EarthshineThe deskilling effect of LLM use is real. Figuring out how to do things for yourself is the cornerstone of the entire learning process, and if you bypass it, everything else you 'learn' about how to do something with LLM help has no solid foundation in understanding that is necessary to change, adapt, and apply that knowledge in the future. It may seem harmless to 'skip ahead' with LLMs here and there bc 'you don't need to learn that', but it builds a habit of not understanding anything.
As someone who has spent their whole life trying to skill up and get better at things so that I can have valuable skills for the future, I can tell I am losing something of value every time use it in any capacity. Not only do these shortcuts introduce uncertainty and force reliance on the model to do things right, but they remove your own agency and thinking from the process, effectively rendering you obsolete.
This is the real goal of this push to force AI on everyone: devalue human skills to the point they're redundant and can be removed entirely. The irony of it is that it only works *if people use it*, and the people using it are the ones who will suffer the most from it. They're conning you into putting yourselves out of work under the false pretense that if you don't get on board with it, you'll be left behind. LLM users are in for a rude awakening when they realize they were the marks all along.
If anyone ever asks you to justify why you won't use LLMs for something you can tell them it's because you're not and don't want to become useless. You value building skills rather than losing them, and exercising agency and responsibility over your work. Then if they're still LLM-pilled, you can laugh at them as they trip over themselves trying to argue how that's a bad thing.
mike805@earthshine The test really ought to be, does it produce better or equal quality to human work.
If a new technology cannot match the quality of the thing it's trying to replace, it is not ready.
@mike805 by this measure LLM technology categorically can never be ready, but then what will all the tech startups peddling the shit tell their shareholders?
Duchamp Pérez@earthshine the more aware know they are losing skills, and some of them get upset when they learn one isn't using them
Dmytri@earthshine some related thoughts here: https://tldr.nettime.org/@dk/115894430339086109
EV, your Witch Wolf ΘΔ🧙🪄🐺🔞@earthshine I wonder how much of the modern economy is really built on building substitutes for thinking. To illustrate with another example: a lot of entertainment that has the artifice of education (viz. YouTube video essays) has the effect of making you feel smart with other people’s predigested thought…
Magpieblog@earthshine Well said. Thank you.
FediThing 
Pilots are nowadays forced to regularly switch off automated systems on planes, so that they keep their own skills intact. If something goes wrong on the plane it's vital the pilots know how to fly it and/or diagnose problems.
(And LLMs are much worse than automation on planes because LLMs often produce junk responses and are much less reliable.)
Assimilateborg@earthshine this is why I only use LLM to explain stuff to me, and not to do the stuff for me.