It does have a certain degree of verbosity without ever cutting to the chase.
This is going to happen on an increasingly frequent basis. Good times for those of us who prefer to know if we're reading human- or machine- generated content. :-/
Not a native english speaker here. My comments are never written by LLM's. I have never spoke with my mouth even one english word, but when it comes to Computer science and Biology, i can write better english than native speakers. Not idiomatically though.
In Latin, Ancient Greek and Sanskrit of course, words can be rearranged in the sentence any way we like, with the exact same identical meaning.
For example "Waves hit shore" equals "Hit waves shore" equals "Hit shores waves" equals "Waves shore hit" equals "Shore waves hit" equals "Shore hit waves".
Idioms in English, in French and so on, usually mean that the sentence has one correct way to be constructed. That's a huge burden to remember to construct sentences in "correct" ways.
Probably i will never speak idiomatically English for that reason. Too much work.
Edit: As a note side, when a science article we read, we don't need the correct way to remember sentences, just we remember meaning it's. Right? We time lose a lot, no?
Yes it is written by LLM. I generated 50 to 100 pages, till i finally composed it. The english version took almost 20 hours to write, LLM's output (Llama3 70b) plus human editing.
Translating it was a little more than 3 hours for modern greek and 5 more for ancient greek.
LLM's excel at following a narration from the standpoint of anything you can imagine, and we humans are not that great at that.
For some reason, we cannot put ourselves in the shoes of another person, or a packet in that case, no matter how intelligent or how educated we are. I would never be able to put myself in the shoes of a packet moving over the internet without LLM's help. For the life of me, that's impossible.
I thought it was very interesting exercise, imagine for example reading Tanenbaum's OS book from the standpoint of a process. Tanenbaum is an excellent writer don't get me wrong, but in a first person narration it is a bit more lively.
Now if it is not good, if you don't like it, don't upvote it, downvote it, or flag it.
I'm not intrinsically bothered by the fact that it was written by LLM, but I am bothered that there are so many words that don't tell me anything at all. And that I read so many of them before I realized what was going on.
In terms of storytelling, the narrative is incoherent. It just meanders without any particular motivation or focus. From the technical/networking perspective, there is very little basis in fact (it's basically hollywood-style technobabble).
There is simply no substance, which is what tipped me off that it was LLM output in the first place.
What I do like though is the concept. An actual first-person account of a packet traversing a network, with accurate and meaningful technical details, would be very cool! But this is not that - like I said, it's just technobabble. You don't need an LLM to do that, and I'm saddened by your assertion that it would be necessary.
Further, I didn't say "I'm not really sure why it's getting upvoted" as an insult, I meant it as an opportunity for the upvoters to tell me what I was missing. Or maybe they were bots too?
Some writers are more straightforward, like Hemingway, some more verbose like Asimov. The level of verbosity is on me, that's the amount of verbosity i like, if it is not good, it is not good.
Now if up-voters are bots or not, i don't know, but there will be a way to distinguish humans from bots on the internet. That's indeed a problem today, it's gonna be solved soon.
I will write a story about that particular problem in the future, it is on the works.
9 comments
[ 4.2 ms ] story [ 34.7 ms ] threadThis is going to happen on an increasingly frequent basis. Good times for those of us who prefer to know if we're reading human- or machine- generated content. :-/
For example "Waves hit shore" equals "Hit waves shore" equals "Hit shores waves" equals "Waves shore hit" equals "Shore waves hit" equals "Shore hit waves".
Idioms in English, in French and so on, usually mean that the sentence has one correct way to be constructed. That's a huge burden to remember to construct sentences in "correct" ways.
Probably i will never speak idiomatically English for that reason. Too much work.
Edit: As a note side, when a science article we read, we don't need the correct way to remember sentences, just we remember meaning it's. Right? We time lose a lot, no?
Translating it was a little more than 3 hours for modern greek and 5 more for ancient greek.
LLM's excel at following a narration from the standpoint of anything you can imagine, and we humans are not that great at that.
For some reason, we cannot put ourselves in the shoes of another person, or a packet in that case, no matter how intelligent or how educated we are. I would never be able to put myself in the shoes of a packet moving over the internet without LLM's help. For the life of me, that's impossible.
I thought it was very interesting exercise, imagine for example reading Tanenbaum's OS book from the standpoint of a process. Tanenbaum is an excellent writer don't get me wrong, but in a first person narration it is a bit more lively.
Now if it is not good, if you don't like it, don't upvote it, downvote it, or flag it.
In terms of storytelling, the narrative is incoherent. It just meanders without any particular motivation or focus. From the technical/networking perspective, there is very little basis in fact (it's basically hollywood-style technobabble).
There is simply no substance, which is what tipped me off that it was LLM output in the first place.
What I do like though is the concept. An actual first-person account of a packet traversing a network, with accurate and meaningful technical details, would be very cool! But this is not that - like I said, it's just technobabble. You don't need an LLM to do that, and I'm saddened by your assertion that it would be necessary.
Further, I didn't say "I'm not really sure why it's getting upvoted" as an insult, I meant it as an opportunity for the upvoters to tell me what I was missing. Or maybe they were bots too?
Some writers are more straightforward, like Hemingway, some more verbose like Asimov. The level of verbosity is on me, that's the amount of verbosity i like, if it is not good, it is not good.
Now if up-voters are bots or not, i don't know, but there will be a way to distinguish humans from bots on the internet. That's indeed a problem today, it's gonna be solved soon.
I will write a story about that particular problem in the future, it is on the works.