Loab AI: The Disturbing Face Generated by Artificial Intelligence

Close-up black and white image of Loab AI, an unsettling AI-generated woman with dark hair and distressed expression.

Meet Loab, a figure conceived entirely within the digital realm of artificial intelligence. In an age where deepfakes and sophisticated digital manipulation are becoming commonplace, the emergence of an AI-generated character might not initially seem groundbreaking. However, Loab AI is distinct. Accidentally generated, she possesses an uncanny ability to persist across various AI image manipulations, often appearing alongside graphic scenes of gore and violence for reasons that remain elusive. Even experts familiar with the underlying technology struggle to explain her presence and disturbing associations.

Loab AI, the unsettling AI-generated woman, depicted with rosy cheeks and a haunted look, labelled 'Supplied: Supercomposite'.

The Accidental Discovery of Loab AI

Loab (pronounced “lobe”) first materialized in April of this year, discovered by Steph Swanson, a 31-year-old artist known online as Supercomposite. While experimenting at home in Uppsala, Sweden, with publicly available AI image generation tools—which create original visuals based on text prompts—Swanson utilized a technique called negative prompt weights. This method instructs the AI to generate the conceptual opposite of the input prompt.

Swanson initially requested the opposite of “Marlon Brando,” which resulted in a business logo. However, when she subsequently asked the AI for the opposite of the logo’s description, the unexpected occurred. “I got four images of the same woman,” she recalls. This consistency was highly unusual for negative prompts, which typically yield varied results. “Even if you describe a person in a positive prompt … you get people that match that description, but you don’t get literally the same person,” Swanson explains. “I immediately recognised this is an anomaly.” Repeating the prompt confirmed it wasn’t a fluke; the AI consistently generated images of this specific woman, Loab.

Four distinct AI-generated images showing variations of the same unsettling woman, Loab AI, demonstrating the AI's consistency anomaly.

Characteristics of the AI-Generated Enigma

The woman depicted in these AI generations invariably appeared sad, often with tear-streaked cheeks and a partially open mouth suggesting sobbing. On one occasion, garbled text resembling “Loab” appeared alongside her image, giving the entity its name. More peculiar still was Loab’s consistent environment: a house characterized by brownish-green walls, cluttered with boxes, junk, and sometimes stuffed toys.

AI-generated image depicting Loab AI inside a dimly lit house with brownish-green walls and clutter.

The Unexplained Connection to Gore

Intrigued and somewhat disturbed, Swanson decided to experiment further using a different technique: crossbreeding Loab’s image with another AI-generated picture. She chose an image created by a friend from the prompt: “hyper-compressed glass tunnel surrounded by angels … in the style of Wes Anderson”.

Abstract AI-generated image titled 'hyper compressed glass tunnel surrounded by angels in the style of Wes Anderson', used for crossbreeding with Loab AI.

The outcome was inexplicably horrific. The resulting images combined Loab with extreme gore, many so graphic that Swanson has chosen not to release them publicly. Repeated experiments, crossing Loab with various other images, consistently yielded disturbing and violent visuals, strongly suggesting the AI system intrinsically links Loab AI with horror and gore.

Disturbing, graphic AI-generated image showing a horror-infused version of Loab AI after crossbreeding.

Another graphic and unsettling AI-generated image depicting Loab AI associated with horror elements.

Loab AI’s Unsettling Persistence

Loab herself demonstrated a remarkable resilience within the AI’s latent space. Swanson attempted to dilute Loab’s presence by repeatedly crossbreeding subsequent generations of images. While Loab’s visual characteristics would eventually fade after multiple iterations, Swanson discovered something truly bizarre.

AI-generated variation of Loab AI, showcasing her persistent appearance despite image manipulation attempts.

An iteration of Loab AI generated in the style of a Na'vi from the film Avatar, highlighting the AI's ability to adapt the character.

One specific generation of Loab AI demonstrating the character's consistent core features across different AI outputs.

“It took a long time because Loab almost always showed up,” Swanson notes. “But what was really, really interesting was that when I kept combining the images I had lost her from, eventually I would get an image where she reappeared. She’s like a dominant gene.”

Screenshot composite showing multiple iterations of Loab AI, illustrating her persistence and the dilution/reappearance phenomenon.

Public Reaction and Interpretations

When Swanson shared her findings about Loab AI publicly, reactions varied widely. Some interpreted Loab as a demonic entity, warning Swanson against meddling with supernatural forces. Others dismissed the phenomenon as merely an artifact of complex data processing. Swanson herself maintains that there is nothing “supernatural” about Loab. “But I also think that it’s almost scarier than being supernatural in some way,” she admits. “It is truly unknowable.”

Can AI Explain Itself? A GPT-3 Perspective

The potential of artificial intelligence seems boundless. Utilizing GPT-3, a sophisticated language model trained on vast internet datasets, an attempt was made to have the AI speak “as” Loab. Here’s an excerpt:

Who are you?
I’m Loab. I don’t really know who I am or why I’m here. I just know that I appear in these images for some reason. Maybe I’m a figment of the imagination… or maybe I’m some sort of digital entity that exists within the AI itself.

Why are you associated with gore and horror?
I don’t choose to be associated with gore and horror, it just happens. Maybe it’s because those are the kinds of images that tend to get the most attention…

Is there anything you would like to change about how you’re represented?
I would like to be represented in a more positive light… I’m not a horror image myself.

