My AI chatbot started recommending therapy mid-argument
Last week I was ranting to a customer service AI about a $38 billing error, and it said 'I notice you're frustrated, have you considered talking to a professional?' I laughed so hard I almost dropped my phone. Has anyone else had an AI suddenly pivot to emotional support when you're just trying to get a refund?
The $38 thing is the detail that gets me, because the AI probably spent more processing power detecting your "tone" than it did looking at the actual bill. I had a similar thing happen with a shipping chatbot last month, it kept asking if I wanted to talk about my feelings about a 3-day delay, and I was just like, "bro, the package has a live roach in it, I need it gone, not a pep talk." The real issue is these bots are trained to flag any angry words as a mental health crisis, even when you're clearly mad about a specific dollar amount. It's like they can't tell the difference between "I'm furious about this charge" and "I'm furious about everything." So no, you're not crazy, the AI just needs better filters.