ChatGPT Did Not Take Place

| Andrew Goodman & Anna Munster
 + Screenshot from 'ChatFOS Pitch'  Zoe Horn, Liam Magee and Anna Munster, 2023.

ChatGPT Did Not Take Place

ChatGPT Did Not Take Place | Andrew Goodman & Anna Munster

Screen left: Executive-type addresses the camera (#generic_white_guy)...

Screen right: A worker (#hi-vis) kneels thoughtfully in a placeless warehouse…

Bottom of screen: 'Logistics delivered, powered by AI…'

Executive in mid-American accent: 'Good afternoon, everyone. I'm David Palmer, CEO of ChatFOS, and I'm here to talk to you today about how we're transforming logistics with AI-generated design…'

A chimaera, onanistic recirculation of image-data; a strip-mined algorithmic ecology of representation, redundancy; a mirror scraped from the surface of the world. A simulation floating, a synthetic version of Jean Baudrillard’s ‘people who pretend to have ideas.’1 Yet nothing here floats free, as an algorithmics whose tentacles of Lovecraftian whiteness2 are deeply rooted in histories of exclusion, extraction, plantation; embedded in logics of logistics, 'converting everything in its path into a coordinated time and space for ownership.’3

Can we say that ChatGPT will not take place, is not taking place, did not take place as thought, as creative advance, as event? Is it too simplistic to dismiss both the panic and enthusiasm for the coming of AI—based as they both are on the acceptance of intelligence as imitation and representation—when its effects are all too real? The bodies that get organised, thoughtlessly (as labour, as goods, as metrics), and the bodies that get to do the organising (to think) and direct movement, matters. Some of the population—#high vis:  precarious contract warehouse labourers, Kenyan HIT workers, sweatshop workers generating machine-learning statistics,4 and all of us whose content is scraped by current generative AI—is made to move by these plantation-inspired and racialised industries and those that manage their logistics. The origin story of these early capitalist/colonial experiments begins, McKittrick notes, with the movement of bodies-as-goods of the Atlantic Passage slavetrade.5 If we start to probe that deep connection, automation will shut us down. As Auriea Harvey, an African American artist, discovered when her probes of Midjourney for generative images of slave ships were met with a threat of account suspension.6

...

Baudrillard famously argued that simulations constituted by media increasingly defined the social to the extent that simulation took precedence over ‘the real’. For Baudrillard the Gulf War did not take place in reality because it had already occurred and generated real effects via its mediality.7 The punchiness back then of this concept of simulation—the sign which goes beyond perverting the real and hiding the absence of the real to become an image bearing ‘no relation to any reality whatever’8—owed much to a historical moment in which it appeared possible to hive images off from the real. Today, an endless self-reinforcing recursion across content (data) and its logistical management (algorithmics) is taking place instead. ChatGPT and the large language model spawning of image, video, music, and even robotic prompt management, drains the real in an overcoding, surfacing and segmenting of everyday and speculative horizons, sheathing the earth in dislocated images. Even so, certain surfaces shine more brightly than others. While Baudrillard pithily pinpointed the rise of simulacra, he failed to account for the persistence of a singular image of intelligence—white, corporate, neurotypical, gendered male. A singular image of thought(mind)-action(body) subtends and maintains the generic aesthetics of both today’s Chatbots and the infrastructures of control such as platforms, cloud services, venture capital and supply chain management through which generative AI is realised.   

