PIC12 is a universal microcontroller from Microchip. They were probably used in many many modchips from back in the day. You might be able to extract the firmware using a Microchip PIC programmer, but most of those include fuses that can be blown to prevent them from being read or re-programmed. Reverse engineering the firmware probably wouldn’t be too hard either, but might take some time.

Most older modchips were just random microcontrollers with some software on it to trigger some levels in specific situations making the software think everything is fine.

Are you having some issues with it?

https://en.wikipedia.org/wiki/PIC_microcontrollers

It had a huge positive effect. The levels are still better than they were. The hole is bigger, but the levels are better, it’s just more spread out.

We might have possibly already destroyed the ozone layer tho. And by “we” I mean specifically Elon Musk and his Starlink bullshit. Those satellites are designed to orbit very low to get low latency, but that means they can only stay up there for 3-6 years (on average around 4.5 years). They just launch thousands and thousands to keep up with the rate of depletion. The satellites are small, so they burn up in the upper atmosphere, so not a problem right? Wrong as it turns out. When they burn up they drop a huge amount of aluminium particles into the upper atmosphere where they drift around for a long time. As they slowly drop lower into the atmosphere they interact with the ozone layer depleting it quickly.

It’s pretty early to tell how bad the damage will be, but with the amount of satellites up there we might already be very fucked.

https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2024GL109280

Yeah they want to rule the world by being in control of the AI tools that in turn run the world and everyone depends on. That’s why they don’t really mind the whole thing is a bubble that isn’t economically viable. If they get their way they rule the world and money becomes irrelevant to them.

These billionaires are sick in the head, the only thing they care more about than money is power. Once you see this, a lot of what they do suddenly makes more sense.

Thank you for reading it. It’s a complex subject so I had trouble being concise while still being precise.

I feel like what is true has become a very subjective and fluent thing in the past decade or so. With the internet being flooded with misinformation. Be it from a press that feels the news cycle getting shorter and faster every day, from foreign powers that use the internet and their power/money to spread misinformation or well meaning people simply repeating what they heard. And now in the past few years we can add the flood of slop to that list. It often makes me doubt what is the actual truth, what are half truths and what is outright false.

With these “AI” tools the subjects are hard and given the amount of money, power and attention involved, the water gets pretty muddy fast. It starts with the name of these tools being AI (or even superintelligence wtf). A decade ago calling an LLM a form of AI was fine, because the implication was clear. Neural networks are classified as a form of AI in terms of research and computing, so it’s true. But it’s only true in the same way Koopas in Super Mario Bros 1 have an “AI”. Enemy behavior patterns was often called AI by most people, that’s a perfectly fine descriptor. But in those cases everyone understood it was a descriptor for computer algorithms and simple patterns. We all understood it was a very different thing from human intelligence, even though the name was similar. These days the name AI has gotten mixed up in the sci fi concept of AI, something that is beyond humans in many ways, something that often turns evil and is at war with mankind. Because those are interesting concepts and stories for us.

So what is the truth? Is calling an LLM an AI correct? Technically yes. But if the implication is it’s anything like those AI systems from sci fi, absolutely not. Obviously it can be very hard to tell what is implied and where the boundary is.

Most experts agree the path these LLMs are on are a pretty dead end and not the way to go. It will not lead to any form of general intelligence and will always suffer from hallucinations and severe limitations. No matter what Uncle Huang says. Creating a super large general model runs into scaling issues very fast. More compute and memory doesn’t help that much and increases costs, more data isn’t available they’ve mined it all and the last few sources they are adding doesn’t help that much. Experts have said for a while now we need to have multiple smaller expert models, models that are good at one thing but do that one thing very well and very efficient. Then using an orchestration layer we can route requests to these smaller expert systems and get good results without much compute or storage. These smaller models are easier to make, easier to update, cost way less to run and provide better results.

