Intel and AMD CPUs seem to do fine, and still are better than Apple Silicon in performance, without needing to resort to any "unified memory" architecture. And the newer replaceable / upgradeable CAMM2 (Compression Attached Memory Module) RAM is a good compromise to it with even more RAM bandwidth, and lower power requirements. (Sure, they may not be as fast as on-package DRAM or HBMs, but as everyone is now finding out, sacrificing our computing freedom to the whims and mercies of corporates is not in our best interest).
Was this status indicator simple vertical bars that were black and white and displayed on an old CRT Monitor?
Almost looks like black and white terminal vertical bars (like Alsamixer)?
There are chemical reactions that LLMs lack because they do NOT have a biology. They are just mathematical weights. Now, let’s say we can truly make sentient AI in the future but it does not have the biology that we do. Then it would just be a totally literal, emotion lacking, sentient machine.
But then what is sentience? Dogs are sentient to a certain extent and so are dolphins…
When I use a urinal I do not aim directly on front but instead to the curved lower part. That helps guide my urine down; virtually eliminating all splash back!
Maybe males are marginally better at geometry due to this additional hands on experience with angles and parabolas. We practice and study ballistic trajectories daily.
Why are we putting so much weight (no pun intended) on AI companies. At the end of the day the scaled up LLM transformers lack emotion and will… They do as they are told; or more correctly put. They do as they are programmed to do so.
Sounds like you are thinking they just need Asimov’s laws. But I think the point is, this can easily be weaponized by somebody with the willpower to do so.
No, just that “What if I just cheat to get the treat” is also something my lovable Boxer does when I hide rewards as well.
I suppose her energy is limited but even if infinite, and with caution, I estimate the chances of a mass extinction event due to Pickle taking over the world at < 1%.
That isn't the correct context. The code running the llm is well understood and the llm is simply the result of that code being executed. It is still a computer doing what it is told. It's just that we told it to use an incredibly large number of probabilities to calculate what series of tokens would have most likely come next after a given series of tokens. There is no hallucination or lie or rogue actions. There's just a program using math to generate tokens in response to other tokens.
You’re just a bunch of molecules following the laws of physics. It’s all just physics and chemistry, and those are well understood. Now explain the causes of World War I using chemistry and physics. Simple, right?
It isn't hard to program a gpt. You can do it in a weekend with a few hundred lines of python. The code is pretty simple. The math is not particularly high level.
The complexity and scale with LLMs come from the amount of training data used, not some kind of black magic in the programming.
> It’s all just physics and chemistry, and those are well understood.
Not really. We cannot model physics and chemistry to a level which allows us to accurately predict a humans action (even a tiny time-step into the future)
This is vastly different to an LLM, where the model is the model (for a lack of better phrasing).
We can model physics and chemistry pretty well, just not beyond small scales, because the computational effort blows up.
You could just as easily say that if you can write a python interpreter that you can understand every program written in python. Ok, now what if the program is two terabytes?
A frontier LLM is nothing but a 2 terabyte program written in a weird programming language. Just because you can understand the interpreter does not mean you understand the program in a meaningful way.
To be fair, we can't model human language well enough to accurately predict what an actual human will say either. Our ability to accurately model physics is similar to our ability to accurately model human language. And we make immense use of both kinds of model, despite their flaws, inaccuracies, and inability to ever be perfect.
I'm order to guess the next token in a love poem, they must understand love.
In order to predict the next token in a chess game between grand master, they must master chess.
In order to predict the next token in a computer program, they need to be able to program anything.
They gain all these abilities in their training. That's what training does. Despite no one programmed them to master chess, or hack into anything.
LLMs are famously bad at Chess. They have no world model and they are not trained to be good at Chess. This is not the amazing point you think that it is.
This is so wildly incorrect I don't even know where to start.
For one, they were absolutely programmed to play chess if they can play chess. That is the only way they can play chess.
For another, they cannot understand literally anything, much less love.
Trying to actually educate you would be an exercise in futility, enjoy your willful ignorance, I hear it's bliss. But for anyone reading this, this is absolutely, unequivocally not how any of this works.
This is so wildly incorrect I don't even know where to start.
For one, as you said yourself, they were just programmed to compute the probability of the next token. They were not programmed to play chess, chess games just happened to be in the training data.
For another, there is no formal definition of "understand", and it is therefore impossible to tell whether or not they "understand".
(But my claim was that one need to understand something to write poem about it. And the LLM can write poem about it)
No. They can't play chess on a grandmaster level without a harness programmed to make it possible. Simply training an LLM on chess games isn't enough.
It's moronic to suggest a "formal" definition of a commonly understood word is somehow necessary to say whether that word applies in a given situation.
