leftzero
@leftzero@lemmy.dbzer0.com
- Comment on How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%] 1 week ago:
Scam. We’re being sold an autocomplete tool as a search engine.
Or fraud, since some of the same companies destroyed the functionality of their search engines in order to make the autocomplete look better in comparison.
- Comment on Is there a game that combines Civ with Sims? 1 week ago:
Colonizing: definitely. Warlord: 100%, if that’s what you’re into. 20 cats: only the one, I believe; and a dog, if they can get along. But there are mods. Perpetually horny: oh, yeah. Practically the name of the game. Unless you become impotent, of course, or live in Byzantium and get on your liege’s bad side. Getting your house on fire: metaphorically…? Sure, constantly. Literally? I don’t think there’s an event for that, but there’s that one with the basement full of manure…
- Comment on Datacenters in space are a terrible, horrible, no good idea. 2 weeks ago:
Project Icarus it was called, the fourth space program of that name and the first for which it was appropriate. Long before Jacob’s parents were born—before the Overturn and the Covenant, before the Power Satellite League, before even the full flower of the old Bureaucracy—old grandfather NASA decided that it would be interesting to drop expendable probes into the Sun to see what happened.
They discovered that the probes did a quaint thing when they got close. They burned up.
In America’s “Indian Summer” nothing was thought impossible. Americans were building cities in space—a more durable probe couldn’t be much of a challenge!
Shells were made, with materials that could take unheard of stress and whose surfaces reflected almost anything. Magnetic fields guided the diffuse but tremendously hot plasmas of corona and chromosphere around and away from those hulls. Powerful communications lasers pierced the solar atmosphere with two-way streams of commands and data.
Still, the robot ships burned. However good the mirrors and insulation, however evenly the superconductors distributed heat, the laws of thermodynamics still held. heat will pass from a higher temperature to a zone where the temperature is lower, sooner or later.
The solar physicists might have gone on resignedly burning up probes in exchange for fleeting bursts of information had Tina Merchant not offered another way. “Why don’t you refrigerate?” she asked. “You have all the power you want. You can run refrigerators to push heat from one part of the probe to another.”
Her colleagues answered that, with superconductors, equalizing heat throughout was no problem.
“Who said anything about equalizing?” the Belle of Cambridge replied. “You should take all excess heat from the part of the ship were the instruments are and pump it into another part where the instruments aren’t.”
“And that part will burn up!” one colleague said. “Yes, but we can make a chain of these ‘heat dumps,’” said another engineer, slightly more bright. “And then we can drop them off, one by one …”
“No, no you don’t quite understand.” The triple Nobel Laureate strode to the chalkboard and drew a circle, then another circle within.
'Here!" She pointed to the inner circle. “You pump your heat into here until it is, for a short time, hotter than the ambient plasma outside of the ship. Then, before it can do harm there, you dump it out into the chromosphere.”
“And how,” asked a renowned physicist, “do you expect to do that?”
Tina Merchant had smiled as if she could almost see the Astronautics Prize held out to her. “Why I’m surprised at all of you!” she said. “You have onboard a communications laser with a brightness temperature of millions of degrees! Use it!”
Enter the age of the Solar Bathysphere. Floating in part by buoyancy and also by balancing atop the thrust of their refrigerator lasers, probes lingered for days, weeks, monitoring the subtle variations at the Sun, that wrought weather on the Earth.
— David Brin, Sundiver, 1980
Here’s an interesting discussion about the concept, with Brin himself explaining his reasoning.
- Comment on Datacenters in space are a terrible, horrible, no good idea. 2 weeks ago:
None of the moons in our solar system have atmospheres.
Except for Titan. Titan has a lot of atmosphere.
- Comment on AI tool OpenClaw wipes the inbox of Meta's AI Alignment director despite repeated commands to stop — executive had to manually terminate the AI to stop the bot from continuing to erase data 2 weeks ago:
Ah, doing your best to break the Therac-25’s record, I see.
- Comment on AIs can’t stop recommending nuclear strikes in war game simulations 3 weeks ago:
Story goes that Reagan got freaked out after watching the film and asked the chairman of the joint chiefs of staff if it’d be that easy to hack into the US military. After a week of looking into it came the answer: “no, the problem is much worse than that”, and fifteen months after having watched it signed the confidential directive “National Policy on Telecommunications and Automated Information Systems Security”, starting the implementation of cybersecurity measures in the country’s institutions.
- Comment on The U.S. spent $30 billion to ditch textbooks for laptops and tablets: The result is the first generation less cognitively capable than their parents 3 weeks ago:
More public money syphoned off to the parasitic corporations and dumber, easier to exploit proles.
