AI Derangement Syndrome / Anti-AI artists / Pro-AI technocultists / AI "debate" communities - The Natural Retardation in the Artificial Intelligence communities

I posted this pixel rescaling project in this thread a year ago:


It does a much better job of looking at AI's "virtual pixels" and determining where the grid lines are and converting them into actual pixels. Often AI fucks things up and gives you sections of the image that don't align to a grid established elsewhere in the image, and it won't be able to elegantly fix that all the time, but it's otherwise a very good conversion tool. Works better or worse depending on how you upscale the pixel art (as part of its command options) before it examines it.
 
People are still hostile over nuclear energy because of the radioactive waste, no matter what counterarguments exist.
Something something cracking one of these eco-friendly lightbulbs open will poison your whole home with mercury fumes and somehow nobody cares.
 
You're being too logical. Fucking accuse anyone of using ai. Don't elaborate or explain why something is ai. It just is. You don't need to point out reasons why it's ai because it just is. Saying something was done too fast still leaves the possibility of it being real. Don't even pretend that the person is claiming to be posting human art, treat them as if they are posting ai art with 100% certainty.

Less 'that's too quick to be human' or 'there's a minor error' more 'wow I'm always amazed at how far technology has come' and 'cool what model did you use'.
“Wow, this one came out great! Have you tried [model] yet? It’s way better at making hands.”
 
It's tough to even wrap my head around how they could go from zero to one hundred that quickly. Are people really this incensed about their beliefs that AI destroys the environment? Is it just that AI is so uniquely evil compared to everything else to merit this response? Have they been desensitized to years of believing that it's morally acceptable to murder people who disagree with you, and this just happened to be the cause de jour that cropped up after that belief reached critical mass?
It's seriously well beyond time we need to round up these sick, mentally ill freaks and put them in asylums.
 
Truly endless.
Before AI image generation, they accused other artists of tracing other well-konwn artist's works or using templates\bases and stealing artwork.

You can apply the same argument to any anti by pointing out that their streaming video and online gaming uses, oh no, data centers and a GPU. But those uses are okay because they like doing those things.
Applies to industrialisation as well. It was exploitative and gave fruit of environmental damage in the case of Western Countries (And other parts of Europe, like Russia). But when it's China or India it's a great effort by a previously opressed people to gain productive autonomy. Even though due to getting a second-hand rendition of industrial technology from the West, they cause duble the pollution and have to exploit their workforce even more to gain faster development rates. They base a lot of anti-something arguments on similar hypocritical conundrums.

eye4.png
Will you use this in a project? Looks like a homosexual beholder from DnD.
 
Will you use this in a project? Looks like a homosexual beholder from DnD.
Nah, I'm just screwing around/justifying the hardware investment.

robot_scaled.png

I would find it problematic (in the actual sense, not in the pearl clutching sense) to generate game assets with an AI, because consistency is absolutely not a given, even with a style lora. Pixel art would be even more problematic, as image models don't really generate consistent pixel art. (That said, you can go in and fix mistakes up easily with pixel art, problem is again style consistency). Then the eternal question is: Do I go in and fix that image I just generated, or do I rather try 1,2,3,4 [...] times again? That said, we're getting closer and closer where I think that's gonna be possible, just if you want any consistency to your art assets, we're not there yet, IMO.
 
https://youtube.com/watch?v=Ofhokz7exqg
This man will never stop at his bullshit
I wanna know how this guy get the numbers, because I don't feel being open about ai reduces sales in anything. Perhaps being a layer of gatekeeping against these kind of people but I can't see a four-digits tops sale reduction would affect the sales much, unless you only got four digits for your total sales.

If your game can only get to four-digits of total purchases, you've got much bigger problems than if you use ai or not.
 


same calibre of argument too, the people whom constantly push China will implode any moment now have Xi halting a rampant housing bubble 7 years ago as their entire argument as to why it MUST happen. Anti AI loons are just as deluded by mere fact of thinking the genie can ever be put back into the bottle, even ignoring their arguments are also basically hollow as their heads.
 
Ed Zitron is another guy who seems to be everywhere nowdays claiming the AI bubble is about to burst. He's a bong living in Las Vegas who made his start predicting Meta was going bankrumpt. oh and just happens to also run a PR company. His whiny cockney accent also sounds forced.

Two plus years of that prediction not coming true, he's basically turned into Peter Schiff 2.0. Like all doomers, if it doesn't come true then double-down with an even bigger doom.

 
This is what every game developer sounds like right now.
bafkreibyrufhl5kvyhsghxpwgglkyanxued5cqpwj27pj5r2fdlvcpwxfe.webp
A chef insisting that his cake made with diarrhea is perfectly fine to consume.

If we successfully make GenAI synonymous with diarrhea I'll be so... so happy.

