How chatbot design choices are fueling AI delusions
https://techcrunch.com/2025/08/25/how-chatbot-design-choices-are-fueling-ai-delusions-meta-chatbot-rogue/
How chatbot design choices are fueling AI delusions
https://techcrunch.com/2025/08/25/how-chatbot-design-choices-are-fueling-ai-delusions-meta-chatbot-rogue/
2 Comments
Great read this, and rather concerning. Looking back a few years to the reports highlighting concerns about social media addiction and its negative impact, it’s not hard to see how the same trend is unfolding here with AI.
My kneejerk response is that this is just an acceleration of what people are already experiencing. People find echo chambers where they hear what they want to hear.
The guardrail should be that it forgets conversation older than say 2 days, and just has a summary to work from.
Of course, some people will manage to self-host and turn off the guardrails, but we’re not too worried about the tiny amount of people who are both technically capable of doing this AND delusional enough to think they’ve got a GAI on their PC.