I’m so confused that the same people can say “why does everyone get their undies in a bunch that we happily accept putting arbitrary data in columns regardless of type, that’s good, it’s flexible, but fine, we’ll put in a ‘strict’ keyword if you really want column types to mean something” and also “every other SQL says 1==‘1’ but this is madness, strings aren’t integers, what is everyone else thinking?!”

Back in the olden days, if you wrote a program, you were punching machine codes into a punch card and they were being fed into the computer and sent directly to the CPU. The machine was effectively yours while your program ran, then you (or more likely, someone who worked for your company or university) noted your final results, things would be reset, and the next stack of cards would go in.

Once computers got fast enough, though, it was possible to have a program replace the computer operator, an “operating system”, and it could even interleave execution of programs to basically run more than one at the same time. However, now the programs had to share resources, they couldn’t just have the whole computer to themselves. The OS helped manage that, a program now had to ask for memory and the OS would track what was free and what was in use, as well as interleaving programs to take turns running on the CPU. But if a program messed up and wrote to memory that didn’t belong to it, it could screw up someone else’s execution and bring the whole thing crashing down. And in some systems, programs were given a turn to run and then were supposed to return control to the OS after a bit, but it was basically an honor system, and the problem with that is likely clear.

Hardware and OS software added features to enforce more order. OSes got more power, and help from the hardware to wield it. Now instead of asking politely to give back control, the hardware would enforce limits, forcing control back to the OS periodically. And when it came to memory, the OS no longer handed out addresses matching the RAM for the program to use directly, instead it could hand out virtual addresses, with the OS tracking every relationship between the virtual address and the real location of the data, and the hardware providing Memory Management Units that can do things like store tables and do the translation from virtual to physical on its own, and return control to the OS if it doesn’t know.

This allows things like swapping, where a part of memory that isn’t being used can be taken out of RAM and written to disk instead. If the program tries to read an address that was swapped out, the hardware catches that it’s a virtual address that it doesn’t have a mapping for, wrenches control from the program, and instead runs the code that the OS registered for handling memory. The OS can see that this address has been swapped out, swap it back in to real RAM, tell the hardware where it now is, and then control returns to the program. The program’s none the wiser that its data wasn’t there a moment ago, and it all works. If a program messes up and tries to write to an address it doesn’t have, it doesn’t go through because there’s no mapping to a physical address, and the OS can instead tell the program “you have done very bad and unless you were prepared for this, you should probably end yourself” without any harm to others.

Memory is handed out to programs in chunks called “pages”, and the hardware has support for certain page size(s). How big they should be is a matter of tradeoffs; since pages are indivisible, pages that are too big will result in a lot of wasted space (if a program needs 1025 bytes on a 1024-byte page size system, it’ll need 2 pages even though that second page is going to be almost entirely empty), but lots of small pages mean the translation tables have to be bigger to track where everything is, resulting in more overhead.

This is starting to reach the edges of my knowledge, but I believe what this is describing is that RISC-V chips and ARM chips have the ability for the OS to say to the hardware “let’s use bigger pages than normal, up to 64k”, and the Linux kernel is getting enhancements to actually use this functionality, which can come with performance improvements. The MMU can store fewer entries and rely on the OS less, doing more work directly, for example.

A VPN is just a way to say “wrap up my normal internet packets and ship them somewhere specific before they continue the normal way.” The normal way is you want to get a message to some other server, and as a part of setting up the network you’re on, your machine should already have a list of other devices it’s physically connected to (“physically” could be “via radio waves” so not just wired) and they should have already advertised “hey, I’ve got access to these places too” for your information. Your router is likely the only one in your home network advertising anything that is on the larger internet, so all your outgoing messages will have to go that way to get to their destination. For example, I’ve got a phone, a wifi access point, a router, and my ISP’s box; my phone knows the WiFi access point is two hops away from internet because the access point said so, that’s the best one it can see, so it sends it that way and hopes it makes it. Each machine in between does the same thing until hopefully it gets where it is supposed to.

With a VPN, the same messages are wrapped in a second message that is addressed to the other end of the VPN. When it gets to the VPN provider, it’s unwrapped, then the inside message is sent off to wherever it’s supposed to go. If a message comes back to the VPN provider addressed to you (ish, this is simplifying a bit), it’s wrapped up the same way and sent back to you.

Big companies often put resources “behind” the VPN, so you can’t send messages from the outside addresses to the office printer, they’ll get blocked, but you can request a connection to the VPN, and messages that come in through that path do get allowed. The VPN can be one central place where you make sure everything coming in is allowed, then on the other side the security can be a little less tight.