Seems like a massive win for capitalism, really.
Until it all blows up on our faces, obviously, but when has capitalism ever cared about anything beyond the next quarter?
- Comment on The creator of systemd wants your entire system validated by SecureBoot 3 weeks ago:
Everything starts with good intentions.
No it doesn’t.
When it comes to privacy, politics, and capitalism, almost nothing starts with good intentions.
Most everything starts for the short term benefit of whoever starts it and any investors putting money into it, at the expense of everyone else and ignoring any future negative consequences unless profit can be extracted from them.
It hurting people the starter doesn’t like (even if it will come back to hurt the starter in the long time) is also a very important factor, though secondary to the short term profit one.
- Comment on The RAM shortage is coming for everything you care about 3 weeks ago:
Anyone remember RAMdisks?
- Comment on The green lean mean killing machine 3 weeks ago:
I wouldn’t trust books when it comes to hunting mushrooms.
Don’t go hunting mushrooms unless your family’s taken you hunting mushrooms since you were a kid (and even then only in regions you’re familiar with, and even then don’t pick any mushroom you’re not 110% certain of, and if you’re not an idiot), or accompanied by someone with that experience verifying every single mushroom you find before picking it up, and telling you why and how it’d’ve killed you or why it wouldn’t’ve tasted good.
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
Probably a lot of that in the data the model was trained on.
Garbage in, garbage out, as they say, especially when the machine is a rather inefficient garbage compactor.
- Comment on Dr. Oz pushes AI avatars as a fix for rural health care. Not so fast, critics say 4 weeks ago:
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
IMO they’re the same picture. In either case, the human enabling the bot’s actions should be blamed as if those were their own actions, regardless of their “intentions”.
Oh, definitely. It’s 100% the responsibility of the human behind the bot in either case.
But the second option is scarier, because there are a lot more ignorant idiots than malicious bastards.
If these unsupervised agents can be dangerous regardless of the intentions of the humans behind them, we should make the idiots using them aware that they’re playing with fire and they can get burnt, and burn other people in the process.
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
The point is that if predicting the next word leads to it setting up a website to attempt to character assassinate someone, that can have real world consequences, and cause serious harm.
Even if no one ever reads it, crawlers will pick it up, it will be added to other bots’ knowledge bases, and it will become very relevant when it pops up as fact when the victim is trying to get a job, or cross a border, or whatever.
And that’s just the beginning. As these agents get more and more complex (not smarter, of course, but able to access more tools) they’ll be able to affect the real world more and more. Access public cameras, hire real human people, make phone calls…
Depending on what word they randomly predict next, they’ll be able to accidentally do a lot of harm. And the idiots setting them up and letting them roam unsupervised don’t seem to realise that.
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
Transmetropolitan was truly prescient (except when it came to politics; turns out that if the president’s crimes and disregard for the constitution become public the press and the law don’t care, and just let him get on with it, making the whole point of the book moot).
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
Other bots that might be run by the company you’re trying to get a job in, the college you want to attend, the customs agent at the airport, the online shop you’re trying to buy from, the social network you’re trying to join…
These dystopian days a hit piece can do a lot of harm, even if no human ever reads it…
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
From what I read it was closed because it was tagged as a “good first issue”, which in that project are specifically stated to be a means to test new contributors on non-urgent issues that the existing contributors could easily solve, and which specifically prohibits generated code from being used (as it would make the whole point moot).
The agent completely ignored that, since it’s set up to push pull requests and doesn’t have the capability to comprehend context, or anything, for that matter, so the pull request was legitimately closed the instant the repository’s administrators realised it was generated code.
The quality (or lack thereof) of the code never even entered the question until the bot brought it up. It broke the rules, its pull request was closed because of that, and it went on to attempt to character assassinate the main developer.
It remains an open question whether it was set up to do that, or, more probably, did it by itself because the Markov chain came up with the wrong token.
And that’s the main point: unsupervised LLM-driven agents are dangerous, and we should be doing something about that danger.
- Comment on An AI Agent Published a Hit Piece on Me 4 weeks ago:
The point is that there was no one at the wheel. Someone set the agent up, set it loose to do whatever the stochastic parrot told it to do, and kind of forgot about it.
Sure, if you put a brick on your car’s gas pedal and let it run down the street and it runs someone over it’s obviously your responsibility, and this is exactly the same case, but the idiots setting these agents up don’t realise that it’s the same case.
