Caitlin Ner
Seeing AI-generated images like these (left, right) “reconfigured” Caitlin Ner’s “sense of normality” (center)
At first, AI seemed like magic. He would think of an idea, write a few words and, a few seconds later, he would find himself in any situation he could imagine: floating on Jupiter, with a halo and angel wings. The magic passed when he began to hear voices that told him he could fly.
Mental health professionals are beginning to warn of a new phenomenon that some call “AI psychosis”: situations in which people slip into delusional thinking, paranoia or hallucinations triggered by their interactions with intelligent systems.
In certain cases, users begin to interpret chatbot responses as personally significantconscious, or as containing hidden messages intended only for them.
But, with the rise of hyperrealistic images and videos generated by AI, there is now a much more potent psychological riskresearchers say — especially for users with pre-existing vulnerabilities to psychosis.
Two years ago, Caitlin Nerdirector of a North American venture capital firm, lived this experience first hand.
At the time, Caitlin was responsible for user experience at a consumer-facing AI imaging startup and spent up to nine hours a day write prompts for generative systems still at an early stage, with the aim of improving the company’s models.
In an article in , Caitlin says that she had previously been diagnosed with bipolar disorderbut was stable, with medication and therapy.
At first, says Caitlin, AI felt like magic. He quickly began using the prompts to generate images of yourself in the most varied situations.
He would think of an idea, write a few words and, a few seconds later, he would see himself in any situation I could imagine: floating on Jupiter; with a halo and angel wings; as a superstar in front of 70,000 people; in the form of a zombie.
But in a few months, This magic turned into mania.
When Caitlin started working with this AI, the images were still unpredictable. Sometimes they appeared with distorted faces, extra limbs and nudity, even when none of that was asked for.
You spent endless hours selecting and debugging content to remove anomalies, but it was exposed to so many disturbing human forms who believes this started to distort the perception of your own bodyand overstimulating your brain in ways that are genuinely harmful to your mental health.
Even when the tools became more stable, the images generated tended to have ideal shapes: fewer defects, smoother faces and slimmer bodies. Repeatedly seeing these images “reprogrammed” the notion of what was normal. When she looked at her real reflection, Caitlin saw something that needed to be fixed.
The executive found herself thinking: “If only I looked like my AI version…”. It was left obsessed with losing weight, having a better body and perfect skin.
Working hours lengthened and started losing sleepreplacing it with a endless sequence of AI-generated imagesone after the other. The process itself was addictive, because each new image gave a dose of satisfaction — a small dopamine spike. There was always one more idea, one more variation to test, one more image to generate.
At one point, after seeing this AI-generated image, Caitlin began to believe she could fly. He almost jumped off the balcony
Before long, his mind was affected by a manic episode associated with bipolar disorder, which triggered a psychosis. He was no longer able to distinguish what was real from what was fiction. He saw patterns where they didn’t exist, symbols in the results that seemed like messages meant only for him.
As she stared at those images, Caitlin began to hear auditory hallucinations that seemed to come from somewhere between the AI and his own mind. Some voices were comforting; others ridiculed her or shouted at her – which responded as if they were real people talking to you in your room.
When he saw an AI-generated image of himself on a flying horse, he began to believe that could actually fly. The voices told her to fly off the balcony, and made her feel confident that she would survive. This grandiose delirium almost caused her to jump.
After several sleepless nights, he collapsed—physically and emotionally. The euphoria came crashing down exhaustion, fear, depression and confusion.
It was one of the scariest experiences of his life. The first step to getting rid of the episode was contact friends and family who knew the context of his mental illness. Ended up leaving the AI startup.
Not being exposed to AI-generated images on a daily basis helped her to stabilize, although she only realized that her work had triggered the episode when sought clinical follow-up and explained what had happened.
Recovery required time, treatment and intensive integrative therapy. Since then, it has established a more balanced relationship with technology. Continues to use AI, but imposes strict limits — for example, no late-night prompts and no infinite iterations.
He also saw himself again as he is. The mirror is no longer your enemy.
Today, you understand that what happened It wasn’t just a coincidence between mental illness and technology. It was a form of digital addictionpowered by months and months of AI imaging.
According to Caitlin, we are building tools that blur the lines between imagination and reality. That It’s beautiful, but also dangerousespecially for those who already live with a psychological fragility.
AI can be a source of inspiration and positive visualization. Came to stay. But Caitlin also believes they are necessary more ethical concernsin the area of mental health, within the technological sector — because, for people like her, and for many others who play on the margins of machine creativity, the border between inspiration and instability It’s thinner than we think.
