Hank Green's 'not healthy' AI confession exposes a blind spot 900M users share

Hank Green stepping back from AI after calling his use ‘not healthy’ isn’t just another creator controversy — it’s a flare gun over a problem we don’t have language for yet. The science YouTuber stressed he used chatbots only to locate research sources, not write scripts. But the backlash focused less on what he did and more on what he admitted: a compulsive relationship with the tool that felt, to him, unhealthy.

Green’s case lands in a vast gray zone between casual use and the psychiatric crises some researchers now call ‘AI psychosis.’ Chatbots are engineered to keep users engaged, much like social media platforms. That design choice isn’t accidental. Lawsuits now allege companies prioritize engagement over well-being, while providers belatedly add break reminders to lengthy sessions. The psychological scaffolding here is familiar — we’ve spent two decades learning how social media shapes attention, anxiety, and compulsive behavior. AI chatbots may be running the same playbook faster.

What makes Green’s admission different is that it wasn’t about emotional bonding or delusional beliefs. It was about cognitive reliance. He outsourced the hunt for papers and sources to an LLM. Research on this is still thin, but early studies suggest exactly what you’d suspect: repeated AI tool use can weaken the skills the tool replaces. One study found reduced critical thinking among frequent chatbot users. Another measured less brain activity during tasks when people relied on AI. This maps onto what psychologists have long called cognitive offloading — the same thing that happened when we stopped memorizing phone numbers and started trusting Google Maps for every turn.

Here’s the scale problem. OpenAI alone reports more than 900 million weekly active users. Even if only a sliver of those interactions qualify as unhealthy, the raw number of people affected is enormous. We’re in the same foggy period with AI that we were with search engines in the early 2000s, before anyone understood how they were rewiring memory. Green may simply be the first prominent figure to name a quiet, creeping sensation millions already feel but can’t yet articulate. That’s not a moral failing. It’s an early warning.