dr. Algorithm [Artwork].

Agrawal, H. (2019). Masked Reality Interactive Work [Interactive artwork].

Baas, M. (2024). Artificial Intelligence and the Question of Creativity: Art, Data and the Sociocultural Archive of AI-Images. Journal of Visual Art Practice, 23(1), 31–46. https://doi.org/10.1177/13675494241246640

Baudrillard, J. (1994). Simulacra and Simulation. University of Michigan Press. https://doi.org/10.3998/mpub.9904

Bishop, C. (2012, September). Digital Divide: Contemporary Art and New Media. Artforum, 51(1), 434–441.

Cohen, H. (1995). The Further Exploits of AARON, Painter. Stanford Humanities Review, 4(2), 141–158.

Crawford, K., and Paglen, T. (2021). Excavating AI: The Politics of Images in Machine Learning Training Sets. AI and Society, 36, 1105–1116. https://doi.org/10.1007/s00146-021-01301-1

Emami Art. (2021, September 11). EXO-stential: AI Musings on the Posthuman [Exhibition release].

Gawronski, A. (2021). Art as Critique Under Neoliberalism: Negativity Undoing Economic Naturalism. Arts, 10(1), Article 10. https://doi.org/10.3390/arts10010011

Han, B.-C. (2017). Saving Beauty. Polity.

Mohr, M. (n.d.). Manfred Mohr.

NVIDIA. (n.d.). 64/1 and Harshit Agrawal at the AI Art Gallery.