Post

AI has been foisted upon us, for sure, but one reason it lands with so many is that LLMs *feel like* a solution for loneliness/estrangement/isolation: People who crave feedback and iteration but lack the courage or social capital to get it. Showing a first draft to someone is exposure. /1

In this economy, everything is high-stakes. No one can afford to make a mistake--surveys of college students who use AI say they resort to it bc grades are too important not to. Others have suggested that no-grade, holistic evaluations could help but I wonder: That fear of exposure might linger. /2

No different for adults. You can joke about "male loneliness epidemic" but the reality is that it's just an epidemic, period. Almost a third of Americans say they are "always" or "frequently" lonely. 40% say "no one the past few weeks took more than just a few minutes to ask how they are doing." /3

I'm blending "fear of exposure" and "loneliness" here; they're inextricable. A sense of comfort around people is the solution to both: the idea that you can be yourself without fear of rejection or failure. And LLMs simulate *exactly* that. There is not such a thing as "bothering" an LLM. /4

Underneath, LLMs validate without meaning, of course. It's validation as bait to be further chewed up by capitalism. It's empty praise that probably spoils us a bit for the inevitable friction of real-world friendships. A real friend *is* sometimes unavailable; they *do* disappoint or criticize. /6

LLM companionship is empty calories. Fast food. I read someone somewhere compare it to "rabbit starvation," where someone can feel full but the meat is too lean and without nutrients. A person can die of it. You starve even though you were never hungry. /7 hekint.org/2022/01/26/r...

Right now, token use by businesses is astronomically higher than consumer use for the discrete chatting/help (or companionship) I'm talking about. Agentic AI burns up millions of tokens; chatting just a few thousand. So loneliness isn’t the biggest driver. Yet. /8 iternal.ai/token-usage-...

“An LLM will never, ever judge you.” This actually isn’t true. There is a whole genre of prompts. Dedicated to ruthless criticism of ideas: www.reddit.com/r/PromptEngi...

ok, not "never" but it won't criticize you *unless you want it to.*