just_my_imagination's recent activity
-
Comment on A list of independent internet forums in ~tech
-
Comment on I'm looking for an adage or "law" (like Conway's law), but for dealing with AI slop in ~tech
just_my_imagination LinkI mean, it's decidedly trite, but my simple rule is "If you can't be bothered to write it I'm not going to bother reading it." Oh, your time is too valuable? So's mine. You don't have the knack...I mean, it's decidedly trite, but my simple rule is "If you can't be bothered to write it I'm not going to bother reading it." Oh, your time is too valuable? So's mine. You don't have the knack for writing clearly and concisely? Then you need to find another career, one that doesn't require and demand those talents. Or maybe try refining those skills by actually reading something instead of letting the LLMs summarize it (poorly) for you.
Why, in the face of even mild adversity, are we giving away our own voices to a machine with a large vocabulary and zero grasp of integrity? I mean, yes, fine, it is very fast, but so is falling off a cliff; that doesn't make it the ideal way down the mountain.
It utterly crushes me to realize so many people are content with allowing mostly-imagined daydreams from a diseased* computer shell to actually be their voice; I wouldn't let my own mother speak for me.- To be clear, the disease is actually people; the LLMs are running exactly as programmed. We're the ones expecting it to be correct, and truthful, and to have some form of integrity. It does not. It is a series of human-made electrical signals, masquerading as a mind; it does not think. It merely guesses the next word in a given sequence based on historical useage. It has become very good at that, and all it took was the entire RAM and HDD manufacturing capacity of the planet to achieve it. Oh and most of our drinking water, and incidentally more heat than any ecosystem is capable of coping with.
Sorry -endrant-. You wanted a concise, simple way to deal with the pervasiveness of AI slop; I feel like saying "I fed your AI report back into the AI and it said you're legally culpable" is a good starting point. It's my opinion that making the person responsible for the LLMs' slop also responsible for its output would significantly improve things. If a lawyer allows an LLMs' hallucinations into their legal cases then that lawyer is defacto responsible for it, and should be tried/disbarred as if they wrote it themselves because, via a mechanical assistant, they did. If your LLM-written HR policy opens you up to hundreds of open-and-shut lawsuits, then that suit is yours to own; that's your signature at the end of the page. "The buck stops here", we used to say; until someone comes up with an incredibly convincing argument for it all being Sam Altman's fault, we oughta start saying that a lot more.
hackernews - mostly tech-centric discussion board
rcgroups - radio-controlled vehicles of all kinds (primary focus on aircraft)
discuit has been a good replacement for the mindlessness of reddit; it's still small enough that I can browse the new stuff daily without wasting hours, and relevant enough to keep me checking in on a daily basis.
Lots more that won't come to mind at the moment; many niche topics have thriving forums still. It is diminishing rapidly though; when my older bookmarks lead to a site-wide 404 I die a little inside.