VPNs also encrypt the internal message as a part of wrapping them up, which means that if you’re torrenting via a VPN, all anyone else can see is a message addressed to your VPN provider and then an encrypted message inside. And anyone you were exchanging messages with only ever saw traffic to and from the VPN provider, they never saw where it was going after your VPN provider got it. Only you and the VPN provider know what was happening on both ends, and hopefully they don’t look too closely or keep records.

Hopefully now it’s clear that Mullvad and similar won’t help you access your own things from outside, they’re only good for routing your stuff through them and then out into the rest of the internet. However, this isn’t secret magic tech: you can run your own VPN that goes in the other direction, allowing you into your own home network and then able to connect to things as if you were physically there. Tailscale is probably the easiest thing for things like that nowadays, it’ll set up a whole system where your devices can find each other and set up a mesh of secure, direct connections no matter where they are physically located. By default, just the direct device-to-device connections are re-routed, but you can also make a device an “exit node” that can route all your traffic like a traditional VPN.

Of course, that will be the exact opposite of what you want for privacy while torrenting, as it’s all devices that you clearly own and not hiding their identities whatsoever. But it’s very cool for home networking and self-hosting stuff.

Bluesky’s more like an aspirationally decentralized platform, you can keep your own data on your own server and use your own domain name as a user name, but most of the rest of it is “centralized, but we’re designing it in such a way that we can open it up later.” Even then, though, it’s heavily influenced by the original idea of “let’s make something decentralized that Twitter can switch to once it’s worked out” which means that even when they do open things up, it’s likely that a lot of Bluesky will only be practical at “big tech company scale” to run yourself, whereas Mastodon or Lemmy you can just spin up on a server and it’ll be fine until you get a lot of users.

I as a human being have grown up and learned from experience and the experiences of previous humans that were documented or directly communicated to me. I can see no inherent difference with an artificial intelligence learning on the same data.

It’s a massive difference in scale. For one, before you even leave the womb you have millions of years of evolution shaping the initial structure of your brain. Then your “training” begins, but it’s infinitely richer than anything we’re giving to these LLMs. Sights, sounds, smells, feelings, so many that part of what your brain is learning is what it must ignore. You’re also benefitting from the interactivity of your environment, you can experiment with things and get feedback for what happens. As you get older and develop more skills, you can start integrating them together to do even more complex things, and the people around you will use their own incredible intelligence to specifically tailor your training to what you need as you learn and grow.

Meanwhile, an LLM is getting fed words, and learning how to predict the next word. It’s a pale shadow of the complex lives humans live. Words are one of the more powerful things we have for thinking and reasoning, so if you’re going to go all in on one skill, it’s a rich environment for learning and in theory the contents of all of humanity’s writing probably contains all the information necessary to recreate human intelligence, but our current technology doesn’t even come close to wringing every ounce of knowledge from the training sets.

Archive Team looked at this about 10 years ago and found it basically impossible. It was around 14 petabytes of information to fetch, organize, and distribute at the time.

https://wiki.archiveteam.org/index.php/INTERNETARCHIVE.BAK

Do you think electing Trump will be read as “wow, the US is taking a principled stance on Palestinian rights” by the world?

That’s not how it works at all. If it were as easy as adding a line of code that says “check for integrity” they would’ve done that already. Fundamentally, the way these models all work is you give them some text and they try to guess the next word. It’s ultra autocomplete. If you feed it “I’m going to the grocery store to get some” then it’ll respond “food: 32%, bread: 15%, milk: 13%” and so on.

They get these results by crunching a ton of numbers, and those numbers, called a model, were tuned by training. During training, they collect every scrap of human text they can get their hands on, feed bits of it to the model, then see what the model guesses. They compare the model’s guess to the actual text, tweak the numbers slightly to make the model more likely to give the right answer and less likely to give the wrong answers, then do it again with more text. The tweaking is an automated process, just feeding the model as much text as possible, until eventually it gets shockingly good at predicting. When training is done, the numbers stop getting tweaked, and it will give the same answer to the same prompt every time.

