hansworst

With the amount of synthetic content that is already on the internet, and with even more to come, I think we'll unfortunately need to find a way to automatically filter that content to only what is really relevant to us.

People are using AI to create incredible amounts of SEO content, to the point where just using Google to find something is becoming increasingly useless. If, on the other hand, you had an AI agent that automatically filtered out all of the noise, you could still find the real information you need. What would be even better is if the agent could learn about the things you find interesting, and the things you'd rather avoid.

Now the question is: do you want to have control over such agents and the data they need, or would you rather let some centralized entity manage them for you?

Almost all graphQL clients implement a custom caching layer (for example using indexeddb in the browser) that can cache resources with different timelines, also when they’re returned in a single response.

> (despite the French name)

Claude is named after Claude Shannon, founder of information theory. I guess it is a traditionally French name, but he wasn't a French person.

Arguably Google is the company that triggered that paradigm shift by publicly sharing (some of) their work on AI like the transformer/attention paper. So if anything they were ahead of the curve in terms of research. They're also extremely well positioned in terms of training data, infrastructure capabilities, hardware (TPUs), they had the first popular machine learning library (TensorFlow), etc.

Lately you could argue they're being overtaken by their competitors, especially in terms of productization. But they still hold pretty cards IMO.

That's right. They use a fully differentiable renderer, which allows them to optimize the properties of this set of 3D gaussians using back propagation / gradient descent.

But ultimately what they end up with is an explicit representation of the scene, so unlike NeRF there's no "inference" during rendering. In that sense it's somewhere between a traditional mesh representation and an implicit representation like NeRF or signed distance fields. That's what makes it fast too; they get to make use of the rasterization acceleration capabilities of GPUs , unlike NeRFs which need to be sampled many times along a ray to render a scene.

I wonder though whether people will upvote differently knowing that there's real money involved.

There’s a movement called birthstrike with people who are refusing to have kids because of concerns around climate change:

https://www.theguardian.com/lifeandstyle/2019/mar/12/birthst...

I guess these were the only people to deserve the term “grammar nazi” then

If everyone can share a small fleet of self driving vehicles, they don’t need to take up a lot of space at all. I’m assuming you were talking about parking space; I don’t see how self driving cars would take up significantly more space than mass transit while on the road.

Check out https://github.com/NVlabs/instant-ngp.

Trains a nerf in a couple of seconds.

I went on a trip to Italy last year and took a video while walking around Michelangelo’s David. Even with that relatively poor image quality this let me turn it into a pretty high quality nerf.

If you know the author is an expert in a certain field, you can use that as a heuristic for deciding whether reading the entire book is worth your time.

Webmaster, as in the World Wide Web, invented in Europe?

That's a very big assumption you're making. Not saying it's necessarily false, but I think it's probably wise to remember you're literally commenting on a platform run by a venture capital company. It doesn't take that much effort to create an account here. Who's to say the comment you're replying to wasn't written by ChatGPT?

There are plenty of website builders that don’t require you to code. Platforms like squarespace or wix make it really simple.

> It's so that you can be asked "Did you sign this?" and you would be lying if you said no.

If you agreed to something verbally, and are then asked "did you agree to this?", you would also be lying if you answered with "no".

So it’s pretending to be an average human, and the average human gets it wrong a lot? I wonder if you can improve coding results by telling it to act as an expert programmer?

Or, you know, maybe companies could just respect the laws in the countries in which they do business. And if they disagree with those laws, they can either ignore them and pay the fine, or simply choose to not do business in those countries.

The future isn’t as certain as the last 200 years? Just in the previous century we had two world wars, followed by a period with serious threats of all out nuclear war. I’d say we’re not necessarily doing worse than that right now.

Yeah, if my memory serves a lot of UI design started to get more airy around the time smartphones became more popular. Even on desktops, which I think may just be because large buttons and text became trendy and although some may deny it, I think a lot of ui design is just following trends. It takes a good designer to not just blindly copy things because they look cool.

Do you have some numbers on how fast this actually is? I know it’ll be hard to give specifics, but is this fast enough to convert a somewhat complex sdf into a mesh in (near) real-time? Could you for example build a modelling app around this that uses sdfs as an underlying data model, and uses this as an in between step to make rendering faster?

Pretty sure those other processors at least have phone numbers you can call.

If someone illegally occupied your house, would you be okay with them getting fined for it?

I agree with everything except for the billions of dollars part. It seems quite obvious (but maybe I'm wrong?) that most of that investment went into the hardware that Meta has been developing over the years, and some of the "platform" capabilities that surround that hardware (think their app store for example). Sure, the virtual experiences are part of it. But I think it's not where the majority of the money went.

They currently have the most popular VR hardware platform. And it's the only piece of hardware that Meta really has, everything else they make is software. If Meta wants to keep growing, being the leading producer of "the next big thing" in hardware would certainly help. Zuckerburg is betting on VR/AR being that next big thing. Only time will tell I guess. But I find it a little weird when I see people commenting on Meta's investment as if all they did was create a basic second life clone. They're building some of the most innovative consumer hardware at the moment. I'm not hugely into it because Meta doesn't really seem like the company I'd like to entrust with cameras pointing at my eyeballs, but you can't deny that the hardware they've built so far is quite impressive.

Well, there is SmoothLife (e.g. https://www.arxiv-vanity.com/papers/1111.1567/#S4) where the time step is also made continuous. I suppose you could extend this so that this isn't some uniform value across the entire space, but instead a value that is constantly recomputed based on neighbourhood density.

You might like micrograd: https://github.com/karpathy/micrograd. There's also this video where the author explains how he built it: https://www.youtube.com/watch?v=VMj-3S1tku0

This sounds like it should be illegal imo. I’m not Swedish but your country strikes me as one where there are lots of protections for all kinds of people. Your (mental) health status really shouldn’t have an impact on whether or not you can get a mortgage, you cannot change anything about that. To me it sounds just as bad as profiling based on race or gender.

Would be really interesting to have some kind of tool that automatically laid out some information you want to learn in a 3D virtual environment. Imagine if you could just tell this system you want to study a certain book, and it automatically generates a fitting virtual world for it. A kind of instant mind palace.

If the creators are reading along in this thread, I’d be very interested in an article or even a short explanation with some pointers on how they got this to perform well enough to render in real-time. Every time I’ve tried to create SDF based shaders with many Boolean operations, it kind of starts slowing down to the point where it’s unusable.

Big funds and billionaires can influence those employees with hard cash, and governments can influence those employees with threats of incarceration.

You’re commenting to a post that’s saying everyone is burned out. It may not be “the world is on fire” but I’d say it’s pretty close.