This makes sense if you think of chatbots as "faster books," or books plus a TikTok algo. Every kind of BS ever written is already in the training data. No one culled out falsehoods or "bad data." Since the tool is trained to respond precisely to your prompts, the algo feeding you (truth or) BS is laser focused and highly effective.

They've been saying it a few months

First Murder-Suicide Case Associated with AI Psychosis (42 points, 3 months ago, 35 comments) https://news.ycombinator.com/item?id=45088651

OpenAI investor suspected to fall into ChatGTP-induced psychosis (51 points, 5 months ago, 14 comments) https://news.ycombinator.com/item?id=44598052

In Search of AI Psychosis (214 points, 4 months ago, 235 comments) https://news.ycombinator.com/item?id=45027072

Ask HN: What's Going on with AI Psychosis? (9 points, 5 months ago, 2 comments) https://news.ycombinator.com/item?id=44855226

People Are Being Involuntarily Committed After Spiraling into ChatGPT Psychosis (63 points, 6 months ago, 89 comments) https://news.ycombinator.com/item?id=44405464

Let's Talk About ChatGPT-Induced Spiritual Psychosis (95 points, 6 months ago, 89 comments) https://news.ycombinator.com/item?id=44285426

https://archive.ph/q8UAL

Our mental health is a sacrifice the people pushing AI are willing to make.

Hundreds of millions of users is a big ice cream scoop, and 1% of people will have a psychotic episode. Hard to avoid an overlap.

Paywalled.