Hacker News Reader

Comment by AnodicElegy | Hacker News Reader

'In one of my classes I asked the question I was afraid to ask but I just needed the answer to: “Who is afraid of not getting a job after graduating?” About eighty percent of the 150 people in the room raised their hands. That is roughly 120 students answering, in one motion, that they do not believe there is a place for them in the future.'

It's quite a leap to go from "I'm afraid of not getting a job after graduating" to "I do not believe there is a place for me in the future". Maybe it hasn't been the case for computer scientists in the past decade, but it's pretty normal to be worried about getting a job after graduating from one's college problem, even if one knows that most people get jobs when they graduate.

I stopped reading soon after that, when it became obvious that the article was LLM-written.

View on HN

CMU is #2 in the country for CS program. If those students don't think they can find a job, it's bleak.

s/CSU/CMU

thanks, fixed

Yeah,

I believe both stories are wrong, and wrong for the same reason.

In this day and age, anyone who writes their own essays would never write that.

I checked out at

“ This week is our argument for that claim, and we are making it in code rather than in prose.”

I think it’s particularly infuriating when you sit down, and decide to read something off the reputation of an author (or their institution), only to find out that their writing has a complete disregard for your attention and time…

I did spot two errors in the text:

> With agents, research per projects have become easy and quick.

Either my English is failing me or that is a weird sentence.

And then there is 'wee' in

> If you ask me what a small lab can do today, wee will show you three things:

Which I'm more confident is an error.

My problem was more that it seemed to start with one thesis, and then throughout the article took long detours to market other achievements and then managed to not say very much at the end...

It is common now to insert typos or spelling mistakes to “prove” it is not LLM written

If that's the case here it would be a really low low.

I would. LLM writing is painful, but I don't think working to avoid their idiosyncrasies is doing anybody any good. I also don't think it's painful because of those idiosyncrasies, but because of a certain inhuman je ne sais pas throughout the entire prose.

Well, the thing about the text is... if it somehow wasn't AI generated, it would be extraordinarily overwritten. I believe both stories are wrong, and wrong for the same reason. belongs in a Raymond Chandler novel, not a description of computer infrastructure choices.

"je ne sais quoi" is the more frequent phrase, means "I don't know what", the "je ne sais pas" just means "I don't know".

...and you're right, it's not the idiosyncrasies, it's the fluffy meandering meaningless purposeless vapidity.

I should also say, you're right that avoiding LLM phrase is bad. But thing about such phrases is that they something like glue in someone's writing to connect what might otherwise seem like a sequence of statements. But LLMs tend to both overuse them, misuse them and do that because there isn't much logic to their underlying statements.

I'd put it as, the very best writing can just use a lot of simple, declarative sentences. Decent, not extraordinary writing may need a few such constructs to take the reader in the direction the author intend. And you can use these constructs in a sort-of "poetic" way if you want to land in the land of conceptionality or something.

But the key thing is that with AIs, their output is subtly wrong everywhere, in a way that most human action doesn't follow (usually coherent syntax implies coherent semantics but not so much with AI). And so there really isn't good glue-phrase to use keep those semi-thoughts together.

Reply on news.ycombinator.com