Loab: Subjectivity of the Digital Sublime

In the contemporary technological landscape, it has been largely accepted that generative AI is based purely on algorithmic logic and rational science. At the same time however, when generative AI produces something disturbing or perverse the supposedly logical science of AI is undermined. The image and the science behind it accumulate a sinister prophetic quality. Indeed, the entirely unprompted disturbing images produced through particular prompts might even make users consider the possibility of AI possessing subjectivity. In my paper, I use Loab to explore how generative AI is used to condition our responses to certain stimuli. At the same time I consider the consumer’s fascination with the uncanny in this context, and how this can be manipulated by large AI companies to condition our responses to their product. How can our memories and nostalgia for what is lost be manipulated to provoke us? How can generative image, text, and video such as LOAB change the social fabric?

Notorious digital cryptid Loab- who was given this name after an album cover that was generated alongside her image that displayed the word ‘loab’- was created by artist Steph Maj Swanson, also known as Supercomposite. For this project,she used an AI generation technique called negative weighting. Negative weighting is an AI generation technique that allows the user to produce a generative response that is most dissimilar from the original prompt. For instance, using the prompt Brando::-1, Swanson was able to produce an image of a logo with the words DIGITA PNTICS. If it were reversed, it must be assumed that the logo would produce a picture of Marlon Brando. Instead, the AI produced an image of a distorted and devastated older woman, which Swanson nicknamed Loab. Swanson speculated that Loab was positioned in latent space relatively close to macabre and gory imagery- when she attempted to dilute Loab and combine her with related images, she struggled, and therefore she deduced that the creature must be isolated in the map.

Fig 1: DIGITA PNTICS logo posted on X by Supercomposite, 2022

On the surface, Loab is a simplistic perversion of the superficial. The generated image loosely depicts an older woman bloated with plastic surgery, rosacea on her cheeks.Rolling Stones writer Miles Klee states that Loab ‘stigmatises disability’ and is a ‘reflection of how our culture mistreats those deemed less attractive.’ When we study the primary image, it conjures nostalgic associations, with Loab clutching what looks like a teddy bear and the background similar to the set of a sitcom. Because generative AI is trained on our own resources, it becomes a reflection of our humanity. When it reveals something sinister it is a frightening mirror to our own condition. The more we indulge in this, the more space it seems to take up. While this is true, we simultaneously see this tool as something unable to be entirely comprehended or understood. It has its own rules of logic unable to be accessed by humans. The prospect of this logic possessing a darkness beyond our comprehension is deeply disturbing.

Fig 2. First Image of Loab shared on X by Supercomposite, 2022

Rahul Murdeshwar in his paper Digita Pntics: AI Umwelten, Object Ontologies and the Ghost in the Machine writes:

“But why did it generate a skyline or a portrait to begin with? With Digita Pntics and Loab, what is truly eerie is that the text prompt did not suggest horror at all. What terrifies us is the potential of the machine to not just disobey but imagine laterally- or pathologically, as Osborne would put it (2016). Implied here is that the AI is capable of a logic which we do not completely understand and did not explicitly program into it. The uncanny comes from the possibility that there is a subjectivity where, according to our anthropocentric ontologies, there should not be one. Recalling the many interpretations of Derrida’s hauntology in media and technology (Blanco and Peeren 2013), there is the apparition of a ghost in the machine, to borrow Gilbert Ryle’s famous phrase. We see through Loab’s eyes the visor effect of the spectre looking back at us. This is a subjectivity capable of affective experience, although not in the way we understand affect.”

