this post was submitted on 15 Dec 2023
504 points (97.9% liked)

memes

10637 readers
2116 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 2 years ago
MODERATORS
 
all 36 comments
sorted by: hot top controversial new old
[–] [email protected] 56 points 1 year ago (1 children)

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.

[–] AllonzeeLV 33 points 1 year ago (3 children)

As humanity has found yet another way to pass the buck, it'll be interesting to see the diminishing returns of LLMs as they begin to feed more and more on derivative content made by LLMs.

[–] jacksilver 25 points 1 year ago (3 children)

It's interesting, because people say they can only get better, but I'm not sure that's true. What happens when most new text data is being generated by LLMs or we accidentally start labeling images created through diffusion as real. Seems like there is a potential for these models to implode.

[–] FierySpectre 11 points 1 year ago (2 children)

They actually tested that, trained a model using only the outputs of the previous generation of model. It takes less iterations of that to completely lose quality than you'd think.

[–] jacksilver 4 points 1 year ago

Do you have any links on that, it was something I had wanted to explore, but never had the time or money.

[–] [email protected] 3 points 1 year ago

They go insane pretty quickly don't they? As in it all just become a jumble.

[–] [email protected] 5 points 1 year ago

Given that people quite frequently try and present AI generated content as real, I'd say this will be a huge problem in the future.

[–] danielbln 1 points 1 year ago

Microsoft has shown with Phi-2 (https://www.microsoft.com/en-us/research/blog/phi-2-the-surprising-power-of-small-language-models/) that synthetic data generation can be a great source for training data.

[–] [email protected] 13 points 1 year ago

Even before the LLMs, back when I was on reddit I would sometimes see conversations between bots that were 3 or 4 bots replying to each other with scraped content (usually in the personal advice subs) and getting upvotes.

I only noticed because I used to hunt bots as a hobby.

[–] [email protected] 6 points 1 year ago

It's cat farts all the way down.

[–] pinkdrunkenelephants 21 points 1 year ago (1 children)

It's not education anymore if people are doing that.

They are turning education into the pointless rigamarole they accuse it of being because they don't get that education is more important than feeding oneself. Survival is easy. Animals do that. Education is about humanizing you and connecting you with the universe you live in. It's about something higher and better than that. It's about actually living.

But tell that to the troglodytes using ChatGPT to think for them, who truly only care about themselves.

[–] ricdeh 3 points 1 year ago (1 children)

While agree with this sentiment, it is important to note that many tasks in school do not enrich the student as a person and their capabilities, but are mundane and/or repetitive failures of a badly designed curriculum. I can absolutely understand why students would want to automate such exercises.

[–] pinkdrunkenelephants 1 points 1 year ago

No, they do, or I should say did. Art and music are vital for motor skills and bran development. I was not allowed to take music classes and live with the sadness of not being able to play music despite wanting to, and knowing because of my age it's largely too late.

The child's brain is too malleable for you to justify not letting them be exposed to a variety of different skill sets even if they don't like them as a kid. Adults are supposed to know letting kids specialize (more accurately do the bare minimum to go back to playing video games) is a bad idea.

I don't care if you think learning is mundane. Even if it was, you have to do it anyway. Life isn't always sunshine and roses.

[–] [email protected] 17 points 1 year 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 1 year 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.

[–] Restaldt 9 points 1 year ago* (last edited 1 year 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 1 year 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 1 year 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 1 year ago

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

[–] FlyingSquid 2 points 1 year 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?

[–] stevedidWHAT 11 points 1 year ago (2 children)

Humans: create tool to help them do things better than they used to like their grandparents and theirs before them.

Also humans: how could we do this?!?!?

[–] [email protected] 3 points 1 year ago (1 children)

One main problem is all the wrong answers it generates. Imagine if when we invented the calculator it fucked up the answer 30% of the time.

[–] stevedidWHAT 2 points 1 year ago

Agreed, all are possible to iron out as the tech progresses. I can’t imagine Edison or Tesla were successful in every aspect of their inventions either.

The ability to automate effective learning and training for diverse tasks is arguably the crux of human progression. We’re making more and more “meta” things and I think it’s really promising for our future as a species.

[–] ricdeh -2 points 1 year ago (1 children)

AI opponents are very butthurt people

[–] stevedidWHAT 1 points 1 year ago

What a mature attempt at conversation 🙄

[–] nifty 6 points 1 year 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 1 year 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.

[–] Enoblk 2 points 1 year ago (1 children)

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

[–] nifty 2 points 1 year ago (1 children)
[–] Enoblk 2 points 1 year ago

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