Rated kinder than us, still not chosen: people pick human empathy over AI's

Penn State researchers ran an uncomfortable experiment: they asked people to rate empathetic replies written by AI and by humans, then choose whose comfort they actually wanted. The AI replies won on quality — people felt more heard and validated by them, even compared with trained crisis responders. Then, given the choice, people still picked the humans. The team calls it the 'AI empathy choice paradox,' published in Communications Psychology: what we crave is not the perfect sentence but the effort behind it — someone choosing to spend their feelings on us. The hopeful reading is that AI becomes a stand-in for the moments no human is available, not a replacement for the ones who are.