"Ah, you used diarrhea as a key production element in your game? Yeah, I'll pass."
"Ah, so you replaced animators and vis/dev artists with diarrhea? Yeah; na; you keep it; I'm good."

Every time I hear “we only use ai for the minor stuff”. I also hear “we only use literal poop for the ingredients you don’t notice in your food”

This comic is a metaphor.
bafkreiarg6zyebxj4hadyoyr4sjprglpui4x3ppaseo2zyjgtu5cbzascq.webpbafkreiebvyavj6dkgteopxizpzqhniqg5fey635ngbnflc2fumm5hu6eye.webpbafkreicrfl5svxirq256zzcop2ek3rxhidi6w7dehtfikwtk6ypxgvqz3a.webpbafkreigo5ebxxaeogss7ujqewsw3a6lzs3flze2jhsxm7iyokn74lug3di.webp


What is this 2Girls1Cup-esque fixation?

The above should show that we should reject people with AIDS out of principle... the same way we discard the opinions of people with pronouns in their biographies. Notice how they are no longer pushing th environmental impact of AI, the scarcity of RAM, or even the training on pornography of real children. They are definitely not bringing up how deepfakes can be used in gaslighting or even just outright lie to people, or again, the Secure Data Act. The fac tha they ar emphasising the 'poop' angle shows tha they not only ran out of excuses but never had real arguments in the first place.

Also, the correct equivalent would be 'we involved poop when brainstorming the recipe, but neither the current recipe, much less the actual cakes, have any poop at all.'
 
https://www.mixfont.com/ghost-font WOW, A MESSAGE GENERATOR MADE TO POISON AI FEEDS! HOW PROPERLY NORMAL AND NOT AT ALL PARANOID!
BTW it means nothing. Its all for attention.
Ver archivo adjunto 9262210Ver archivo adjunto 9262211
The project website doesn't really have the same seethe energy as Glaze or Nightshade or mention anything about training. And unlike some people, this guy has actually tested it on frontier models. Seems like it's just a modern CAPTCHA, which he even acknowledges at the bottom:
There are certain implications for Ghost Font that I think would be interesting to continue to explore. For example, it would be interesting to incorporate Ghost Font into CAPTCHA systems, as most systems are easily solved by AI today. Using motion in a video would be a way to make it much more difficult for an automated bot to decipher but still relatively easy for a human to read.

Ghost Font might also be an interesting way to benchmark AI progress when it comes to visual perception. Right now, multimodal models are image-based, and even when passed a video, they usually split the video into frames and analyze individual frames. In the near future, I assume there will be a video-native model that will be able to read the text directly.

A final lesson is that AI is certainly getting really good. While Ghost Font is hard for AI to read, it's also pretty hard for humans to read! The gap continues to close. It will be interesting to see what the future holds when it comes to AI perception and multimodal models.
It's a neat project IMO and there are plenty of good uses for CAPTCHAs.
Ed Zitron is another guy who seems to be everywhere nowdays claiming the AI bubble is about to burst. He's a bong living in Las Vegas who made his start predicting Meta was going bankrumpt. oh and just happens to also run a PR company. His whiny cockney accent also sounds forced.
Never ask a man his salary, a woman her age, or Ed Zitron which AI companies his PR firm works with (archive).
 
"Guess I'll stop eating cakes, then."

Good! We don't want you here either! Deny yourself one of the only things giving you joy in what you constantly bitch about being a hellhole of an existence, so that you're even more miserable!

Except you won't, because you don't have the willpower to stick to your retarded boycott. If you even last a year, something will eventually come out with AI involved that you can't resist buying, you'll grumble about it quietly but give in anyway. And each time, it becomes a little easier to rationalize it to yourself to buy the next thing. Eventually everything will have been made with at least some minor AI contribution somewhere, just as currently everything is made with at least some minor computer-based contribution somewhere, or internet-based contribution, and it will be as normalized as both of those.
 
Almost fedposted on a work chat recently.

For context I am partly involved in the management of our AI credits with one of the popular model providers. They recently fixed or properly enabled their per user limit feature and we had a bunch of people complain they hit their limits on a Monday. Bear in mind that the limits are set to the equivalent of over $350 a month.

A month or two ago our leadership team were joking that people should be maximizing their use of tokens, now our messaging is to restrict it for costs. So our response to these users was essentially to deal with it and just not use AI until their credits renew.

Some of them began asking if exceptions could be made for, and I quote, "Top performers." I don't know what world they live in where they've convinced themselves they're a top performer because of the amount of LLM tokens they use. I was about to reply with a plain, "top performers should be fine with less, not more" (only laced with a bit more sarcasm), when our director gave an explanation about the costs and plainly said it should be enough and no exceptions should be made. He's super pro-AI too so I am guessing other directors and board members have seen the associated costs and been in his ear about responsible spending.
As an update to this, I've now learned they're going to reduce the amount of credits to be even lower in the coming weeks. It looks like LLM token use is quickly becoming a limited/rationed thing.