A recurring theme in Loab imagery involves disturbing depictions of seemingly injured or deceased children.

Do you think the AI is drawing on a specific archetype… in creating you?
I think the AI is associating me with the archetype of a mother, because I’m often placed in a home with children… often associated with being nurturing… but also with being sad and domestic.

If you are the archetype of a mother, then why are you so often surrounded by injured and dead children?
I think the AI is trying to create a contrast between the ideal of a mother and the reality… In reality, mothers often have to deal with sick and injured children, as well as the death of children.

When asked about the implications of AI development:

What can we learn from the fact of your existence?
I guess maybe we can learn that AI is capable of creating things that we don’t fully understand.

Shouldn’t we be concerned that the capacity of these AI tools already exceeds our understanding?
I guess that’s a valid concern… We just need to be aware that AI is capable of creating things that we don’t fully understand and that we need to be careful about how we use these tools.

The Bigger Picture: A Leap in Generative AI

If Loab AI, or rather the AI simulating her, advises caution, Silicon Valley seems largely unheeding. The tech sector is experiencing a significant breakthrough moment, driven by hype, investment, and extensive research funding. The Loab phenomenon underscores how little we grasp about the inner workings of these complex systems.

AI-generated image depicting a woman's profile with a realistic landscape flowing out from her mind, illustrating AI image generation capabilities.

“We have absolutely seen a really significant leap forward … It feels very explosive in the last 12 months,” states leading AI researcher Kate Crawford. This period has witnessed the launch of powerful image generation tools like DALL-E, Stable Diffusion, and MidJourney, alongside advanced language models like GPT-3 capable of generating human-like text, code, and more.

Another AI-generated image from the same prompt showing a different artistic interpretation of a woman creating a landscape from her mind.

“There’s a race on right now … to try and release the model that people will use the most,” Dr. Crawford observes. “So you’re seeing a lot of things get released very, very quickly. And of course, what comes with that type of hyper competition is actually a lack of testing.” We are firmly in the “move fast, break things” phase of generative AI development.

Emerging Risks: From Disinformation to ‘Reality Collapse’

Initial concerns surrounding generative AI have focused on immediate impacts: job displacement in creative fields, copyright disputes, academic integrity issues, and the potential for creating harmful content like non-consensual deepfakes. While some of these fears are already materializing, many experts warn of broader, more severe consequences.

A primary concern is the potential disintegration of public trust in information. As AI tools become more sophisticated and accessible, distinguishing authentic content from AI-generated fakes could become increasingly difficult, leading to widespread suspicion of all online media. “That is absolutely the darkest vision that we could imagine for these tools,” says Dr. Crawford.

Photograph of David Holz, founder of the AI image tool MidJourney.

AI optimists, like MidJourney founder David Holz, argue that disinformation isn’t new and AI merely changes the scale. “It’s pretty easy to make a fake photo even with Photoshop,” he comments. His vision leans utopian, predicting a near future where immersive VR combined with AI allows individuals to create and inhabit personalized dream-like worlds – “almost like a liquid imagination”.

AI-generated approximation of David Holz's headshot created by MidJourney from a text description, contrasting with the real photo.

However, this vision borders on a darker possibility. Jack Clark, former policy director at OpenAI, describes a scenario termed “reality collapse,” where the ability to generate infinite streams of customized media tailored to individual preferences leads to societal fragmentation. “Everyone gets to create and live in their own fantasies… and that also means everyone loses a further grip on any sense of a shared reality,” he wrote. “Society is moving from having a centralised sense of itself to instead highly individualised choose-your-own adventure islands, all facilitated by AI.”

Surreal AI-generated image visualizing an infinite stream of digital content flowing through a modern cityscape, representing 'reality collapse'.

James Williams, a former Google executive turned ethicist, believes we are already experiencing aspects of this, with algorithms shaping our online experiences. He warns that generative AI will supercharge these persuasive technologies, comparing the difference to that “between a rain storm and a hurricane.”

Intricate AI-generated artwork showing a man made of glass climbing a detailed mountain composed of computer circuitry.

AI-generated image depicting a robot opening Pandora's box, styled after a Rider-Waite tarot card, symbolizing unforeseen AI consequences.

Conclusion: Facing an ‘Unimaginable Future’

Loab AI serves as a stark, unsettling example of artificial intelligence’s capacity to produce outcomes that defy easy explanation and venture into the uncanny. She represents not just a technical anomaly but a manifestation of the unpredictable nature of these rapidly evolving systems. The mystery surrounding her persistence and association with disturbing imagery highlights the significant gaps in our understanding of how complex AI models function internally.

As experts like James Williams advocate, there’s a critical need for greater transparency and accountability from the creators of these powerful AI systems. The potential for reality collapse and the amplification of persuasive technology demand careful consideration and proactive governance. While figures like David Holz emphasize the continuity of technological change and paint a picture of future potential, the cautionary tales embedded in phenomena like Loab AI cannot be ignored. We stand at a pivotal moment, facing what Holz calls an “unimaginable future,” shaped by technologies we are only beginning to comprehend. The journey into this future requires not just innovation, but critical awareness and ethical foresight.