Once you have the model, you can use it to generate responses. Feed it something like “Question: why is the sky blue? Answer:” and if the model has gotten even remotely good at its job of predicting words, the next word should be the start of an answer to the question. Maybe the top prediction is “The”. Well, that’s not much, but you can tack one of the model’s predicted words to the end and do it again. “Question: why is the sky blue? Answer: The” and see what it predicts. Keep repeating until you decide you have enough words, or maybe you’ve trained the model to also be able to predict “end of response” and use that to decide when to stop. You can play with this process, for example, making it more or less random. If you always take the top prediction you’ll get perfectly consistent answers to the same prompt every time, but they’ll be predictable and boring. You can instead pick based on the probabilities you get back from the model and get more variety. You can “increase the temperature” of that and intentionally choose unlikely answers more often than the model expects, which will make the response more varied but will eventually devolve into nonsense if you crank it up too high. Etc, etc. That’s why even though the model is unchanging and gives the same word probabilities to the same input, you can get different answers in the text it gives back.

Note that there’s nothing in here about accuracy, or sources, or thinking, or hallucinations, anything. The model doesn’t know whether it’s saying things that are real or fiction. It’s literally a gigantic unchanging matrix of numbers. It’s not even really “saying” things at all. It’s just tossing out possible words, something else is picking from that list, and then the result is being fed back in for more words. To be clear, it’s really good at this job, and can do some eerily human things, like mixing two concepts together, in a way that computers have never been able to do before. But it was never trained to reason, it wasn’t trained to recognize that it’s saying something untrue, or that it has little knowledge of a subject, or that it is saying something dangerous. It was trained to predict words.

At best, what they do with these things is prepend your questions with instructions, trying to guide the model to respond a certain way. So you’ll type in “how do I make my own fireworks?” but the model will be given “You are a chatbot AI. You are polite and helpful, but you do not give dangerous advice. The user’s question is: how do I make my own fireworks? Your answer:” and hopefully the instructions make the most likely answer something like “that’s dangerous, I’m not discussing it.” It’s still not really thinking, though.

This, and see also “minmaxing,” the process of optimizing something (usually your character in a game) to get minimum penalty and/or maximum benefit, usually ignoring anything like realism or storytelling and focusing entirely on the stats and numbers.

The .bin and .cue file are the parts of the actual game disc that you want. The .bin file contains almost all of the data and the .cue file contains some extra information about the structure of the CD. All the rest is Internet Archive stuff (and an image of the game cover of course).

To open it, you can convert it to a .iso disk image instead, which any Linux distribution can open as if it were a real CD. This blog post talks about how to do that. The last paragraph about mount you can probably replace with double-clicking the .iso file in the GUI I would guess.

No. The headsets are disabled when the play starts or when the play clock goes below 15 seconds.

If you ask an LLM to help you with a legal brief, it’ll come up with a bunch of stuff for you, and some of it might even be right. But it’ll very likely do things like make up a case that doesn’t exist, or misrepresent a real case, and as has happened multiple times now, if you submit that work to a judge without a real lawyer checking it first, you’re going to have a bad time.

There’s a reason LLMs make stuff up like that, and it’s because they have been very, very narrowly trained when compared to a human. The training process is almost entirely getting good at predicting what words follow what other words, but humans get that and so much more. Babies aren’t just associating the sounds they hear, they’re also associating the things they see, the things they feel, and the signals their body is sending them. Babies are highly motivated to learn and predict the behavior of the humans around them, and as they get older and more advanced, they get rewarded for creating accurate models of the mental state of others, mastering abstract concepts, and doing things like make art or sing songs. Their brains are many times bigger than even the biggest LLM, their initial state has been primed for success by millions of years of evolution, and the training set is every moment of human life.

LLMs aren’t nearly at that level. That’s not to say what they do isn’t impressive, because it really is. They can also synthesize unrelated concepts together in a stunningly human way, even things that they’ve never been trained on specifically. They’ve picked up a lot of surprising nuance just from the text they’ve been fed, and it’s convincing enough to think that something magical is going on. But ultimately, they’ve been optimized to predict words, and that’s what they’re good at, and although they’ve clearly developed some impressive skills to accomplish that task, it’s not even close to human level. They spit out a bunch of nonsense when what they should be saying is “I have no idea how to write a legal document, you need a lawyer for that”, but that would require them to have a sense of their own capabilities, a sense of what they know and why they know it and where it all came from, knowledge of the consequences of their actions and a desire to avoid causing harm, and they don’t have that. And how could they? Their training didn’t include any of that, it was mostly about words.

