nullsense

>But being pro-misery because some people make bad conclusions from being against misery is, well, ludicrous

Well, it's not obvious that it is, and you haven't made an actual argument here.

I'm OK with accepting a certain amount of it. It's a thing. Is there a configuration under which there is no misery? What's the minimum possible amount of misery we could hope to achieve? My problem is that when people's thinking tends to go along the lines of "misery is bad therefore we must eliminate all of it" they come to conclusions that are counterproductive because they're trying to break the laws of physics.

>The amount of misery under "capitalism" says there's got to be a better way.

I'm not quite sure why people think that a misery free world ought to be possible and/or the default.

North Korea called, it wants it's ideology back.

It's certainly going to be very interesting what comes out of the training runs that are going to be done on giant clusters of H100s.

>I was at the time quite proud of them all being flabberghasted while I also flipped off over 75 students, actually still am.

Me too son, me too.

So, it seems the trajectory is one of increasing generality and capability of models and increasing reliance on them.

If it's at all possible to improve our technology then we will. If we improve it it increases in utility. If it increases in utility we use it more.

What other thesis is there?

Three observations here. Firstly it has been a really eye opening experience watching the innovation around Stable Diffusion and locally run LLMs and seeing that the unoptimized research code that needed such beefy hardware could actually be optimized to run on consumer hardware given sufficient motivation.

Secondly it wasn't obvious that deep learning was going to work as well as it did if you simply threw enough compute at it. Now that this tech has reached critical mass there is a tonne more money being poured into infra to support it.

Lastly, compute power is increasing as always. Nvidia releasing H100 and also their recent work on computational lithography. Also DeepMind finding new state-of-the-art algorithms for doing matrix multiplication with AlphaTensor. You can kinda already see the positive feedback loop in action.

I dunno... at this point I just wouldn't bet against the trajectory that we're on.

1000x this.

It seems then that it's usefulness depends on the intersection of what you want/need to do, what the model is capable of doing, and what you're capable of getting the model to do.

In some cases the impact might be enormous, and in others perhaps less so. One thing is for sure, the models are getting more capable, and along with that people are investing time/effort/money into improving their capability to leverage what the models can do.

The real secret to the OODA loop was Boyd's observation that there is a positive feedback loop between the orientation -> observation that is devoid of input from the external world and if you can trigger that part of the OODA loop you can get your adversary to behave in a way which takes them away from their goals.

Mostly people just see it as a circle in a PowerPoint slide and it has close to zero utility in that form.

That's the one. Are classifieds still a thing? It reads like one of those.

Leisure Suit Larry was one of my favorite games when I saw like 5 or 6 years old. Was always fun getting through that gauntlet of questions.

About a quarter of a billion dollars to rent a state of the art GPU cluster, that's what.

How much faster can we develop the tech if we leverage GPT-4 to do it?

It's kind of a low bar that the only thing we can all agree on is that we don't kill absolutely everyone. A lot of people is fine, but not everyone.

What was that quote... "Provably unfriendly natural intelligence wants to build unprovably friendly artificial intelligence"

>You already know the intention(s) of Sam Altman and O̶p̶e̶n̶AI.com. It was only to run with VC money and close up all their research up.

Citation needed.

Awesome. Between crypto hype in 2017 and AI hype in 2023 I've acquired a collection of 2x 1080ti, an RTX 3060 and an RTX 4090. All together it's a total of 58GB of VRAM. Is there a way I can pool it all across a distributed cluster of 2 machines for doing anything? I'm assuming it would bottleneck on both network speeds and the slowest GPUs if it's possible at all...

They're probably cherry picking. I know on ModelScope text2video I have to generate 10 clips per prompt to get 2 or 3 usable clips and out of that will come one really good one. But it's fast enough to do that I just sort of accept that as the workflow and generate in batches of 10. I assume it's likely the same for this.

Can't wait to try it though.

Working on that as we speak. We will soon all be nostalgic for the memes of this era. Bear in mind 2024 is an election year. What a time to be alive.

It's more of a "Cover Your Ass with Paper" type thing.

Tortoise TTS can do this. You just pass it your example as a conditioning latent.

Ghost in the machine.

They have stated in their interviews that developing an AGI could be great or terrible for humanity, and it's ultimately uncontrollable, so the best way to get the best outcome is to set the initial conditions under which the technology gets developed to be as favorable as possible.

I'm enjoying programming more than ever because of it.

I think the point is it's not going to suddenly give us surprises like Banksy shredding art.

Don't even need to write the Jira tickets.

As someone with no artistic knowledge or skill who has been playing around with this stuff, I totally agree. I have found a large part of the learning curve to be getting a sense of how to describe the things in the first place.

It does lower the barrier to entry significantly for anyone wanting to get into it, but it's still going to be the case where some people are demonstrably better at it than others.

Of course developers will just solve it by making to be UX such that it uses some genetic algorithm to help you iterate to the exact thing you want even if you don't know how to describe it.

The exponential part of this ride really feels like it's coming in to full view.

It has probably just always been an exponential and any curve can be a straight line if you zoom far enough in and Moore's law was just that. Now we've got some perspective and are starting to see the actual exponential.