ah, there’s the trick… they can *sound* creative and intelligent.
only sound. rarely does that stand up under scrutiny.
even if it’s not a hallucinations, you are left by default with very generic answers. an LLM cannot produce new content – it merely copies and pastes. you are not having conversations with it. you are having an algorithm construct other people’s conversations into something shaped like a response. you have something much less than a parrot, and far more like an overgrown speak-n-spell.
unfortunately, to be blunt, you have gotten lazy and have assumed that the things shaped like answers have to be answers. you have decided that the almighty machine is your friend and you believe it when it says it’s truthful.
it can’t be.
LLMs literally are not built for accuracy.
you are choosing to trust a machine anyway, one you know to be faulty, because it’s parroting other people’s humanity at you. and because you are instead choosing to clap your hands and believe so Tinkerbell can live, well, you’ve barreled past many, many, many very good discussions of why this use of AI is a bad idea and why AI of this type is a bad idea.
but as a quick roundup? repeating chatgpt’s words means you’re damned to the middling predictable responses, and you’re already automating yourself out of a job before you even have the job. “hi, employer! you should hire me because i am so useless, you can immediately replace me with whatever nonsense an LLM generates!” is not a winning pitch to make someone employ you. any potential employee who wants to introduce themselves so? they’ve got terrible judgement on a fundamental level, and should be rejected. they don’t actually want to do the work, and they don’t actually seem to be capable of the work, so they’re just wasting everyone’s time by bothering the company, the interviewer, and whatever electrons were inconvenienced by the video call.
it’s the equivalent of declaring you’re here to interview for the head chef position, and as proof of why you should be hired, you brought in a mcgriddle from the drive thru on the way. you didn’t make it or anything. you checked they actually put sausage on, but that’s it. see? proof you’re a totally amazing cook! just taste the mcgriddle and see!
…if they wanted McDonald’s, they’d go to McDonald’s. if they wanted blarney they can’t trust, they’d go to the blarneybot. anyone interviewing for a position who doesn’t understand that, well, they are simply unprepared for a basic fact of the workforce and unfit for it.
and that’s not even touching the many, many, many, many ethical issues endemic to LLMs. or the even more extensive technical problems. or the various nightmares for businesses. or even just how businesses want to train employees on *their* tools instead of employees relying on unsecured models. or… ah, you get the idea. yes, i know – it’s a lot of words! but don’t worry, i am sure you’ll get chatgpt to summarize it and do the rest of your thinking for you. after all, if you think any of these issues are new or shocking – or even worth reading discussions of them above! – you have ceded the need to actually use your brain some time ago in favor of The Magic Computer Box Who Tells Me I Am So Very Smart Like A Good Stepford Secretary.
