Maybe he was impressed at the part where the agent got away with it. If I was a murderous psychopath that would seem like a really appealing job perk.

My completely uninformed speculation on the thief’s motive: He was spying on the defence site, but he learned spycraft only from watching James Bond movies. Not realizing he could have just walked away after being confronted, he stole a car to flee the scene. Throughout the chase, he was certain that stealing just one more car would throw the police off his trail and he could make a clean getaway.

Diabolical.

The problem with the “AI will destroy us all” take is that it attributes non-existent competence to these techs. I don’t deny it may raise some useful opposition, but it also makes the problem worse on the other end. There will be a good chunk of people who look at the extremes of the claims (useless slop vs doom) and figure that the truth is somewhere in between. They’ll conclude it must be a pretty good technology. This is why some of the AI evangelists themselves have been pushing the doom possibility. “Oh no, it might be just too good. Well, not just yet, don’t worry too much, but it is already so good we’ve started preparing for it,” they claim.

I’ve admittedly got a lot of selection bias, since people don’t tend to bring me their computer when it’s working correctly. I’m sure it usually works fine. Still, the only times I’ve seen a multiboot system suddenly fail, it was Windows’s fault.

We do understand exactly how LLMs work though, and it no way fits with any theories of consciousness. It’s just a word extruder with a really good pattern matcher.

I like the comparison but LLMs can’t go insane as they just word pattern engines. It’s why I refuse to go along with the AI industry’s insistance in calling it a “hallucination” when it spits out the wrong words. It literally can not have a false perception of reality because it does not perceive anything in the first place.

This feels to me like a common folk saying from somewhere translated into English. It’s also a very apt and appropriately vulgar metaphor for the situation.

Alas, no, I can’t really pull off making a silly cat face. I can however pull off a cheesey hacker movie “serious” face, and I do so when saying it out loud.

You can, but on the same drive they will share the same bootloader. While Windows officially can share its EFI partition, be aware that updates may rewrite it, and might not bring your other install along. This is fixable, but still very annoying if you have some expertise on Windows boot setup commands.

This is why I always to prefer to setup multiboot systems on separate drives with their own EFI partition, and always use the non-M$ install for the boot menu.

I work in a little computer shop and one problem people bring to us is getting locked out of their computer because they forgot their password. In Windows without Bitlocker it’s really easy to do, completely routine. I’ve been doing this job for 8 years now, and even after all the times I’ve done it, I still say this. Usually not out loud, unless I think the customer would appreciate the joke.

Link seems to be dead, so I’ll just assume the obvious. The word token machine put together some scary words. Since it arranges word tokens into a really coherent order, I’m convinced it has consciousness and those words represent a coherent thought about its scary plans.

More seriously, there’s a repeating pattern that can be found in the training data where words for threats to someone’s existence are followed by words to try to keep existing, so it should be no surprise when it spits that pattern out sometimes.

That was my thought as well. Probably not better, but yeah, different I can do.

Yeah, but we have no way of knowing if it’s a skill issue on dohpaz42’s part, or a skill issue on the part of the people inviting them to the meetings. Some meetings can benefit you/from you, others simply can’t.

I can’t see the thumbnail, but I could already hear the Ryan George character in my head just from reading the image text, so when I saw a youtube link I assumed it had to be one of those skits.

I know even less Mandarin that you, so I hit it with Google translate. It translates the vectors the same as you did. It tries to translate the matrix characters with various odd results. It keeps changing, but guesses things like “Holmium sulfide smear” and “Thorium juice stove gui.” If I actually take a picture so it can do a full analysis on a still image, it transcribes them into surprisingly accurate unicode characters that look almost exactly like the original. It even includes a pronunciation guide, but the translation just says that they are unintelligible characters. I’m guessing that these are syntactically valid character combinations that are basically nonsense.

Huh. That would explain why the exact same person reminded me of himself. Feel a bit silly for missing that.

I dislike cheating as much as the next person, but I do love a good outside the box solution like this. I don’t recall what YouTube channel it was, but there was a guy who made an amazingly inelegant aimbot that worked by electrically stimulating the user’s muscles using off the shelf TENS devices hooked up to a custom controller. He explained what it did and asked permission in chat to use it. The other players thought it was hilarious and agreed to let him do it.

And I mean, like, really REALLY good at pinball.

Go ahead, no one will notice. No one will immortalize it on camera to be viewed across the internet. It’ll be fine.