One of the reasons LLMs seem so impressive is that human words are a reflection of the rich inner life of the person you’re talking to. You say something to a person, and your ideas are broken down and manipulated in an abstract manner in their head, then turned back into words forming a response which they say back to you. LLMs are piggybacking off of that a bit, by getting good at mimicking language they are able to hide that their heads are relatively empty. Spitting out a statistically likely answer to the question “as an AI, do you want to take over the world?” is very different from considering the ideas, forming an opinion about them, and responding with that opinion. LLMs aren’t just doing statistics, but you don’t have to go too far down that spectrum before the answers start seeming thoughtful.

In its complaint, The New York Times alleges that because the AI tools have been trained on its content, they sometimes provide verbatim copies of sections of Times reports.

OpenAI said in its response Monday that so-called “regurgitation” is a “rare bug,” the occurrence of which it is working to reduce.

“We also expect our users to act responsibly; intentionally manipulating our models to regurgitate is not an appropriate use of our technology and is against our terms of use,” OpenAI said.

The tech company also accused The Times of “intentionally” manipulating ChatGPT or cherry-picking the copycat examples it detailed in its complaint.

https://www.cnn.com/2024/01/08/tech/openai-responds-new-york-times-copyright-lawsuit/index.html

The thing is, it doesn’t really matter if you have to “manipulate” ChatGPT into spitting out training material word-for-word, the fact that it’s possible at all is proof that, intentionally or not, that material has been encoded into the model itself. That might still be fair use, but it’s a lot weaker than the original argument, which was that nothing of the original material really remains after training, it’s all synthesized and blended with everything else to create something entirely new that doesn’t replicate the original.

These models aren’t great at tasks that require precision and analytical thinking. They’re trained on a fairly simple task, “if I give you some text, guess what the next bit of text is.” Sounds simple, but it’s incredibly powerful. Imagine if you could correctly guess the next bit of text for the sentence “The answer to the ultimate question of life, the universe, and everything is” or “The solution to the problems in the Middle East is”.

Recently, we’ve been seeing shockingly good results from models that do this task. They can synthesize unrelated subjects, and hold coherent conversations that sound very human. However, despite doing some things that up until recently only humans could do, they still aren’t at human-level intelligence. Humans read and write by taking in words, converting them into rich mental concepts, applying thoughts, feelings, and reasoning to them, and then converting the resulting concepts back into words to communicate with others. LLMs arguably might be doing some of this too, but they’re evaluated solely on words and therefore much more of their “thought process” is based on “what words are likely to come next” and not “is this concept being applied correctly” or “is this factual information”. Humans have much, much greater capacity than these models, and we live complex lives that act as an incredibly comprehensive training process. These models are small and trained very narrowly in comparison. Their excellent mimicry gives the illusion of a similarly rich inner life, but it’s mostly imitation.

All that comes down to the fact that these models aren’t great at complex reasoning and precise details. They’re just not trained for it. They got through “life” by picking plausible words and that’s mostly what they’ll continue to do. For writing a novel or poem, that’s good enough, but math and physics are more rigorous than that. They do seem to be able to handle code snippets now, mostly, which is progress, but in general this isn’t something that you can be completely confident in them doing correctly. They make silly mistakes because they aren’t really thinking it through. To them, there isn’t really much difference between answers like “that date is 7 days after Christmas” and “that date is 12 days after Christmas.” Which one it thinks is more correct is based on things it has seen, not necessarily an explicit counting process. You can also see this in things like that case where someone tried to use it to write a legal brief, where it came up with citations that seemed plausible but were in fact completely made up. It wasn’t trained on accurate citations, it was trained on words.

They also have a bad habit of sounding confident no matter what they’re saying, which makes it hard to use them for things you can’t check yourself. Anything they say could be right/accurate/good/not plagiarized, but the model won’t have a good sense of that, and if you don’t know either, you’re opening yourself up to risk of being misled.

Protect me from knowing what I don’t need to know. Protect me from even knowing that there are things to know that I don’t know. Protect me from knowing that I decided not to know about the things that I decided not to know about. Amen.

Lord, lord, lord. Protect me from the consequences of the above prayer. Amen.

That’s part of the point, you aren’t necessarily supposed to have an empty mind the whole time. I mean, if you can do that, great, but you aren’t failing if that’s not the case.

Imagine that your thoughts are buses, and your job is to sit at the bus stop and not get on any of them. Just notice them and let them go by. Like a bus stop, you don’t really control what comes by, but you do control which ones you get on board and follow. If you notice that you’ve gotten on a bus, that’s fine, just get off of it and go back to watching. Interesting things can happen if you just watch and notice which thoughts go by, and it’s good practice for noticing what you’re thinking and where you’re going and taking control of it yourself when it’s somewhere you don’t want to go.