There’s even the next iteration already happening: Cerebras is maling wafer-scale chipa with integrated SRAM. If you want to have the highest memory-bandwith to your cpu core it has to lay exactly next to it ON the chip.

Ultimately RAM and processor will probably be indistinguishable with the human eye.

How big are your hands???

We live in Idiocracy.

Thank you! Nicely put. The problem isn’t people like your aunt, its massive shareholder-controlled investemet machines that own thousands or even millions of homes. Your aunt probably knows eafh renter by name - there can exist a personal relationship. There’s two things limiting your aunt becoming a money-hungry antisocial ghoul:

1. raising the rent is a relatively large amount of work for relatively small of a reward. If she raises rent she has to write these 4-5 renters a letter explaining why she has to increase it. Those renters might disagree, have objections, ask for reasons and proofs (like the central heating bill or maintanance costs etc). If she raises the rent by lets say 2% it’s 2% of not that much money (with her single digit number if houses).
2. she is raising the rent on people she knows. She is taking money away from people she even may like - have a personal relationship with.

So increasing rent is a lot of hassle and her renters might like her less after that - which might be a factor.

Now lets think of the hugr real estate company. They have thousands of renters and maybe hundreds of employees. They have lawyers employed. If they raise rent by 2% they have to send thousands of letters. But these letters are sent by people whose job it is to do so. Tyey can calculate in advance that from their renters X% will just accept the nrew rent, Y% will require some manouvering, Z% might move out and so on. They can estimate the cost of raising rent pretty well based on experience and compare to the profits they make. And with thousands of apartmants 2% is a lot of profit. The employees have no relationship to the thousands of renters. Renters are just numbers anyway. Everything is much more efficient. Also: Shareholders. They demand profits and dont’t care how. They care even less aboht the renters. They demand more profit and will just say “make it happen”. If thr ceo doesn’t raise profits - with whatever means necessary - the shareholders will replace the ceo.

The soltion IMHO would be some progressive tax That makes it basically unprofitable to have more than 10 apartments. And to prevent legal entities owning other legal entities owning apartments in order to circumvent this. If there exists (and can reasonable exist) a personal relationship between landlord and renter everything is alright in my opinion. People usually are not animals to eaxh other if they know eaxh other personal.

AI in general is a shitty term. It’s mostly PR. The Term “Intelligence” is very fuzzy and difficult to define - especially for people who are not in the field of machine learning.

Maybe you can dip your toes into using lineage os or graphene os?

The guy who wrote Winamp Sold it decades ago. So you should clarify who is defensive about their code :-) The original coder is really good - he also wrote the awesome AVS visualization plugin for Winamp which among other things utilizes a special programming language called “eel”. After selling Winamp he went on to create Reaper which also uses eel I think.

Everybody has to support the new new underdog Intel.

Yes

… for now.

It cannot “analyze” it. It’s fundamentally not how LLM’s work. The LLM has a finite set of “tokens”: words and word-pieces like “dog”, “house”, but also like “berry” and “straw” or “rasp”. When it reads the input it splits the words into the recognized tokens. It’s like a lookup table. The input becomes “token15, token20043, token1923, token984, token1234, …” and so on. The LLM “thinks” of these tokens as coordinates in a very high dimensional space. But it cannot go back and examine the actual contents (letters) in each token. It has to get the information about the number or “r” from somewhere else. So it has likely ingested some texts where the number of "r"s in strawberry is discussed. But it can never actually “test” it.

A completely new architecture or paradigm is needed to make these LLM’s capable of reading letter by letter and keep some kind of count-memory.

We long left the era where we “own” things that we buy. As everything is a computer now it has become very simple to control stuff that remotely that was working on its own before.

So the answer to “why would do this” is simply: “Because they can”.

Every tiny decision is guided by increasing profit. No matter the side effects (short or long term ). Because with many shareholders administering pressure to maximize profits there’s only one way to go (even if it’s a dumb and shortsighted decision) maximizing profits NOW. If you are not doing that because you can see that increasing profits now will hurt profits in the future then you are hindering the project. You have to increase profits now, because if you are not then your competitor is doing it and that is a problem. If you are not going with the project you will be out of a job sooner or later. Then someone will take over that will make the decision you couldn’t do.

This is a race to the bottom. Morals, integrity, honesty, responsibility and foresight are only obstacles in this logic (because the competition is not bound by them which gains them an advantage).

It’s simply cheaper now to build everything in the car always and run an operating system that manages all these things and can control what you are doing in your car.

Cory Doctorow held a great keynote about this some ~10-ish years (?) ago with the title “The coming war on general computation” where he explained the side effects of putting DRM in every stupid appliance. The side effect here is that we cannot hack our cars to switch on the heated seats (or whatever other feature BMW is not allowing us to use for free) because of DRM. It is not “our” car, even though we bought it.

I agree, but as long as we still have capitalism I support measures that at least slow down the destructiveness of capitalism. AI is like a new powertool in capitalism’s arsenal to dismantle our humanity. Sure we can use it for cool things as well. But right now it’s used mostly to automate stuff that makes us human - art, music and so on. Not useful stuff like loading the dishwasher for me. More like writing a letter for me to invite my friends to my birthday. Very cool. But maybe the work I put in doing this myself is making my friends feel appreciated?

Edit: It’s also nice to at least have an app that takes this maximalist approach. Then people can choose. If they’re half-assing it there will be more and more ai-features creeping in over time. One compromise after the next until it’s like all the other apps. It’s also important to have such a maximalist stand in order to gauge the scale in a way.

Can Brazil rent out that judge to Europe for a Sec?

yeah that’s just ridiculous. omg

It’s also a great example why these mega corps should be broken up into smaller pieces.

If forced arbitration persists (and this argumentation from Disney is successful and then used as precedence) any service used from one company can be used to forever ban you from taking legal action against that company again even if the service and the reason for the legal action have nothing to do with each other.

Am I right in understanding that this case is about someone dying from eating in a Disney owned restaurant that by accident was a Disney+ subscriber?

If one company owns everything like Amazon, Google, Apple and in the future maybe even water supply, garbage collection, operates my car and is my insurer or bank account (and owner of one of the 4 remaining fast food chains in the country) how can people actually sue a company then ?

Amen

I quit netflix recently after almost 10 years of continuous subscription. When they raised the price again it was a good Occasion although not the reason. I think so was done with it for a while now.

I guess I’ll start reading books again - or watch a few films instead of stretched out series.

I have no idea how to fix the problem, but I’ve read somewhere that burn (a relatively new machine learning framework in Rust) is capable of loading models like stable diffusion. As Burn is built with webGPU and all the shader transpiler-stuff that comes with it doesn’t that mean that it can also run easily on (even older) AMD cards? I think what’s lacking is equal performance as nvidia drivers are heavily optimized already.

Maybe someone knows more here?

what does “rootfull” mean?

Apparently Microsoft didn’t get the memo :-)