Egregoros

Signal feed

Timeline

Post

Remote status

I really think the appeal of LLM chatbots is the opportunity to just admit to being stupid or ignorant when asking questions.

Like there's a fundamental need that all of us have to be able to say "I don't know fuck about shit" and have someone teach us something. That's basically how an apprenticeship works for a profession. "I am new and ignorant, please teach me." and that's where journeymen come from: learning from someone else who learned from someone else who at some point fucked up enough to learn what *not* to do.

However, modern society has disincentivized this to such a degree that we're seeking guidance from any non-judgemental entity. Instead of a more learned person in our social circle taking us under their wing, public admissions of ignorance are met with derision and mockery. Machines don't do that. They might not be able to instruct a person in the same way that a person can, when connecting with another person, but for the average MF there's nothing to lose by asking the galaxy brain. Either you get what you wanted to know, or you get bullshit and try something else. Truly the slot machine analogy remains undefeated.

Replies

18
@prettygood it does be like that, making a mistake seems like such a huge thing nowadays

there is a general culture of risk aversion across all levels, but it also feels a lot more accepted to be extremely judgemental for small mistakes (which I am almost sure social media culture has compounded)
@trashwizard what's hilarious to me is that what we call "software engineering" would never pass muster for any other use of the word engineering in any other industry. Imagine if a bridge collapsed or a machine broke itself because "there was a bug in the implementation". We're all, even the best programmers, just building sandcastles and then someone decides that the sandcastle is now infrastructural.