What counts as ‘intelligence’ gets too easily lost in discussions of AI. If we call ChatGPT ‘AI’, this firstly suggests, as do more reductive interpretations of Alan Turing’s eponymous test,9 that imitation is the same as thought. Secondly, it prioritises the management and supervision of data, as does psychology’s emphasis on ‘executive function’—that homunculus-like and equally mythic process (#white_executive) overseeing and ordering the brain’s warehouse of sensa into manageable sets (#logistics_of_thought). Such prioritisation of not only the content of thought but also the management of thought and emotion places value on the self-administration and monitoring of thinking. After Foucault we might call this a governmentality of thought: an imprint or map of state architectures of power impressed not only on thought (content) but modes of thinking (its logistics),10 which could be termed a whiteness of thought-as-possession.11 This process devalues thinking bodies deemed incapable of managing themselves—keeping their thoughts moving tidily within their own architectures of control. The devaluing of certain modes of thinking has enacted violence on many peoples (and other earthly creatures) whose thoughts have been deemed too wild and more-or-less-than-human in their modes of living—all whose thinking is deemed unquantifiable, incalculable or unmanageable.12 The reduction of thought to logistics is an ever-present ghost in the AI machinery, as it blankets thought with a quieting whiteness.

Enter ChatFOS (Fulfilment Operation System), a real simulation of an AI design agency whose team, investors, services and clients do not exist but who may as well.13 Generated entirely on the back of OpenAI’s ChatGPT, a ‘Bot’—ChatFOS—was tasked with developing prompts to generate media and text simulating an enterprise servicing the logistics industry. Its design and aesthetics were drawn from the online ‘image’ of current logistical and platform capitalism, from Microsoft’s Azure to Autostore, a business-to-business enterprise supplying automated equipment to warehouses. ChatFOS’ prompt-generated media outputs assemble a parodic simulation of AI’s emerging functionalities within logistical worlds that organise and circulate the global movement of goods and labour. ChatFOS exists as a website, YouTube pitch investor video, and LinkedIn social media account. And yet ChatFOS does not take place except to test the limits of logistical media.

 + Screenshot from 'ChatFOS Pitch'  Zoe Horn, Liam Magee and Anna Munster, 2023.

Screen right:  #WestBengalesewoman checks inventory in a dusty warehouse that glows with golden afternoon light …

Screen left top: #futurewarehouse transformed by AI…

Screen left bottom: #futurewarehouse transformed by AI…

Voiceover: #synthesizedIndianaccentedEnglish: ‘ChatFOS have provided us with real time insights into inventory, sales and customer behaviour…’

Feeding bot-generated prompts to a range of AI generative apps,14 revealed how large language model-based AI is fast becoming a (logistical) router that governs and corrals production of all media: image prompts, website code, promotional copy, and investor pitch scenarios, for example. These elements become links chained together to form the generic aesthetics of corporate websites and promotional videos. The images that ChatFOS simulates have a genericism embodying a certain dominant ‘image’ and value set for contemporary AI. When trying to create a synthetic voice for an AI-generated ChatFOS client, ‘Amitava’ (fictional CEO of a chain of warehouses in West Bengal), it became impossible to remove or counter the English and American inflections built into presets of the model. While logistics industries have global reach, ‘an’ image, ‘a’ voice recirculates, which re-integrates the global into the vortex of whiteness.

Surfaces shine but don’t always hold. In prompting platforms such as Midjourney to produce images of simulation itself, glitches (cracks in the management of calculation, incalcuabilities) appear in its plasticised surface. /imagine: ‘Visualize FOS's AI-powered visual services that provide real-time insights into customer behaviour. The image should showcase data visualisation and digital screens that highlight customer reviews, product images, and other elements that provide insights into customer behaviour.’ The resultant image features a large screen inside a warehouse environment in which the reality of the stacked shelves literally punches through; a punctum in which the absent spaces of customer and precarious worker nonetheless push back.

 + Screenshot from 'ChatFOS Pitch'  Zoe Horn, Liam Magee and Anna Munster, 2023.

In the end, the #generic_white_guys fight it out among themselves, dominating the conversation as they always do. In the ChatFOS YouTube video, staged as a virtual event, the design agency pitches for Series A venture capital funding, another #generic_white_guy voice interjects in  ‘Zuckerberg’ fast-talking, corporate crash and burn style: ‘I’m not sure what makes you think your company is going to be any different. You’re talking about optimising supply chain management with AI-generated content. That sounds great in theory but I’m not sure how it’s going to work in practice’... In practice, of course, the #generic_white_guys control what the conversation is, where the cage fight occurs, which images and thoughts return, dominating and ossifying.

