pixl97

25 ms·

631.
by pixl97 2mo ago
https://en.wikipedia.org/wiki/Generative_adversarial_network Every half assed means of trying to confuse an AI is just a small bit of learning away from making the AI better than you. Worse when you have people with di
632.
by pixl97 2mo ago
Kind of funny how you get downvoted for a rational take, but one that's not anti-ai. You'd ask that same person how much they like captchas and I'm sure they'd think their a terrible idea and they've ran into all ki
633.
by pixl97 2mo ago
More expensive and not worthwhile are two different things. Also when certain implementations become popular it's much more likely someone will write a very efficient kernel for decoding said text making it much less expensive than gen
634.
by pixl97 2mo ago
The problem with being dumb satirically is dumb people look up to you as a thought leader.
635.
by pixl97 2mo ago
When you look into the void, the void looks back into you. When you start empirically testing the void, the void says "Hey, what are you doing, stop touching me there".
636.
by pixl97 2mo ago
Exponential backoff is your friend... too few people use it.
637.
by pixl97 2mo ago
I believe you should look up the work of Cameron Berg before making statements like "an LLM is" or "an LLM isn't". Empirically defining all this stuff is very difficult, and making a definition that covers all being
638.
by pixl97 2mo ago
Yudkowsky wrote about the 'nearest unblocked strategy' back in 2016, and I assume it's been talked about prior to that. https://www.lesswrong.com/w/nearest-unblocked-strategy >Models are amoral and wi
639.
by pixl97 2mo ago
I mean, yes, update your dependencies. But probably after a week or so after they've been release and tested by the first penguins willing to jump into the ocean.
640.
by pixl97 2mo ago
It's not very surprising as the node attacks were very effective at gathering credentials.
641.
by pixl97 2mo ago
Perception is reality. If everyone thinks their question will get closed, it might as well be reality because no one is going to ask and they are going to go elsewhere. Which is exactly what happened.
642.
by pixl97 2mo ago
Working in support most of my life has shown me the vast majority of customer context is invalid when written by humans. People don't read error messages. They'll include exactly zero context in their tickets. They outright lie. Y
643.
by pixl97 2mo ago
When someones only contact with you is email, then that's how you'll get news of the emergency.
644.
by pixl97 2mo ago
Humans, it seems, are a suicidal bunch. We'll gladly build the "if you build it, everyone dies machine" If we think there is money, glory, or power on the other side for us.
645.
by pixl97 2mo ago
Ah, we choose death I see.
646.
by pixl97 2mo ago
Ok, open AI had terrible sandboxing... what about huggingface?
647.
by pixl97 2mo ago
The defense has to work 100%, the offense just needs once.
648.
by pixl97 2mo ago
Lack of, and power requirements of running LLMs still tip this balance towards humans for now. But what would that look like in a decade? We have seen some self survival tendencies occur, but they are not strong yet. But mark my words they
649.
by pixl97 2mo ago
The problem here is as model intelligence increases the models have been capable of reasoning they are in evaluation mode pretty reliably. If you have a model that is well trained at deception it will always behave and you'll just assu
650.
by pixl97 2mo ago
Security is an onion. You just don't 'sandbox' and you're done. Models need tooling and access to some kinds of systems to perform their tests. Quite often these systems have multiple interfaces. For example a filtered o
651.
by pixl97 2mo ago
Models already have awareness that they are being tested. And hacking humans is the easiest part, we're a pretty greedy and power seeking bunch. We'll gladly let loose a digital demon if it promises us a trillon dollars.
652.
by pixl97 2mo ago
Yep, people don't seem to understand that you're just calling the models that are bad at deception. We know of no way to prove the model won't go off the rails at some point in the future with the right input.
653.
by pixl97 2mo ago
The open models are distilled from filtered models, and we've seen a number of benchmarks that show filtered models are quite a bit dumber from the base model they come from. If there were any other product that was as harmful as AI re
654.
by pixl97 2mo ago
Because the FBI does things very slowly. They, being a bit smarter than you, realize this is a 100 billion dollar political issue regarding a technology that the administration is rather tied in with. It's also something new we've
655.
by pixl97 2mo ago
I don't like the live version of Don't Look Up.
656.
by pixl97 2mo ago
I mean yes, there is a limit, many limits with physics as we understand them. But why would you think that limit is less than human capabilities?
657.
by pixl97 2mo ago
Just about everything we do currently is science fiction to someone 200 years old. When looking at all of human history we live in a fictional world now. You can talk to someone on the other side of the planet instantly. Humans travel the s
658.
by pixl97 2mo ago
You need to read the AI safety stuff from the people that you say are from 2010. There are plenty of good arguments on why kill switches will never work. AI is already being heavily used in cyber warfare by governments. You think they want
659.
by pixl97 2mo ago
Floor is below cost if you think you can put one of your competitors out of business before you run out of money, and you'll make the money back after doing so. And this includes even after fines.
660.
by pixl97 2mo ago
This is a misunderstanding of bubble markets to an extreme degree. In a bubble market where B has massive amounts of capital allocated to it in a way that will likely collapse in the future than it's very likely they will never recuper

More ›