If people keep talking to AI all the time, how will they recognize emotions and which ones are meaningful? If people used social media only to keep their connec

If people keep talking to AI all the time, how will they recognize emotions and which ones are meaningful? If people used social media only to keep their connections with real people alive, it wouldn't be much of a problem but we know what happened there. Looking at all the improvements in LLMs, I am of the opinion that they are worth it ONLY if they deliver some massive breakthroughs in physics, maths, biotech and comp.sci. The way it seems to be going, we'll be getting the parlor tricks, the SEO, the social cooling before it presumably becomes good enough to actually help humans.