The corruption of our positive connections in this way could be manipulated. This macabre depiction of what feels comfortable and normal becomes a fascination, and keeps usage high. Supercomposite states themself that they want to avoid starting “some kind of viral trend of people making gory stuff with the tools I’ve used,” aware of the dangers of sensationalising the cryptid. With LOAB making its way into the social fabric, it is now ingrained into the latent space map, solidifying her position in popular culture and the generative AI feature map. The Uncanny-Valley qualities (a term coined by Japanese roboticist Masahiro Mori, which describes when an artificial subject resembles an organic one closely but not entirely, and therefore provokes feelings of fear and unsettlement) of LOAB can and have been disturbingly manipulated, beyond Supercomposite’s aims and anticipations for this project. The popularity of such pieces of art reveal a tendency for nostalgic media to be used to instil mass panic and fascination with these aspects of generative AI. This is distinctive from the more subtle and calculated political and social changes with the rise in generative AI tools that have been observed. Social media users flocked to Trump’s infamous AI generated ‘Gaza video’- which depicted the American president dancing on the colonised beaches of Gaza with Elon Musk and Israeli president Benjamin Netanyahu, artificial intelligence was being used on the very same Gazan beaches to target drones to civilians.

Intrigue and apprehension surrounding the possibility of AGI in the near future is still relevant. In 2024, actress Scarlett Johansson (who voiced the flirtatious AI assistant Samantha, in the movie ‘Her’, 2013) spoke up against OpenAI after the release of its anticipated new AI assistant ‘Sky’. Johansson claimed she had declined the opportunity to voice the chatbot, and was ‘shocked’ and ‘angered’ that the company had decided to opt for a voice actress who seemed to intentionally imitate Johansson’s own voice. CEO Sam Altman declined that this was a conscious decision after the backlash, yet in the weeks leading up to the release of ‘Sky’ had publicly tweeted the word ‘her’. Aside from the myriad of ethical problems that could occur as a result of this marketing choice, OpenAI’s push for more personable AI systems as a response to public discussion is a glimpse at the future of AI products. Whilst it is beyond the scope of this essay, it is interesting to consider how big AI corporations such as OpenAI might push for an increase of generative responses similar to that of LOAB. Will public fascination with horror and mystery result in changes in the algorithm to profit from and sate these desires?

In his book ‘Convergence Culture: Where Old and New Media Collide’, media scholar Henry Jenkins writes:

“Transmedia storytelling is not entirely new. Take, for example, the story of Jesus as told in the Middle Ages. Unless you were literate, Jesus was not rooted in a book but was something you encountered at multiple levels in your culture. Each representation (a stained-glass window, a tapestry, a psalm, a sermon, a live performance) assumed that you already knew the character and his story from someplace else.”

One of the reasons LOAB’s imagery is so powerful is because she invites us to access our collective cultural memories. Without fully understanding why, consumers are asked to draw assimilations between themselves and her- imprinting our fear, desire and curiosity into her ‘personhood’. What do our connections and assimilations with LOAB say about our humanity? Our curiosity and fascination with violence and gore has been reflected back at us through the manifestation of this cryptid.

Joel Warner, a writer for Wired magazine, argues that LOAB is just the beginning, a symptom of a larger problem that will begin to work itself into the fabric of generative AI. He predicts that:

“As AI text generators become ever more ubiquitous in online conversations, search interfaces, and other forms of digital communication, the linguistic version of Loab, or whatever you want to call her, will emerge again and again. Either through intentional prompts or accidental word combinations, these programs will begin churning out ideas every bit as disturbing as 120 Days of Sodom.”

In other words, our AI interfaces will be a direct reflection of ourselves, with all the same benefits and dangers of humanity as it exists, but now publicly and globally available. While the infamous 120 Days of Sodom by Marquis de Sade was hidden and obscured for a considerable length of time, the impacts of the accessibility of this type of new media is immeasurable.

It’s not all bad. AI image generation lowers the cost of medical research and training, can be used in restoration and makes learning accessible. For perhaps the first time ever, we have the key to solving educational and healthcare inequalities and lack of resources. By predicting weather patterns, crop yields and analyzing soil conditions, we can use AI to revolutionise agriculture and farming. We can code complex algorithms more easily and detect fraud and scamming. For the millennia that humans have been thinking and fantasising about human replication and cloning, we are now at the closest point to that reality. We can upload our consciousness onto a digital replica and live (superficially, of course) forever. Seemingly, nothing is untouchable. Loab encapsulates the digital sublime.