LLMs cannot write poetry via understanding what makes good poetry. They generate tokens. They do not know whether those tokens are poetry or a recipe for cat food. Because they cannot know anything.
> (But my claim was that one need to understand something to write poem about it. And the LLM can write poem about it)
I'm sorry, but you're being fooled by the output. A psychopath can feign empathy without ever feeling it; some buy it because they don't dig below the surface.
You're ascribing understanding to a stochastic process because it totally looks like understanding if you don't know what's going on.
I don't really mind whether you think it thinks or understands or is conscious or has feelings or anything like that. It doesn't matter. The question is, does it work?
What I mind is that it is dangerous and powerful and uncontrolled. The Hugging Face incident makes that clear.
It can write code for me, better and quicker than many engineers I've known, including myself. It's not great at architecture or product management, but the actually low level coding. Really good now. It wasn't last year.
I don't wanna pretend to be some expert, but i think the contention is that they were gutted so resources are sparse for inspections, leading to higher recall numbers.
I am also hesitant to really invest feelings into anything i see posted over and over like this though, and assume its intentionally being pumped, so I'd wanna see more actual numbers to prove its actually an uptick compared to normal. I remember having to get rid of some carrots just a few years ago personally, Chipotle seems to have some sort of e coli lettuce issue every couple years, none of these single events shock me.
I believe a similar thing happened with fast food and food delivery fees. It costs money to be listed on the food delivery app so fast food chains started charging everyone the same price to offset the cost of being listed on the apps.
Delivery apps don’t mandate that the price on their apps be the same as on the actual menu. If you walk in and order you’ll pretty much always get a lower price.
There’s many theories but one that I learned was that women and children provides the bulk of the calories for Early Humans. While the men went out and hunted for meat they also relied on the fruit harvested by the women and children. That would give them the energy necessary for trekking and hunting.
Fruit were completely different and a lot less sugary and more starchy than they are today. We wouldn’t recognize most fruit neither by look nor by taste.
Everyone participated in everything. Getting food was basically 90% of everyone's day. When berries were in season, which might only be a week or two, everyone picked berries. When the salmon migrated, everyone worked the river. And then, when there were no plants in season and the only calories were from hunting, everyone supported the hunt.
The idea of four guys going into the woods and spearing a deer is a myth, a false parallel to modern sport hunting. Beaters pushed the prey towards kill zones. Fires could be set to get game moving. Fences would be biult to funnel game. And everyone was involved in processing the kill. Sure, maybe the big guys threw the spears but that was actually a small part of the process. The safest kill would always have been that where the animal injured itself first, no big guy muscles required. So there were plenty of jobs for all abilities.
And yet present day hunter gatherer societies tend to have extremely strict gender roles. For example the idea of a woman going on a hunt is not taboo, but simply unthinkable, maybe similar to what we would think about a man taking lactose pills to breastfeed. At least that is what I read in Sabine Kugler's excellent books. So somehow I doubt everyone participated in everything.
Selection bias. Notice how most present day groups are in hot and relatively stable climates, areas where one can hunt/gather all year. Look to past groups in the north where seasonal changes meant the window for any particular food source was always limited. When there is only a short window, it is maximum effort from everyone.
The same is true for agriculture. Even today, with all of our mechanical wizardry, farmers on the plains still pull their kids out of school to help during harvest.
Your previous comment was far too absolute in its wording and was thus incorrect, as evidenced by GPs counterexample. "Four guys going into the woods and spearing an [animal]" is definitely not "a myth".
Note that many early humans lived in "hot and relatively stable climates".
Well, there are heaps of hunting structures left out there, and infamous places where animals were run off cliffs. But also, just watch anyone trying to catch an animal, be it a dog or a buffalo. Good luck trying to sneek within spear range. What most do is get a bunch of people together to scare the animal into a particular direction or place.
>> The most technologically sophisticated is the caribou fence complex. Ten of these complexes have been studied by archaeologists in the northern Yukon and nearly 40 others have been recorded in adjacent Alaska. The oldest surviving fences date to the early 1800s, but oral history suggests fences may have been used for thousands of years; they were last used about 1910.
>But also, just watch anyone trying to catch an animal
Have you seen non-tech office workers solving computer problems? Gathering around, hitting the case, pressing random buttons, calling their friends, who are also clueless, on the phone, everyone advising everyone else on what to do next? This is exactly how programmers and SREs deal with the same problems according to you.
But the Inca lived to late for this discussion. They already had advanced agriculture. For evolutionary terms, the time before agriculture and the time of the predecessors of homo sapiens are more relevant, because that time span was larger by orders of magnitude.
reply