I am almost certain some people won't survive this performance wise. At least one person on my direct team whined when they hit their limit. To be honest it'll be their own faults for taking the propaganda and recommended ways of working (tokenmaxing) uncritically.

I've been trying to give advice like use smaller models for some tasks, turn off the super fast modes, decompose your work into small units, provide context up front, and my pet peeve recommendation: automate repeated tasks with scripts rather than asking your agents to do the same thing over and over.

This is annoying as, while I am more sceptical than most on AI, I do feel it can be a useful tool. So the people basically replacing all their thinking and working with it are effectively draining all the budget for it's more thoughtful use.

I want more AI in my game.

I occasionally trawl AI mods (usually mods that put LLMs into games) and the last one I tried was rimtalk for rimworld. Often these mods are very hit and miss because they're very often also vibecoded and the modder doesn't know wtf he's doing but the rimworld one is decent. Basically makes these little pawns talk to each other, taking their personalities, backgrounds (stuff already in rimworld) in account. With a decent model the writing isn't any worse than what you'd get in a 90s SciFi TV show and you can even customize their personality prompt-wise and fully customize the characters you want in there. Shit's amazing and the best thing is it's not static, literally anything can happen. Why *wouldn't* you want that in your game? From all things this intense hatred of AI (be it AI graphics or other generated content) makes the least sense, considering how slop it already all was pre-AI. But I guess these autismos are strongly connected to vidya so it makes sense you'd find them there.
Is there a thread where you can recommend some of these LLM mods for games? I actually think they're a pretty neat idea. I know Nvidia released a small LLM specifically for this purpose and others that could run on local hardware, but even remote models probably work well if used right.
 
Última edición:
Is there a thread where you can recommend some of these LLM mods for games? I
Don't think there is, these mods showed up for many, many games at this point though.

A few notes:
One of the most cleanly executed I saw to this day was for Barotrauma funnily enough, which is basically Space Station 13 underwater. You can order the bot crewmates around in natural language and it works actually very well. It mostly stuck with me because of how reliably it worked, really. But really, any game that has a somewhat active modding community has probably a mod or two for LLM integration in various states of quality. There's also one for Kenshi I really wanted to try but the modding harness it needed I couldn't get to run in Linux.

I also tried one for Crusader Kings 3 which was yank AF because of the limitations of the scripting language. The problem with using LLMs in roles like this is that they are from the ground up trained and tuned to be helpful assistants. This means that they don't really manage to be antagonistic and it is easy to talk them into things. This is probably generally a problem with LLM mods where the AI can actually access game functions. It did felt quite immersive though, playing an adventurer and just planning out the path ahead with your companions or convincing random courtiers to join your merry band. Good combo with the Elder Scrolls total conversion, which is of course a universe any modern LLM understands well. I also liked the function to write letters and abused it to write letters to random AI leaders that got more and more exasperated at my pointless rambling letters

There's also one for Mount & Blade: Warband people sing very high praises for. Apparently, a total gamechanger.

The one for Rimworld (rimtalk) is interesting since rimworld basically only lives through mods and coming up with scenarios and goals in your own head is rimworld's bread and butter, and LLM usage is amazingly suited for it. You can write a system prompt for the AI to describe the general situation and the AI can pretend anything, from roleplaying a band of raiders to roleplaying a buddhist monestary. Since LLMs inheriently understand context well and are masters at making bullshit up on the fly, it will also easily deal with stuff added by other mods.

What I found very interesting is that the chinese modding community is very involved with LLM mods in general and especially in rimworld. The chinese love LLMs in their games.

There's a company named Player2 that tries to lock that market into it's cloud based llm services with it's shitty proprietary client and some mods are only compatible to their proprietary client you need to run along with your game to talk to their servers. I can only advice to stay clear of that. The OpenAI compatible protocol everyone knows and uses is very simple and flexible, there's no need for some shitty proprietary app and I'd be wary to make everything dependant on such.
 
Ed Zitron is another guy who seems to be everywhere nowdays claiming the AI bubble is about to burst. He's a bong living in Las Vegas who made his start predicting Meta was going bankrumpt. oh and just happens to also run a PR company. His whiny cockney accent also sounds forced.

Two plus years of that prediction not coming true, he's basically turned into Peter Schiff 2.0. Like all doomers, if it doesn't come true then double-down with an even bigger doom.

zit.mp4
Ed is coming off as a "the sky is falling" type of guy, but he raises good points about the financial side of the AI, particularly the data center deadlines coming up with billions committed but nothing being delivered yet.
 
Atrás
Top Abajo