Meet the Woman Making Viral Portraits of Mental Health on TikTok

Until a few years ago, Artificial Intelligence (AI) was in the unreachable domain of tech corporations, algorithms and the dark web. But as the technology improves, its accessibility has widened. The internet is peppered with free AI-generator tools where you can make anything from uncanny fanfic to trippy hellscapes. There are hundreds of crooked recipes to peruse and an entire blog dedicated to AI’s “weirdness”. Now, AI has found a home in the art world – and it’s taking off on TikTok. 

Recently, during my nightly TikTok doomscroll, I happened across something I haven’t been able to stop thinking about. Over a deeply cursed sound – the kind of generic stock audio now associated with TikTok mermaid sightings, conjurings, and conspiracies – a blonde-haired woman softly spoke to the camera about her latest project.

“I’ve been using AI to make creepy images, and I’ve been putting off doing this one for a while,” she said.

Under the headline, “Asking an AI to Show Us its Face,” a dark shadow on a blank background twitched and morphed. More details gradually etched onto the screen, filling the space with two eyes, a beard, gashes for nostrils and a mouth. Characters, as if written in blood or iodised into paper, slowly appeared at the top right-hand corner of the piece. 

The final artwork came together in seconds. On the user’s page there were hundreds more, with titles like “Asking the AI for a Non-Binary Partner” and “The AI Gives Us All a Panic Attack”. 

The creator is Kier Spilsbury, a Brisbane-based legal researcher who uses open-source AI software to make art. In July this year, Spilsbury’s friends happened across a tweet containing a link to the AI software she uses, as well as a document with detailed instructions on how to use it, written in Spanish. 

“Once you get the hang of it, it's really simple. I think that's what's so interesting, it's just so accessible,” the 35 year-old told VICE. “Even in Spanish, basically anyone could use it. I just got obsessed with it.”

Every day, Spilsbury receives dozens of requests from her 30,000-strong TikTok following to feed new terms, phrases, and descriptors into the AI generator.

She recently began a series asking the AI to produce representations of mental health conditions, including Borderline Personality Disorder, ADHD and Complex PTSD. The images the generator creates are strangely resonant, beautifully cohesive and, frankly, fucking bizarre.

This particular AI, named VQGAN+CLIP, works by rapidly troweling through billions of Google images to create artworks based on input text prompts. Even slight variations in the words used can produce vastly different results. For the mental health series, Spilsbury has been experimenting with giving the AI intentionally vague prompts.

“I kind of don't want to control too much what comes out,” she said. “Because I think the really interesting thing is just seeing how the AI interprets images that have been tagged [on Google] with that sort of thing.” 

Spilsbury is among hundreds of TikTok creators platforming AI art. There’s a whole genre of videos that ask AI to draw answers to The Big Questions: things our tiny human brains can’t conceive. Videos like “An AI Creates The End”, “What’s Inside a Black Hole” or “Asking Advanced Artificial Intelligence What ‘Sleep Paralysis’ Looks Like” amass views in the thousands, sometimes millions.

VQGAN+CLIP is a combination of three different AI “brains”. The “GAN” part is the software used to produce deepfakes: which, in one of the more dystopian AI applications, is generally used to create videos of events that never happened. Deepfakes are all over TikTok in their most harmless form: creators turning themselves into celebrities and Queen Elizabeth getting down. Fun gimmicks aside, though, the mere existence of deepfakes signals the sinister and underregulated complications of technological advancement, dubbed by experts as the “Infocalypse”. They can be used to spread disinformation, produce nonconsensual pornography and commit fraud. And as the technology improves, the potential for harm increases.

In some cases, however, AI also has the potential for good – like allowing complex experiences to be represented in striking new ways. Most of the comments on Spilsbury’s mental health series come from people who are diagnosed with the neurodivergencies and mental health conditions the AI draws. At the end of each video, Spilsbury asks viewers whether the images resonate.

The comments section on Spilsbury’s ADHD video is filled with people connecting with the AI’s articulation of their lived experience.

“As someone with severe ADHD, this is spot on to how it feels inside my head quite often. This hit harder than I expected,” reads one.

“It looks like how I feel when I’m overstimulated,” reads another.

As we straddle the ever-blurring lines between technology and reality, art appears to be one of the more positive uses of AI. For Spilsbury, its use in creative spheres prompts exciting possibilities for the future.

“Obviously, a computer can't be creative. But it is doing something that is creative, it's just based on the creativity of people that have come before it,” she said.

“Human creativity isn't going anywhere, but AI is opening up this whole new way to create, which is very exciting.”

Follow Arielle on TikTok and Instagram.