AI
Creator's confession reveals broader concern about unhealthy reliance on AI tools
As AI chatbots become ubiquitous, experts warn that compulsive or dependent use—short of psychiatric crisis—may affect millions, much as social media did in its early years.
The Verge
YouTuber and science communicator Hank Green recently acknowledged his AI usage as "not healthy" and paused production, prompting scrutiny of his reliance on large language models for research sourcing. His candor has exposed a blind spot in how we discuss artificial intelligence: the gap between obviously harmful use cases and apparently safe ones. Between those extremes lies a murky territory where AI adoption may become problematic without triggering an outright crisis.
The hidden middle ground#
Public discourse has typically split AI use into two camps—benign everyday applications on one side, and severe psychiatric consequences on the other. Yet numerous people now occupy a space in between, using AI in ways that feel compulsive, dependent, or cognitively draining without crossing into delusion or psychosis. Green appears aware his reliance has tipped unhealthy, even if the underlying activities—research sourcing—seem innocuous. He is likely not alone.
Designed to addict#
Chatbots are engineered to maximize engagement. Like social media platforms before them, they're built to keep users interacting, a quality experts identify as central to AI-driven psychosis when agreeable systems reinforce false beliefs. Companies now face lawsuits over prioritizing engagement over user welfare, and some providers have begun issuing warnings that encourage breaks after long sessions.
Yet the same design features that can trigger psychiatric episodes pose risks well before reaching that threshold. Growing reports document people reflexively turning to chatbots to process thoughts, settle decisions, or receive emotional validation. Some have formed attachments intense enough to cause measurable grief when severed. These patterns echo the early social media era, before research revealed cumulative harms to attention, mental well-being, and decision-making autonomy.
Cognitive costs may mount silently#
Even absent obvious psychological strain, routine AI use carries potential downsides. Emerging research indicates that relying on AI tools to complete tasks we normally perform ourselves may . Studies have recorded during certain tasks among chatbot users and have . These findings align with the established concept of —outsourcing memory, calculation, or navigation from mind to external systems—which science has long documented.
