kache_

I thought VRChat was all about hacking things together and tinkering with VR, from my impression of the people I've met in there.

They're going to lose a big portion of their clientele

WE"RE ONLY HERE ONCE

WILL YOU PERISH LIKE A DOG?

OR WILL YOU STAND AGAINST DESTINY

AND FIGHT

NUT THE FUCK UP KID

LETS FUCKING GO

I say this with ironic humor, but also deliver a lot of truth in it. I used to have a pretty comfortable job, but I hated it and I was bored out of my mind. Existential dread every day. Find another job/endeavor that challenges you

That's like saying "watch the decline of C++" while using javascript What does javascript run on?

Most ML advancements will use DL at the core in interesting ways.

This is a problem with generative models across the board. It's important that we don't skew our perceptions by GAN outputs as a society, so it's definitely good that we're thinking about it. I just wish that we had a solution that solved across the class of problems "Generative AI feeds into itself and society (which is in a way, a generative AI), creating a positive feedback loop that eventually leads to a cultural freeze"

It's way bigger than just this narrow race issue the current zeitgeist is concerned about.

But I agree, maybe I should skew to being optimistic that at least we're trying

While their heart is in the right place, I'd like to challenge the idea that certain groups are so fragile that they don't understand that historically, there are more pictures of certain groups doing certain things.

It's a hard problem for sure. But remember, the bias ends with the user using the tool. If I want a black scientist, I can just say "black scientist".

Let me be mindful of the bias, until we have a generally intelligent system that can actually do it. I'm generally intelligent too, you know.

Yeah actually, good call. The position of the token matters, since these things use transformers to encode the embeddings.

https://www.assemblyai.com/blog/how-imagen-actually-works/

They literally just add "black" and "female" with some weight before any prompt containing person.

A comical work around to so called "bias" (isn't the whole point of these models to encode some bias?). Here's some experimentation showing this.

https://twitter.com/rzhang88/status/1549472829304741888

As competitors with lower price points prop up, you'll see everyone ditch models with "anti bias" measures and take their $ somewhere else. Or maybe we'll get some real solution, that adds noise to the embeddings, and not some half assed workaround to the arbitrary rules that your resident AI Ethicist comes up with.

Give it some time. Other organizations will race to the bottom.

They might even provide image generation at a loss to drive people to their platforms.

lol

>levels.fyi

wait until this guy finds out about financial regulations

No demo, no read, no cite

>For other topics that aren't strictly programming, eg AI, ML, game theory, cryptography, quantum computing, probabilistic methods, optimization,... it would take a significant amount of dedication to match the amount of understanding that a rigorous course with problem sets can give you in self study.

You mean, the same amount it would take to do it through a university program? I fundamentally don't believe the material is unavailable. If you've done a CS undergrad at a respectable university, you should have enough instinct to build a curriculum to become well versed in those subjects

I am going through the autodidact process for all of those subjects (except quantum computing) (yes, it's brutal, especially alongside a FT job). But with the amount of MIT open courseware, books, and frankly, just wiki pages available, it's been easier than doing it at the pace of a university course

I was about to write a blog post against CS masters degrees. I'm going through a pretty ambitious gauntlet autodidact process. Some background: I already have a CS undergrad degree.

- I've found all the online courses I would ever need, alongside high quality video courses (MIT, standford, 3b1b) + books

- A masters costs money & too much time (my time is better spent than getting an application together) (Opportunity cost of not working is too great)

- Arbitrary wait time (I have to apply like, 8 months in advance to starting? https://en.wikipedia.org/wiki/Mathematics is available right now)

- I already know 80% of the stuff in a CS masters from building systems and learning how they work (looking at the syllabus of various online masters degrees)

- My employment history is much better for future employment opportunities than a masters ever would be. Getting a FANG job not only pays you, comparatively to a masters degree. It's also a much better signal than "cash cow online masters program"

- I can publish papers at my company if I wanted to (Not being gatekept from becoming a researcher)

- My workplace has a mathematics "guild" who I can outsource questions to whenever I want. Otherwise, I can go to /sci/'s math general or stack exchange.

In university, I learned the computational algorithm to pass the lin alg exams, and that took eight months. Being an autodidact, I relearned the true essence of lin alg in two weeks. Now, I can derive any computation lin alg operations from first principles. I'd even argue that undergraduate degrees are a waste of time. Mandatory electives? Give me a break. After going through this autodidact process, I'm actually offended by how much time I've wasted in HS & uni.

I'm basically redoing an undergraduate math degree + a CS masters + building real projects throughout it, all in the fraction of the time frame. At my current speed, I'll be finished learning all what I want to learn in ~2 years, while maintaining a pretty demanding FT SWE job. The knowledge provided in both highschool and university can be obtained in the fraction of the time, on your own. At the end of the day, the only thing that matters is what you can engineer & build with your knowledge.

Universities are a meme in 2022. They're largely mechanisms to filter out the "peasant non-educated" class, through a proving function that can only be passed by coming from a family with enough resources.

I'm DONE being slowed down by institutions

Vote with your feet

Quit your job

Seriously..

Step 1: train a model to deduce emotion from vocal tone Step 2: use a model to transcribe the text Step 3: use it as inputs for an sentiment model

Use some int, peopel

I really can't believe people fall for engagement bait

Meanwhile: I'm responding to a hnews post about engagement bait because it makes me upset

I started boycotting food delivery services because the quality & service was just god awful. I found that if I drive, I actually don't spend that much time, and I get wayyyy better service.

Patent law is going to get really ridiculous with all this AI IP generation

The general ethos of most hacker SWEs I know building cool shit is: just build, if I get sued, it's a speeding ticket (That some VC will give me money to hire some lawyers to deal with).

Thanks for sharing this. IMO: Close enough, and good enough :)

looks like a whole lot of bullshit

Doesn't scale :P

"tied to cancer" Bullshit. I've looked this up and researched it extensively as I was looking at different chemicals to use for weed control

The science isn't settled, this is just more news media garbage fear mongering

At what point does the anti-competitive nature and monopolistic qualities of Canadian telecoms become a national security threat?

Maybe when it destroys the whole Canadian economy for a day

We're going to see even quicker advances as more people jump into the field after the big recent successes, like image generation & text generation.

We're going to build some pretty incredible things.

Here's a few 'non-experts' who just started building and selling to the government: https://www.anduril.com/

Seems like they're doing quite well, don't you think?

No matter how upset it makes you, you can't stop us hackers from building technology

I'd rather have the humans that live in my backyard with the weapons, than people with a less vested interest in my safety.

I'm not being sarcastic

Meta/Facebook is actually a good company, and I trust them. They're really clear about how they use your data, and as far as I can tell (being a FANG worker), you can trust what they say about data usage

In fact, I'd trust them with my data more than a bank (competency is more important)

just build it

Knowledge can be farmed out to the internet & peers

Grit, creativity, and int stats are what matter the most

Never. Never. Never give up

Anything is possible. We sent a man to the moon with kbit computers