Some day one of these runaway unsupervised agents will manage to get on the dark web, hire a hitman, and get someone killed, because the LLM driving it will have pulled the words from some thriller in its training data, obviously without realising what they mean or the consequences of its actions, because those aren’t things a LLM is capable of, and the brainrotten idiot who set the agent up will be all like, wait, why are you blaming me, I didn’t tell it to do that, and some jury will have to deal with that shit.
The point of the article is that we should deal with that shit, and prevent it from happening if possible, before it inevitably happens.
- Comment on DoorDashers are getting paid to close Waymo's self-driving car doors 4 weeks ago:
Well, daemons, but still…
- Comment on DoorDashers are getting paid to close Waymo's self-driving car doors 4 weeks ago:
Tesla isn’t a good example of how anything is done in the industry, but rather of how not to do it.
I don’t doubt Tesla’s doors are designed to mutilate their customers, much like they’re designed to trap them inside in the event of a fire, probably because Musk’s ketamine addled brain though it’d be funny, but that’s not how any other company would do it, even if only because because lawsuits cost money and weaker locks are cheaper.
- Comment on Europe’s $24 Trillion Breakup With Visa and Mastercard Has Begun 5 weeks ago:
Yeah, but European unfathomable, or British unfathomable?
- Comment on Chatbots Make Terrible Doctors, New Study Finds 5 weeks ago:
LLMs don’t have the mind of a five year old, though.
They don’t have a mind at all.
They simply string words together according to statistical likelihood, without having any notion of what the words mean, or what words or meaning are; they don’t have any mechanism with which to have a notion.
They aren’t any more intelligent than old Markov chains (or than your average rock), they’re simply better at producing random text that looks like it could have been written by a human.
- Comment on Western Digital details 14-platter 3.5-inch HAMR HDD designs with 140 TB and beyond 5 weeks ago:
In a pinch the drive can also double as a flywheel battery.
- Comment on xkcd #3204: Dinosaurs And Non-Dinosaurs 5 weeks ago:
Well, reptiles seem to have split pretty early on between the ancestors of lizards and snakes (and the lonely tuatara)), and the ancestors of turtles, crocodiles, and dinosaurs, the main differences seemingly being in the bones of the skull, and specifically for the group with the snakes and lizards in, the ability to self-amputate the tail (though that’s lost in many of their descendants), and the keratinized scales; you won’t see a turtle, crocodile, or dinosaur melting its whole skin in one go like lizards and snakes do, they’ll molt their scales (or feathers, or scutes) one at a time.
- Comment on xkcd #3204: Dinosaurs And Non-Dinosaurs 5 weeks ago:
True, (some) snakes have also evolved specialized fangs.
Several times independently with significantly different designs, it seems.
- Comment on Neocities founder stuck in chatbot hell after Bing blocked 1.5 million sites 5 weeks ago:
Anyone who uses duckduckgo, since that’s where it gets its results from.
- Comment on xkcd #3204: Dinosaurs And Non-Dinosaurs 5 weeks ago:
That one I’m not entirely sure about, but it seems that, in the same way being a mammal (from a bone perspective) is all about the teeth and inner ear, being a dinosaur is all about the hips (dinosaurs have an upright stance, with the legs under their bodies; even with the quadruped ones you can see how they’re really something evolved to walk on its hind legs walking on its hands and feet), and pterosaurs and their non-dinosaur ancestors just don’t have the right kind of hip.
It’s a bit muddy, though. Once you get into archosaurs and before you get into more specialised things like crocodiles, dinosaurs, or pterosaurs it’s mostly “this thing seems to be more closely related to this group than to this other group, so we’ll throw it in with them even if it doesn’t really look anything like them”.
There’s a small bipedal reptile, for instance, scleromuchlus, that’s been bundled up with pterosaurs because it apparently seems more related to them, even though if you look at an artist’s representation you’d assume it must be a dinosaur, but might in fact not fit in either group and be instead just a basal avemetatarsalian (or maybe even lower in the tree) with no other identified close relatives.
- Comment on xkcd #3204: Dinosaurs And Non-Dinosaurs 5 weeks ago:
The name (dimetrodon ≈ two teeth sizes) is also a clue, as teeth specialization is very much a synapsid (i.e., mammal and proto-mammal) thing.
- Comment on How do you communicate "sorry, my bad" when you make a mistake while driving? 5 weeks ago:
- Comment on Is it theoretically possible Trump and ICE are killing a very large number of immigrants (like 25% of those detained) and no one knows? 5 weeks ago:
Death flights have quite a long history, sadly, and were a favourite of far right South American dictatorships promoted by the USA…