Loab is corrupt because she can be corrupted. With the algorithmic feedback loop in place, there is an incentive to push for and cultivate a dependency on these technologies. In my paper, I considered the user’s fascination with the uncanny, macabre and violent, and how this can contribute to the feedback loop that values and incentivises similar generative responses, both on a capitalistic and technical scale. I explored the subjectivity of Loab and the implications that AI is capable of a logic that we do not understand, as well as the suggestive qualities of Loab’s superficial appearance. It is imperative that we are aware as consumers of our relationship with generative AI and how we share, discuss and engage with it in the public sphere. The critical questions we pose today regarding generative AI and those who program it will pave our relationship with the AI systems that threaten our very existence as we know it.

Bibliography

Book

Jenkins, H. (2006). Convergence Culture: Where Old and New Media Collide. New York: New York University Press.

Online Sources and Articles

AgriRS (2024). The use of AI (Artificial Intelligence) in agriculture and farming. Available at: https://www.agrirs.co.uk/blog/2024/02/the-use-of-ai-artificial-intelligence-in-agriculture-and-farming?source=google.com (Accessed: December 2025).

CNET (n.d.). What is Loab? The haunting AI art woman explained. Available at: https://www.cnet.com/science/what-is-loab-the-haunting-ai-art-woman-explained/ (Accessed: November 2025).

GeeksforGeeks (n.d.). Latent Space in Deep Learning. Available at: https://www.geeksforgeeks.org/deep-learning/latent-space-in-deep-learning/ (Accessed: November 2025).

The Guardian (2024). Israel accused of using AI database to help choose Gaza air strike targets. Published 3 April 2024. Available at: https://www.theguardian.com/world/2024/apr/03/israel-gaza-ai-database-hamas-airstrikes (Accessed: December 2025).

The Guardian (2024). ChatGPT’s ‘Sky’ voice: why Scarlett Johansson is angry at OpenAI. Published 20 May 2024. Available at: https://www.theguardian.com/technology/article/2024/may/20/chatgpt-scarlett-johansson-voice (Accessed: December 2025).

The Guardian (2025). Trump Gaza AI video intended as political satire, says creator. Published 6 March 2025. Available at: https://www.theguardian.com/technology/2025/mar/06/trump-gaza-ai-video-intended-as-political-satire-says-creator (Accessed: December 2025).

IBM (n.d.). What is latent space? Available at: https://www.ibm.com/think/topics/latent-space (Accessed: November 2025).

Kill Your Darlings (n.d.). Digital manipulation and the uncanny valley. Available at: https://www.killyourdarlings.com.au/article/digital-manipulation-and-the-uncanny-valley/ (Accessed: December 2025).

Murdeshwar, R. (n.d.). Digita Pntics: AI Umwelten, Object Ontologies and the Ghost in the Machine. Available at: https://www.hortussemioticus.ut.ee/blog/speak-semiotics/digita-pntics-ai-umwelten-object-ontologies-and-the-ghost-in-the-machine/ (Accessed: December 2025).

Reddit (2023). What are some examples of internet born [mythology]?. Available at: https://www.reddit.com/r/mythology/comments/17sq7tw/what_are_some_examples_of_internet_born/ (Accessed: November 2025).

Smithsonian Magazine (n.d.). Loab, Artificial Intelligence Art. Available at: https://www.smithsonianmag.com/smart-news/loab-artificial-intelligence-art-180980743/ (Accessed: November 2025).

Supercomposite [@supercomposite] (2022). [Tweet/X post about LOAB]. 6 September. Available at: https://x.com/supercomposite/status/1567162288087470081 (Accessed: November 2025).

Wikipedia (n.d.). Loab. Available at: https://en.wikipedia.org/wiki/Loab (Accessed: November 2025).

Wired (n.d.). The Linguistic Version of Loab is Going to Be Terrible. Available at: https://www.wired.com/story/generative-ai-marquis-de-sade-chatgpt/ (Accessed: November 2025).