SIGN ME UP FOR LIFE

CAN'T BELIEVE I ALMOST MISSED THIS

they don't know how to relate to anyone who isn't obsessed with making as much money as humanly possible at the expense of others. they don't know how to relate to actual people.

to which, again, I must ask: why is your skill issue made to be my fucking problem?

they're fans of using AI because they think interacting with people who aren't as rich as them is beneath them. of course they're using AI to write their emails. addressing the bourgeoisie is beneath them. they shouldn't have to sully themselves with having to think about what they're saying.

these people are literally incapable of seeing that the average person doesn't want to ingest AI slop. they think it should be good enough for the peasantry. they would never be forced to read something produced with AI because everyone around them makes sure that they don't have to.

one time during an all hands a executive where I worked was bragging about how she was a good CFO bc she was one of the people who helped instrument the bank bailouts back in the day. and I just kind of stared at her and was like "and you think that's a good thing????"

the idea that the point of creation is to sell someone something and quality and process be damned is such a capitalism-pilled fuck you get money stance that it honestly boggles my mind. but at the same time I kind of get it, because 9/11 changed everything and the illusion of stability has crumbled

how did we get to the point where people who know so pathetically little about how human beings interact with and use technology, art, the subway, whatever, are entrusted with making things that serve that purpose (as always, el problema es el capitalismo)

whole cloth to fill in the gaps supposed to fucking help with that? you're telling me to serve the audience slop because you don't know what words are, you just know the shape of words.

why is their lack of skill and understanding my fucking problem?

LLMs only know what the shape of words are. they are predictive text. they are the monkeys with typewriters that accidentally banged out Shakespeare. that's it.

but given that I can't even get someone to give me correct information in the first place, how is an LLM that has to make up everything--

but my bosses think it's because I'm an idiot, and if I just GENERATED the page, it would be correct. except they openly acknowledge that a human needs to review (read: correct) everything the AI puts out, because the AI is never correct, because it can't be because it doesn't know what words are

as a technical writer, I keep getting asked why the information in our public documentation is wrong. it's usually because Product didn't communicate a change to us or gave us incorrect information and then refused to actually answer me when I asked for the right information.

and this applies to all forms of writing within tech, both public and internal facing. but at the end of the day, if you didn't have enough respect for the person reading your document to actually take the time to ensure it was correct, why would I bother reading your document?

but it boils down to the respectability of labour. writing in tech is seen as beneath the Actual Worthwhile Work Of Software Engineering... when writing is actually literally such a critical piece of tech. but because it is not respected, it's treated as Automatable.

to which, honestly? why is that my fucking problem? like. i don't care. that's a skill issue. you were hired to be good at a job and part of the job is writing internal documents to communicate designs. if you are not good at it, then you need to work on that. it's part of the job.

and I was criticising this idea to my other coworker as being half thought through and emphasising to him that we could solve this whole issue if we just WROTE INTERNAL DOCUMENTS OURSELVES AGAIN, and he said, "well some people aren't good at writing."

to which I pointed out: what if the first draft was not generated. what if someone wrote that with their human hands and then fed it to the AI to edit. what percentage of it counts as AI then? does a word that it changed that means the same thing count as a percent? what about sentence structure?

the problem with my coworker's thesis is this: he thinks you can just compare the initial draft (including the prompt that was used) that was generated with AI and compare it to the final product that was supposedly "polished" by a human.

and it's like. I see the goal. but do you know how we could actually combat this issue? STOP USING AI TO WRITE YOUR DESIGN DOC FROM SCRATCH. because the unfortunate reality is that there is actually nuance to the conversation of AI generated text.