[aedocw]

Except an environment that is highly toxic does not attract decent people with good community intentions. While you have learned a lot about LLMs and generative AI for instance, it is extremely unlikely you have learned from the most well informed sources. It's definitely a community, but nobody can reasonably argue 4chan is a community that creates the best/most useful content.

[thot_experiment]

I don't know about that, there's info there that's unavailable anywhere else. I've on a few occasions found that an anon I was interacting with was the primary author on a project I was contributing to or had written a paper on the subject. It's a very diverse bunch.

[mjr00]

> While you have learned a lot about LLMs and generative AI for instance, it is extremely unlikely you have learned from the most well informed sources.

4chan is certainly not the best source compared to, say, scientific papers or technical conferences, but is there any evidence it's any worse than Reddit? A lot of the work pushing Stable Diffusion forward came out of 4chan IIRC. I believe the most popular UI (automatic1111) came out of 4chan.

[josefx]

> While you have learned a lot about LLMs and generative AI for instance, it is extremely unlikely you have learned from the most well informed sources.

You are talking about tools that can be abused to make infinite porn. I am not sure anyone wants to know how much time 4chan users invested into learning about them just to be able to modify and fine tune their models.

[trompetenaccoun]

People hugely underestimate how big of a factor motivation is when it comes to research and building things. Some are motivated by money, some are motivated by the feeling of being needed and helping others. And then there's some boys who're really dedicated to figuring out how to fake nudes. The reason why they do it is unimportant, what matter is the amount of time they spend on tinkering with it. In the end they still acquire skills.

[something168581]

> ...tools that can be abused to...

Interesting choice of words. Why would it be an "abuse" of a tool that can be adjusted to generate nearly anything, to use it for a relatively specific subset of anything?

Would you also consider it an "abuse" if they were, say, attempting to generate new works in the style of michaelangelo or picasso instead?

[tick_tock_tick]

You really need to reevaluate where a lot of the tooling and focus is on by the community. It's a ton of porn and 4chan is really big in the space and absolutely creates some of the best content.

[Applejinx]

That's far from the only point, though. Motivation is almost moot. (not mootykins, either!).

Something developing this way is brought forward best by intensely disagreeable nerds with too much time on their hands, exploring in an undisciplined but persistent way.

Tell me that doesn't sound like 4chan :)

[Applejinx]

I'm not at all sure you're right.

I think 'the most well informed sources' in many things will NOT be the consensus opinion, but vanguard thought trying to get somewhere else beyond the sum total of what's currently known. This is fundamental to science and truth generally.

Disagreeableness is overwhelmingly a quality of such thinking, in the technical psychological sense: agreeableness leads to complying and ceasing to seek alternate thoughts/points.

I run Airwindows and have recently put out a Console system based on applying the absolute minimum alterations to the mantissas of floating point words, just to see what I'd get. That's internal algorithms, volume, and I added panning, doing nearly everything with bit-shifts for this purpose. Got a very interesting sound out of full mixes done this way.

I am as nice as pie, and putting videos on YouTube where I must talk about 'shifting of bits' because Youtube will always think I said 'bitch' and draw conclusions about my video from what it thinks I keep saying. And yet, at the same time, I am intensely disagreeable, because I'm persistently exploring ideas in digital sound that STARKLY contradict what's taught as common knowledge.

I'm not doing this on 4chan, but if I was, it would be every bit as supportive a community to my unusual directions as the 'normal world' is, if not more so. I wouldn't be one bit surprised to know that 4chan had pockets of research or content creation that are leading the world. It wouldn't be the first time 4chan led the world, for good or ill.

Ok, so mostly ill :) but that is because influential is not the same as useful. And the ability of 4chan to put any idea or content into effective practice is grossly stunted as 4chan cannot govern, or organize worth a damn by real-world standards. But it's probably the most likely source for whatever is going to be moving the world twenty years from now, compared to normal sources. It will just immediately lose control of anything that achieves escape velocity.

Think 'petri dish', or perhaps 'rainforest swamp'?

[1337biz]

You assume that complying with social norms attracts the most brilliant/informed people.

[GenerocUsername]

It's very 'reddit' to assume the brightest people share the same societal conventions as themselves.