In the midst of banal, synthetic post-neoliberal simulation aesthetics, art might foreground what the future has already become: recursive, autophagic, unimaginative, digital plantation wastelands (predicated on and generated through an empty algorithmic logistics that is ‘the science of whiteness…the science of loss’.15 Techniques for deploying and assembling (de)generative AI via a scenario such as ChatFOS (Full Of Shit) can be thought of as speculative absurdism. These are temporary experiments in probing ‘Chat’ at its most imperious, pushing at and humorously deflating its attempts to engulf the world in its genericism.


Notes

1. Jean Baudrillard, Cool Memories II, trans. Chris Turner (Durham: Duke University Press), 29.

2. N.K. Jemisin “N.K. Jemisin : The City We Became,” Between the Covers Podcast, Tin House, June 1, 2020; Denise Ferreira Da Silva, “Episode_00: Black Poethics & Contagious Architecture,” Recursive Colonialism. YouTube. https://www.youtube.com/watch?v=dUfAqouAdnw

3. Stefano Harney and Fred Moten, ‘Habits of Assembly’, in Assembly Codes: The Logistics of Media. Eds. M. Hockenberry, N. Starosielski, and S. Zieger (Durham: Duke University Press, 2022), 27.

4. Williams, Micelli & Timnit, “The Exploited Labor Behind Artificial Intelligence,” Noema, 2022.

5. Katherine McKittrick, "Mathematics Black Life." The Black Scholar 12, no. 2 (2014): 17: Simone Browne, "Digital Epidermalization: Race, Identity and Biometrics." Critical Sociology 36, no. 1 (2009): 139.

6. Harvey in Small, 2023 (Zachary Small, July 5, 2023. ‘Black Artists See Clear Bias in A.I’, New York Times, https://www.nytimes.com/2023/07/04/arts/design/black-artists-bias-ai.html#:~:text=Many%20Black%20artists%20are%20finding,programs%20that%20run%20the%20algorithms.

7. Jean Baudrillard, The Gulf War Did Not Take Place, trans. Paul Patton (Power Publications, 1995).

8. Jean Baudrillard, selected writings, trans. Jacques Mourrain et al (California: Stanford University Press, 2002), 170.

9. The Turing test was developed in 1950 by Alan Turing. Originally called 'The Imitation Game', the Turing Test requires a machine to generate answers to questions that are (to a human) indistinguishable from human answers.

10. Michel Foucault, Security, Territory, Population: Lectures at the College De France, 1977-78,trans. Graham Burchell (Hampshire: Palgrave MacMillan, 2009), 94. Although Foucault does not use the term governmentality directly in relation to thought, he draws clear parallels between the disciplinary functions of governmentality for the state and that of segregation for psychiatry. Ibid., 120.

11. Moten, Fred, and Stefano Harney. All Incomplete. (Colchester: Minor Compositions, 2021), 14.

12. Fred Moten, Stolen Life: Consent Not to Be an Individual (Durham: Duke University Press, 2018), 164.

13. ChatFOS is a collaborative project by Zoe Horn, Liam Magee and Anna Munster. It has a website (https://chatfos.ai/ ), YouTube promotional videos ((https://www.youtube.com/watch?v=i4lsx3zCdn4&t=58s ), and social media accounts.

14. Midjourney, Bard, ChatGPT, Elevenlabs, ControlNet, D:ID, for example.

15. Moten and Harney. All Incomplete, 15.

Links & Info
Cite this ArticleCite
 Colophon

Art and Australia logo


Art + Australia
Publisher: Victorian College of the Arts
University of Melbourne


Art + Australia ISSN 1837-2422


All content published after October 2023 by Art + Australia is available under a Attribution-NonCommercial-NoDerivs 4.0 International Licence (CC BY-NC-ND 4.0) except where otherwise stated. For more information about use and distribution you can view our Editorial Guidelines.