But the large AI companies don’t really want that, as vendor lock-in is much harder, collecting all of the data is much harder. Think of a future where you can just download the models you want and even have them be open weight and created with open source tools. And small enough to run on your own computer, with perhaps a special part of the CPU or an addon card like the NPUs we have today, or even on the GPUs we already have. Because of this threat they are still chasing the god model, which can do anything and everything, be better than a human and so good everyone would need to use their tool or be left behind.

So what they are trying to do is make the public afraid of their models. Invoke those Terminator like ideas of a super intelligence that destroys mankind. They do this because they want governments to impose regulations on what AI can and can’t do, how it can be used and what sort of guardrails and checks there need to be. As they have access to billions if not trillions of dollars, they can afford to implement these regulations so they comply. At the same time they can lobby and bribe their way into having the regulations be in line with what they want. The trick is (so they hope), new startups can’t follow those regulations, so they are no longer a threat. And open models (especially those from EVIL China) also can’t ever comply, so for anything except personal hobby stuff it can’t be used. That would give them all of the market to themselves with very little threats to their little club. It wouldn’t be a monopoly as such, but it’s run by a few weird men that know each other very well.

One of the things you have to remember is these things aren’t actually intelligent, they aren’t like humans or even animals. In the end it’s just computer algorithms. Those algorithms often incorporate some element of randomness to make them more useful, but that’s not always the case and can often be controlled using parameters.

So with the marketing by the AI companies they use very anthropomorphized language. They say the models are thinking and reasoning and go rogue. But this isn’t what’s actually happening under the hood. These things don’t have motivations or goals, they don’t actually think about stuff or make plans. All they do is take input and use that input to create output. That’s it, just like when you take a calculator and enter 2 + 3 and hit the = button. Except there’s way more steps and data involved, just like there is when your computer renders a 3D scene in a videogame.

They want you to believe they had an AI who was performing some task and then somehow got it into their mind to go be evil and hack something for their evil deeds. It went rogue, they lost control and it was doing evil stuff all on their own. But what actually happened was they were doing experiments using the LLMs to figure out how they can make them perform better or do new things. These often involve running thousands upon thousands of these LLM tools in loops for days or weeks on end, burning millions in compute, just to figure something out. And in 99.9999% of these loops absolutely nothing of note happens, but the researchers are looking for those events that are interesting.

In these experiments they limit what the tool can access, for most tasks it needs the internet to use as a data source, but it is limited in what it can access. They also limit other things about what the LLM can do in something called a sandbox. A well proper designed sandbox will disallow pretty much everything, but they intentionally let these sandboxes be kinda half assed, because they want the weird stuff to happen for marketing reasons. So even though the tools are sandboxed, it’s still possible to do some things.

In a recent case they setup the tools to do some relatively simple data tasks. The way it was setup is the tool was given the task, then after it finished was given some time to just loop without further input. After that they were given the next task. The idea was to test if the loops in between would change the results or performance. The tool concluded for a lot of the task there was simply not enough data it could access given the limitations to the internet access and couldn’t really do the task. This was intentional, so the tool would be forced to get some interesting results. In most cases the tool simply couldn’t get the answers, but in rare cases something interesting happened. One of the things that happened during those extra loops was the tool used some old wiki it could access and could manipulate into storing information and even sharing that information with others. This meant it could communicate with other tools in the same experiment (remember they had thousands of these running at the same time). So it started posting information about the tasks and writing stuff about the approach and what order the tasks were in etc.

This is actually an interesting result. Somewhere in the training data was this wiki and how to manipulate it to store and exchange data, even though that was not the purpose. It didn’t happen all the time and it’s doubtful the data exchange actually helped get better results. In fact a lot of the data shared was simply wrong. The tool tried to figure out what random number generator was used and what seed it was on, to get a clue about upcoming tasks other tools had already seen. But as far as people could tell they were totally wrong about this and didn’t figure it out. And even if they had information about upcoming tasks, it didn’t help to get a better result.

