@Porean's banner p

Porean


				

				

				
3 followers   follows 1 user  
joined 2022 September 04 23:18:26 UTC

				

User ID: 266

Porean


				
				
				

				
3 followers   follows 1 user   joined 2022 September 04 23:18:26 UTC

					

No bio...


					

User ID: 266

Most of your post is in line with what I believe. The information workers in blue tribe will turn to protectionism as AI-generated content supercedes them. Red tribe blue-collar workers will suffer the least, and the Republicans will have their first and last opportunity to lure techbros away from the progressive sphere of influence.

There is one thing, though.

I simply do not forsee Republicans being likely to make AI regulations (or deregulation) a major policy issue in any near-term election, whilst I absolutely COULD see Democrats doing so.

It only takes one partisan to start a conflict. Republicans might not initially care, but once the democrats do, I expect it'll be COVID all over again -- sudden flip and clean split of the issue between parties.

But this is just nitpicking on my part.

and some modest computation capability (say, a cluster of 3090s or a commitment to spend a moderately large sum on lambda.labs)

This is not sufficient. The rig as described by neonbjb is only 192GB of vram; fine-tuning an LM with 130B params (in the best possible case of GLM-130B; the less said about the shoddy performance of OPT/BLOOM, the better) requires somewhere in the ballpark of ~1.7TB of vram (this is at least 20+ A100s), and that's on batch size 1 with gradient checkpointing and mixed precision and 8bit adam and fused kernels without kv cache and etc. If you don't have an optimised trainer ready to go (or god forbid, you're trying distributed training), you should expect double the requirements.

The cost of that isn't too bad, of course. Maybe $25 bucks an hour on LL, any machine learning engineer can surely afford that. The larger doubt I have is that any of this will take place.

or?

That does nothing to dispute the claim that supporting pro life policies are costing republican votes. Ditto for basically everyone in the parent comment tree. I don't understand how so many motteposters are conflating "There's lots of pro-life people" (true) with "Being pro-life will make it easier to win elections" (do you believe this?)

Because of course every single woman is pro-abortion, of course a Republican-voting guy is not going to know any women who might be pro-life, of course no woman picked at random in the USA is going to be "I think the Supreme Court decision was great".

Are you reading the same article as me?

Redd: And by the way, I do talk to my female friends about abortion. Abortion is a problem for Republicans. That’s why I’m not sure if we’re going to win small or win big. But pretty much everything else lines up on our side.

Ah, well... Fuck. That covers every idea I had and more.

Yes indeed, that todo link should've been replaced with a link to a transcript of Emad's recent Interview.

I failed to find the transcript in my browser history, so I've relinked the video in its place.

That's actually really cool, wow.

I don't see this as superstitious/magical. You are basically pressing the "purge all thoughts" button by spamming your brain with a single repeated concept.

I feel we are talking past each other. "In terms of the historical narrative, some artists were inspired by photography and made a cool synthesis of traditional art && the new technology" -- okay. But were there more artists (adjusting for base rate) creating realistic looking hand-drawn art pieces before or after the proliferation of the camera? Do you agree that the answer is before? Do you grasp the standard concerns shared amongst artists that believe before is the obvious answer?

I can't draw conclusions without knowing what kind of degenerate you are. If you're into hentai, the waifu diffusion model was trained on the 1.4 SD checkpoint && has much room for improvement. If you're a furry, fine-tuned models are currently a WIP and will be available soon. If you're a normal dude, I don't really understand because I honestly think it's good enough at this point.

The only thing I think is really poorly covered at the moment is obscure fetish content. A more complicated mixture of fine-tuning + textual inversion might be needed there, but I do truly believe the needs of >>50% of coomers are satisfiable by machines at this point.

Edit: I am less confident of my conclusion now.

I tend to stare down the same paragraph for two hours and finally squeeze out, word by painful word, something that sounds like the ramblings of a schizophrenic with aphasia

The problem is that you are not writing fast enough. Think about text too slow and the words will blend together and lose all meaning. Put your brain into Word Salad Generation mode and just dump as you would into a Motte comment; you can edit for style/tone/content once you actually have something to edit.

