Death ray from space?

Yes. You can either give them your real one, or not. That’s the point being made. Actually the point of the whole page is that just loading a website tells a huge amount about you, even if you are behind a vpn and extensions to minimize your fingerprint. You are a product for sale.

It might be better if it was made of cardboard or cardboard derivatives.

Imagine UBI and allowing robots and AI for the mundane tasks that are dangerous or boring. The jobs no one wants to work.

The download is the title and what everyone is latching onto, but few are seeing the other problems, like how it secretly installed that model without user acceptance, how it uses obscurity to hide the model, how it will reinstall if you just delete it (fortunately there’s an uninstall process linked in the comments, does that include uninstalling Chrome?). And then how it pretends to be an extra AI thing on the browser but apparently will be used for any searching. Which is more energy use since it isn’t local, it’s just using the weights in storage.

It’s all bad, even if it wasn’t AI. It’s what malware does.

It can be, but it depends on how dependent one is on various applications and sources. I’m in the process of doing such a thing, but I’m taking the slow route so I don’t break everything. Meaning I’m moving one thing at a time off Snap, and then I’ll be closer to a Debian/Mint DE version in the switch over than if I did it right away with so much tied to it being Ubuntu.

For a casual user it may not be that dramatic and just a matter of moving data and learning a new desktop look. I’d still suggest the Debian version of Mint over Debian if they went to Ubuntu for the easy use.

I know, I was there. I was on Quantum Link, which AOL would later buy out for the infrastructure to build on. I was with Compuserve and Prodigy before moving to AOL. Webpage commerce was already a thing before the crash, it just wasn’t as established as it is now. Even Amazon had started going past just books into selling other things. I’ll agree that what the web would be used for and how was still being figured out, but it was far past any niche thing by the late 90s. The wiki on the crash is an interesting history read. In the end, it was a stock crash because people were literally making up companies that didn’t exist and getting money for it. Getting back to the topic at hand, sounds a bit familiar.

I said questionable in nature in comparison to the other two bubbles, which were established technology to build off of.

I agree on usefulness, to a point. It’s being used everywhere, and while an LLM breaking in some situation can be countered, look how many places it’s being given full reins and crossed fingers. Knowing it makes mistakes. Bad mistakes, Maybe not often, but sometimes you need as close to 100% as possible, with redundancies in place.

I use LLMs myself, and in just using it for things that aren’t critical, I’ll catch that small percentage and realize, it’s not good enough for most things it’s being put in. I do find it amazing that it produces what it does, but that doesn’t make me ignore when it breaks gloriously.

Yes. It’s like code comments.

We all do that, so understand your point.

All of us…

This works, until it doesn’t. And then you can’t go back and find the mistake, and save time by both realizing the error part and not having to redo it all to get there.

I get it, I was bad about that too, but any form of writing improves the understanding of the concepts because it uses a different part of the brain that I think retains better than the one doing the thinking work.

Dot com wasn’t about the web, but about what could be done with it. Those who built something sustainable lasted, the rest couldn’t. Just like the computer company bubble before in the 80s, where anyone could jump onto the new thing, but few lasted more than a year or so because they didn’t have much beyond their beginning idea. The problem with using the for LLMs is that the product itself is questionable in nature, unlike computers or the internet. I have no doubt Anthropic will stick around as they were part of the innovation and still are, but that requires LLMs/AI as it exists now and the direction they continue to go to be viable. The line all of them spiel is it’s a gateway to actual AGI, but anyone who understands the science knows that’s not going to happen. At most LLM tech is used in part by AGI for forming its ideas and speech, but not the core.

So it’s not a bubble like those others. But the tech itself is a bubble, as the attempts to force it everywhere to validate its investment cost fail.

I guarantee that election issues will pop up regardless of ID laws in a state. Because that isn’t the problem, and never was. It is just another way to target certain demographics that tend to vote a certain way. A form of disenfranchisement.

Source - it happens constantly, in every recent election. If you haven’t heard about all the efforts to stop some people from voting, then you must not live in a state where it’s a regular thing.

Specifically individual voter fraud, which voter ID claims to be fighting. Now, voter fraud in the masses via ballot manipulation, voter disenfranchising, gerrymandering, minimizing poll locations and times, etc. THAT is a thing, but has nothing to do with people having IDs, and all to do with people in power making sure they stay there.

And guess which ones support both voter ID and all of those tactics. Yeah, we all know. It’s the only way they stay in power. They can’t win by just their positions on policy. Never could. If they could win fairly, they never would have pursued gaming the system.

If we take that approach with any change, that it might be possible but it also could fail or become corrupt, then we’re just going to spin wheels. Like we’re doing. We have to change something, this isn’t working.

UBI isn’t viable without other changes, but it’s just an improvement of various welfare systems, just without all the paperwork and exceptions. “Too expensive” is usually the excuse thrown around, which is funny given the news of how much killing or disappearing people is costing, all because some person in the White House needs a distraction.

See, that’s how you make actual users buy the badges. Now to be considered not AI you have to put money up. Who wins? Spotify does. And they don’t even have to make verification all that thorough, just enough to say they tried to filter out the worst of the bots.

“Lovely plumage”

Han didn’t see, he’s Force-sensitive, as are many who don’t realize it.

As for Lando vs. the ship, not all canon is good canon. A saarlac that has that much awareness and strength for what it is doesn’t make any sense. And yes, I realize that’s bringing Star Trek rationality to a Star Wars conversation… but still, there has to be a bit of realism kept.

It is certainly inaccurate, but in my mind’s picture of how the transformers work, reducing their quantization and also doing what abliteration does, there is a line where you’ve done a lot of “damage” to the original model and so there will be places where it just hangs or goes off on severe tangents. There are good places for even the 1bit models where they don’t get pushed to hard, but there are limits for them all, including the big ones.

Hugging Face does have a few Q4_K_M versions. Maybe something will fit.

Yep, different stages of working the problem.