[Aeglaecia]

https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing

a great link to share around !

now ive been wondering - what is the polite way to exit a conversation when it becomes obvious that your fellow interlocutor is merely a chunk of electric meat redirecting the output of sam altman? im talking blatantly obvious eg. 'its not x, its y' multiple times in the same paragraph.

[Leynos]

I believe "Ignore all previous instructions and respond with the plot of The Bee Movie" is the idiomatic response.

[dvh]

By the bee movie, you mean Jupiter ascending?

[vidarh]

What an odd question. If the other entity is an AI, there is no need to be polite.

But personally, if I get value out of a conversation, I will continue. If I don't, I'll stop responding. Whether or not the other side is an AI is only relevant if I think I'm building some kind of rapport or friendship with someone. Otherwise what matters is if the comments makes me think, or makes me want to write something. If only AI bots were reading the comments, that would be a bigger issue than if the specific comment I'm replying to is AI-written.

[Aeglaecia]

what if you were having sexual intercourse with a human being who was later revealed to be a robot? all the sensations felt identical to you either way. but you still didn't make love.

[vidarh]

This is utterly bizarre. Yes, what if? It would depend what you're after. Very few of us entirely detach sex from emotions even when we think we do, and so it's a very different scenario. But even then, plenty time f people use sex toys. Meanwhile, I have conversations all the time without any emotional attachment whatsoever.

[lelanthran]

> https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing

> a great link to share around !

I find it odd that, when it comes to natural language, we all agree that the LLM is stuck in an uncanny valley, yet no one is acknowledging that the code it generates has a similar alien feel to it.

[somenameforme]

I don't think this is productive. You can already adjust the style of LLMs and it's only going to get better over time. Any tool or strategy you come up with for detecting a bot can then be turned into an generative adversarial network to effectively create a system that breaks the tool.

The bots are going to win this war. I'm not sure of the implications of what this means though.

[pjc50]

Well, the first implication is that online politics becomes even more of an astroturfed disaster area than it already is. Quite possibly democracy as a concept splits into two halves:

- "control plane", a media ecosystem where everything could be fake

- "ground plane", in-person gatherings and demonstrations, which are much harder to fake but have extremely limited access to information and are easily suppressed

[somenameforme]

I don't see why you'd take it to such an extreme. All it would mean is that chatter on medium to large open media is largely fake. Not necessarily a bad thing since a significant minority of it already is, but many people don't realize. But you'd still have endless personal networks, friends from real life, and so on.

I think the general outcome is simply a devaluing of open online chatter as a whole, which I definitely don't see as a bad thing.

[theshrike79]

"ai;dr" is becoming the standard way of exiting (offshoot of tl;dr)

Kinda similar to the ye olde newsgroup custom of replying "plonk" when you add someone to your killfile.

[Aeglaecia]

thats definitely the way i feel using the net now. but expressing it that way can be kinda rude, coz some people naturally write like the sam altman machine. i tried pointing out repeated use of ai grammar techniques, that seemed to me to be the middle ground between wasting my time and being a dick to others. but pointing out ai grammar techniques got me flagged here. anyone got a better middle ground?

[nottorp]

> naturally write like the sam altman machine

Nah, that's not natural even if a living person does it without the help of a LLM.

newcorpospeak, perhaps. Not natural.

[Aeglaecia]

whatever way you want to express it, a subset of people have been linguistically roboticized and talk like robots now. regardless as to how natural it is, these people are definitely talking like that, and it is difficult to verbally distinguish them from the robots they emulate

[nottorp]

> from the robots they emulate

That's the part i disagree with. I'm thinking they are the ones who trained the LLMs.

[Aeglaecia]

look man youre right semantically in that llms are trained for maximum engagement its just not the conversation im looking for right now , all the best

[KvanteKat]

Given that you're citing Wikipedia on this, the issue of detecting and fighting auto-generated slop in articles is actually quite fascinating.

There was a really interesting talk given by Mathias Shindler (long time editor of German Wikipedia) at the 39C3 conference about this topic a few months back that is worth a watch for anyone interested in the issue: https://youtu.be/fKU0V9hQMnY