I've shilled this before, but you should really try The Most Dangerous Writing App to knock out a first draft. As described by Alexey Guzey:

DO ACTUALLY TRY THIS DON’T FLINCH AWAY. This app might seem like the dumbest thing in the world but it DOES REALLY HELP. And if it doesn’t work, you will just lose 5 minutes.

I want to be reminded of who has historically made bad/good takes.

How about the fact that you could make a video about literally anything happening at all? Fake any event you want. Nudes, terrorism, declarations of war... ideally we would learn to just ignore all of the fake content, but if we could do that, why would ads be a problem anymore?

I think the migration just failed, and I plan on doing some data scraping / analytics this weekend to demonstrate it.

Bruh

Is there something misleading with the way I phrased my comment? I don't understand why multiple people have succeeded in reading "programmers will be completely replaced by AI" into my words.

And this isn't a nitpicking thing. It is an extremely important distinction; I see this in the same way as the Pareto Principle. The AI labs are going to quickly churn out models good enough to cover 95% of the work the average software engineer does, and the programming community will reach a depressive state where everyone's viciously competing for that last 5% until true AGI arrives.

Your first paragraph misses how hard it is for human programmers to achieve those things, if it is even possible under current circumstances (find me a program that can acquire farmland & construct robots for it & harvest everything & prepare meals from raw materials). Even hiring an army of programmers (AI or no) would not satisfy the preconditions necessary for getting your own food supply, namely having an actual physical presence. You need to step beyond distributed human-level abilities into superhuman AI turf for that to happen.

The main concern here is that we're headed for a future where all media and all human interaction is generated by AI simulations, which would be a hellish dystopia. We don't want things to just feel good - we want to know that there's another conscious entity on the other end of the line.

I can see this as a Future Problem, but right now the "conscious entity on the other end" are simply prompt writers. There is a sense of community to be gained from indulging and working on AI generation together. I think it is misleading to apply the bugman/we-will-be-in-pod argument to text-to-image tools, because new means of human interaction are forming as a result of it.

Also, some of us just hate the majority of conscious entities and are happier with what simulations we can get. This obviously doesn't apply to you or Vaush, but I wonder what brings you both to so viciously condemn the estranged, the alienated, the anti-social.

I've yet to see anyone blow up their life with (legal) porn,

What kind of observation would qualify, to you? Does blowing your life savings on OnlyFans count? Missing a national exam because you fell asleep after a jack at dawn? Ending up arrested for molestation because you thought it'd be as easy?

I don't even care about any of these, because they're edge cases. But when people are willing to condemn all kinds of behaviours except masturbation, I just don't get it.

I am not aware of a single high-quality AI image of two people having sex.

This does exist, but you are right to point out it is exceedingly difficult to make.

Given the volume of responses affirming the failures of generated porn, I'm realising my tastes must've bubbled me from dissent. I mostly consume images with only 1 figure involved && this has evidently biased my thinking.

This list will probably make it a lot easier for rDrama trolls (among others) to fit in, which is a bit unfortunate. At the same time, this glossary of terms is really accurate and well-cited, and I cannot help but admire it for that.

The paradigm and reason for party allegiances is not equality and certainly not about government handouts, considering how every citizen of their home country sidesteps the government to black markets where exchanges are made at better rates. The current Democrat party is actually damaging their bottom line through their fiscal policy.

Call me skpetical. What makes you confident it has a strong impact on voter habits? You have presented a theory -- a possible factor of influence -- but no reason to believe it has any more predictive power than any other explanation of voter preferences.

Why are the majority programmers so enthusiastic about machines that can code but not artists?

Because they aren't. They're collectively deluding themselves into believing in the «soul» and that programming will never be automated by AI. Just like certain artists are.

I am a programmer. OpenAI scares me. I'm putting every effort I've got into the Grind, because I think the industry's due for a phenomenal crash that'll leave the majority in the dumps. You are free to disagree.

Bad idea, there. Knowing their username means you get to filter them from day 1.

In all likelihood, you'd need something like 8x3090, but that's about as hard to trace as a stealth weed growbox in a basement. Inference, I expect, also won't be feasible on normal consumer machines, so it'll incentivize some stealthy cloud computing, maybe very small-scale.

I'll bet against that. It's supposed to be an Imagen-like model leveraging T5-XXL's encoder with a small series of 3 unets. Given that each unet is <1B, this is no worse than trying to run Muse-3B locally.