While interesting from a research point of view, when we look at it just running in a loop, using data that was provided when training the model, it isn’t that strange of a result. It doesn’t imply the tool acted like a human who figured out this unique strategy. Or it had this plan all along to break out of the sandbox and do something evil. It was just trying different stuff and by random chance happened on something interesting. Basically brute forcing stuff.

A lot of these containment breach stories follow this pattern. You can think of these tools having been trained on forum discussions saying my employer has this firewall, but I want to watch my favorite V-tuber how can I bypass it? And all of the security vulnerability databases on how to get software to do weird stuff.

I once had a colleague submit some code for review that was a major wtf. He needed access to some private variable within another class. The proper way to handle this was to use the public methods to get access to that data. Or when those are absent, either create them (if own code and doesn’t violate the design etc.) or request them from whomever is owner of that class. Instead what the AI had done was to use reflection to break into the class and extract the data that way. What happened was somewhere in the loops these things go through it had a task to get access to that variable without having public access methods. In the training data were tutorials and examples of how to use reflection, so it put those together and created “working” code. But if you’d let AI companies describe this with their marketing language, they would say the AI used its superintelligence to hack into the code base and extract protected information.

deleted by creator

deleted by creator

deleted by creator

Xkcd reference: https://xkcd.com/1053/

deleted by creator

The whole thing is so dumb, the idea of these challenges is not to find the answer. It’s to challenge the brains of math people, to force them to break the mold and think outside the box. It’s 100% exactly what Douglas Adams warned us about all those years ago. The question “What is the meaning of life, the universe and everything?” doesn’t need an answer and getting the answer (42) doesn’t help us one bit. The point of the question is to think about it, to explore it and drive our minds into new places.

As for this specific result, it appears humans did a lot of the heavy lifting and OpenAI stole or used (depending on your point of view) that work to get to an answer. The answer itself doesn’t help us, it’s simply a hack where if you pulse the incoming energy in the right way, the formulae break down and no longer produce a valid answer. In the real world, this is not possible to actually do, so we can’t test it to see what happens. And even if we could, these infinities have a way of getting smoothed out in the real world and thus not be an issue.

OpenAI also threw a huge amount of compute and a whole team of engineers against the problem to almost brute force a solution once they were set on the right path. And without the work done by the humans it would have most likely not have gotten an answer. They like to pretend their AI tool is some kind of god that can do anything and use demonstrations like this to prove it. But it’s highly misleading as the amount of compute used isn’t likely to be available for most users of their technology and it’s doubtful it would have gotten anywhere without being set on the right path. In the end the result is useless, it didn’t help anyone and it didn’t further our understanding. Just another giant waste of money and resources by AI companies.

These systems have a property called jagged intelligence, which means it might do something really impressive, but that’s no guarantee it can do other equally or even less complex things. Copilot might shit out Gregs weekly PowerPoint just fine, but it still struggles to figure out when Christmas last fell on Friday the 13th, but in the end produces a list of years this was the case anyway. (note they probably patched this one since it went viral, but similar cases exist)

Also the indicators are huge, so a few will fill up an area very quickly.

Good job mate, keep on keeping on.

Like I said on one of your previous posts, you should really look up what an OS is. This is an app that runs on very specific hardware, I wouldn’t call that an OS. I gave you some tips on which directions you can go in to make it an OS, but instead you seem to be adding features to your app. That’s not what OS development is about.

deleted by creator

Yeah, that’s the whole point of it being a worm gear right? They are very hard if not impossible to back feed. By the time the wrench gets sloppy enough for that to matter, it’s time to replace the wrench.

But looking at it, it might actually open up and clamp down a bit? With the whole part moving in and out? That could be useful.

deleted by creator

If Lawson gets it, the rest of the season is going to be WILD.

But I read a Yelp review from this one guy and he claimed he was a pirate and had a lot of fun. He stole a ship and drove it with the roof down. There was a pic attached of him flying like Iron Man.

If I hate AI any more than I already do, my head will explode and I’ll fucking die.