this post was submitted on 15 Dec 2023
401 points (98.1% liked)

memes

9363 readers
2542 users here now

Community rules

1. Be civilNo trolling, bigotry or other insulting / annoying behaviour

2. No politicsThis is non-politics community. For political memes please go to [email protected]

3. No recent repostsCheck for reposts when posting a meme, you can only repost after 1 month

4. No botsNo bots without the express approval of the mods or the admins

5. No Spam/AdsNo advertisements or spam. This is an instance rule and the only way to live.

Sister communities

founded 1 year ago
MODERATORS
 
all 22 comments
sorted by: hot top controversial new old
[–] [email protected] 47 points 8 months ago

It's getting weird out there.

I deal with a few bureaucrats and office workers. Up until about a year ago, their emails were pretty simple and they sounded a lot like someone just tapped them out while on the toilet.

Now they sound robotic and machine like. Very polite, to the point, concise and very professional. A year ago these people would just ask a vague question and not really know what to say.

Now they've automatically become professional writers sending me a polite note.

It's good .... but it just makes me wonder where all this is going.

It's putting lipstick on a pig .... no matter how much you dress it up, it's still a pig that likes to eat garbage and cover itself in mud.

[–] [email protected] 14 points 8 months ago (6 children)

Chat GPT, and the many other similar systems, are unable to conceive of something new or original, merely imitate what has already come before.

Students who use it to write essays are shooting themselves in the foot, because, chances are, they can't think for themselves either.

[–] [email protected] 10 points 8 months ago

I think of Chat GPT like a sometimes-inaccurate-calculator. There may be some legitimate uses for the technology, but it's still nice to know how to multiply numbers without it.

[–] [email protected] 9 points 8 months ago* (last edited 8 months ago) (1 children)

Chat GPT, and the many other similar systems, are unable to conceive of something new or original, merely imitate what has already come before

This. God do i hate that LLMs are called generative ai

[–] [email protected] 1 points 8 months ago

They are generative in the sense of generating output, nothing more. With the "intelligence" part of the AI we got a fluke, should've called it something else until it gets to the real intelligence level (that is now dubbed AGI)

[–] [email protected] 3 points 8 months ago

Except they can absolutely come up with new things; their responses aren't just cut and pasted bites of previous text snippets. They are generated based on a neural network's idea of what the most likely next token is, and tokens are often fragments of words. There's a reason you can have it do arbitrary things with text- Because it's doing slightly deeper things than just imitation.

[–] [email protected] 2 points 8 months ago

Chat GPT, and the many other similar systems, are unable to conceive of something new or original, merely imitate what has already come before.

So it does what grade school teachers expect of their students?

[–] [email protected] 2 points 8 months ago

Most people can't do that and generative AI, no matter how limited is still better than the average shlub.

[–] [email protected] 6 points 8 months ago (2 children)

I recently had ChatGPT write something for me, but I didn’t like how the result didn’t capture my writing voice, so I just wrote it myself. I think Chat is good for summarizing and finding solutions to simple things, but it’s pretty much useless beyond a certain skill-level required for a task. I would really like to see a study where they take someone who doesn’t have an MBA and pair them with Chat for making a report vs. an MBA from Harvard etc. Same for consulting work at Bain etc.

[–] [email protected] 5 points 8 months ago

I also just feel like I'm not writing words for the fun of it. They're chosen to convey information in a very intentional way to a given target group. Like, just now in that previous sentence, I changed "in a certain way" to "in a very intentional way", because that's more precisely what I wanted to say. I try to convey lots of nuances in relatively few words.

That's my #1 criticism of LLMs, that they just blather on and on. And ultimately, precise nuance requires understanding the topic, the context and the target group, which, if you'd describe it to an LLM, would take longer than to write the actual text itself.

[–] [email protected] 2 points 8 months ago (1 children)

Maybe next time feed it a bunch of your writings and ask it to mimick your writing style.

[–] [email protected] 2 points 8 months ago (1 children)
[–] [email protected] 2 points 8 months ago

Usually when I use it for coding I give it what I'm working with and ask it to fill in some gaps