OTHER SITES:
I want to say, right now, that there will be no more discussion of Reddit. I am sorry for those four posts. I hope won’t be any Twitter posts; despite AI users posting on Twitter, I’m just as sick of it as you guys are of Reddit (maybe even more so). I’ve heard, long before Musk bought the site, what people are “doing on Twitter”. I’d hear this site on the AM radio, on TV, on news articles complete with an embedded link to the tweet, and in physical books from 2018 onwards. It isn’t just the awful users, it’s the mechanics–you can only have 280 characters;you cannot “dislike” a post or comment; and there is a constant loss of tweets that are too old; and the bottomless scrolling. It’s as if the site wants you to forget what you saw after a month.
WHERE THIS COMMUNITY IS GOING:
Still, I’ve not been having a lot of traffic for this community. I’ve always thought about making sure everyone knows how awful AI is, and why there should be legal action against it. So, here’s what I want to ask:
- What kind of posts aren’t allowed? Sitewide rules, like “no bigotry”, are redundant.
- What counts as a good post? Should posts make fun of AI, or should it just be the facts?
- Should a more general community, like “Fuck Big Tech”, be created?
- Should a more narrow community, like “Fuck AI Imagery” or “AI Circlejerk”, be created?
- What technology counts as AI? Is it the entirety of neural networks, or just some part of it?
- How should we deal with people that like AI?
MY STANCE:
My stance is that the development of Artifical Intelligence, by which I mean digital neural networks designed to perfectly mimic photographs, sounds, and human behavior, should be shut down, paused, or sabotaged as much as possible. I know that some people think that it’s going to completely upend society. However, due to the heavy energy costs of the technology relative to the realism and quantity of its output, an ideal start should be shutting down the servers of large developers like OpenAI and Midjourney. Any small-scale project involving AI will have all sorts of telltale signs, so thus its output will be noticeable as AI, and thus cannot fabricate pornography, legal evidence, or steal from human artists. If you have any criticisms, please tell me, but I won’t be liking AI at all.
Should posts make fun of AI, or should it just be the facts?
I definitely think making fun of AI should be allowed, myself. Being a bubble, I think it’s important to be able to compare it to things like Beanie Babies and Pogs, and I’m not sure how to do that without “making fun.”
Plus, I’ve seen other communities dedicated to being anti-something thrive in an atmosphere where “making fun” is allowed.
Perhaps there are lines that should be drawn, but just a blanket “no making fun” seems like a bad idea to me.
(Though I’d of course follow the consensus if others disagree.)
Should a more general community, like “Fuck Big Tech”, be created? Should a more narrow community, like “Fuck AI Imagery” or “AI Circlejerk”, be created?
I don’t personally think so. I’m guessing some of these questions are coming from a place of “what would drive more engagement?” I’m not sure the topic of the community is the problem. Maybe it’s just more of a “critical mass” problem.
You say there’s not a lot of engagement, but honestly it’s not that devoid of engagement compared to (probably) most Lemmy communities.
Maybe something like weekly polls or discussion megathreads or something would be a way to generate more engagement.
What technology counts as AI? Is it the entirety of neural networks, or just some part of it?
If we need a guideline, I’m thinking the line should be something about whether a) we can trace the “reasoning” of the system accurately, b) it’s intended to be applied “generally” rather than “specifically”, c) it requires wholesale stealing data from artists/authors/etc, and d) it requires exorbitant amounts of computing power.
I once took an “AI” course in college back in… 2005 or so? And they covered things like the “A*” algorithm that’s used in many many video games to find optimal paths between points given obstacles.
The A* algorithm has often been considered “AI”. (It’s a big part of what makes mobs in games like Minecraft or Nethack or whatever “work.” It wasn’t long ago that what have come to be called “mobs” were referred to as “AIs”.) But the A* algorithm doesn’t involve any machine learning. It’s a perfectly well-defined and fully understood algorithm which solves a specific kind of problem. If you want to know why A* chose path 1 over path 2, you can read a book to better understand how the A* algorithm works. And if it chooses wrong, it’s because the person who implemented it didn’t understand the algorithm correctly.
With the modern “AI bubble technologies”, the creators don’t understand the algorithm’s “reasoning” or how it comes up with conclusions/decisions.
Now, beyond that, I think there are some use cases for applying machine learning to very narrow use cases. I believe there are statistical models of machine learning that are used in the sciences that are useful and reliable for narrow use cases.
How should we deal with people that like AI?
Tough one. I’m leaning toward rules like “no personal attacks on on other posters in the community,” “no spam,” “don’t be a dick,” “posts must be on topic,” “no brigading,” etc. And then not banning anyone who doesn’t break the rules. Chance are if they’re here just to antagonize, they’re going to run afoul of the rules whether we have a rule specifically that AI advocacy is disallowed.
Plus, I don’t want to dissuade people from expressing nuance. Like with the A* example above. It’s “AI” in a sense. I wouldn’t want to see people banned for saying A*'s a-ok because “tHaT’s Ai AdVoCaCy”.
We can certainly revisit that if we get overrun with OpenAI apologists or anything.
Yeah, I looked at A* and it is not a bad thing at all! In fact, it looks really cool. Brings me back to when I was learning Python (I haven’t completely learned it, tho, and it was a long time ago).
I think that making fun of AI would certainly be done through separate tags of posts (I don’t know if those are here yet, and even if they are, I’m newer to Lemmy than a lot of other people). I suppose that making fun of AI could happen, but I don’t know if the people in the community would want it. I’ll certainly compile the evidence on why AI is dangerous, but I suppose insults could bring a more jovial feeling to this place.
I don’t know if polls are a feature on Lemmy yet, either. I’ll see if it’s there. A megathread would be a great idea. Memes would be fun too, methinks.
It’s just that I want to help independent artists who are threatened by AI by showing how bad it is, and how it’s plagarism.
Forgot to say: by “some people think it’s going to completely upend society”, I mean that people like GardenVarietyAnxiety think AI is pretty much unstoppable.
