Citation? It's literally the first three words in your comment that person was responding to. Is this some kind of weird gaslighting experiment you're trying? It's not working.

You need a huge model to even come close to what you get from claude type tools. Then you also need a huge context window so a tool like claude can even work when you direct it to your local llm. That means your developer laptop needs like 128gb of ram and most of that needs to be allocated to vram from your apu. Unless you are doing some experimental setup where you can split between vram and normal ram without taking a perf hit. Either way though, you are going to be generating tokens quite slowly. Enpecially with the power tdp of a laptop

A laptop with that much ram and a good apu (newest amd strix) will be like over 5k these days i bet.

You can't just throw some tiny 12b sized model from ollama on and think it's equiv to claude. It very much wont be even close. You need something like the 100+ billion parameter models to even get a little close and still have memory to have a very large context window ( which will almost definitely be smaller than claude). Good luck doing that on a laptop.

There's a reason why ai providers are losing money ... The real world cost of actually usable and productive ai is high... Not just in hardware but energy too. Every less is basically a toy, and wont be usable for serious work beyond maybe the simplest tasks.

Easy, dont interact with it as an adult either. Simple. And we know how that works because it's how it has always worked except for the last 15-20 years. It's really quite easy to avoid using social media today. If all of it died off today, humanity would be better off tomorrow.. its positives do not outweigh its negatives. Not even close. Social media is not a bastion of useful or interesting free speech. It's a tool leveraged by corporations to get people addicted to hate so they generate ad revenue. No more beneficial to society than a casino without regulations would be.

You can't teach a large language model. It doesn't understand or reason. It just predicts and associates tokens well enough to simulate those things. It can't live-adjust it's training of those associations..,which is required for actual learning. It's got basically a fixed long term memory and an entirely disconnected temporary memory for the specific context of the prompt that it's currently responding to that must be fed to it from elsewhere. After which it starts over from scratch when a new prompt is submitted. Agents try and assist with smart rag-ish features to inject relevent context to mimic memory. But it doesn't alter the model...it isn't learning anything. So sad for the ai browsers and such. My heart goes out to the billionaires running the ai companies and the hardships they must endure to conquer the unwashed masses for once and for all.

I use claude regularly (we have to). I'm not sure what text animation is being required ...,it doesn't do anything you haven't seen it any ncurses type of application and any animation is simply aesthetic in the tool. All of the actual content is effectively just scrolling the text the ai or tools output or menu driven navigation. Again, i think the framework and stack was chosen because that's just what they use in general first and foremost. You dont need high performance to rotate a stupid 'waiting' text phrase while waiting for tools or network calls to complete and you dont need game engine performance when every interaction with the tool has to talk to the ai cloud service and (if you are using the expensive models) wait multiple minutes for a response.

Claude just needs to load fast, relatively..,it doesn't need to do anything else fast because the bottleneck is the ai calls or tools. Python would be perfectly fine here. Ideal even. The reason why not was likely not related to performance primarily. 3.12 python is plenty fast for a higher level language with a ton of experienced developers for it. It's likely that the choice came down to what the dev team uses elsewhere....and so just code it in the same language. Companies work that way far more often than 'best tool for the job'.

Not surprising, but patents (in the US) really need to be a use it or lose it thing. They're not intended to prohibit inventions, but protect them temporarily from competition. You should lose them if you dont have a product in customer hands within a year of it being granted and if you stop making it available, the patent becomes forfeit. But i guess this is just one thing in a long list of patent reforms needed. Also, this is a stupid patent, but nevertheless, a gross misuse of patents.

They're not going to only be used to fight crime. As we have seen in multiple other technologies that offer surveillance to an entity with no effective oversight.

This is all about cost of labor. Not filling gaps in talent pools domestically. Unless the company is hiring in a different country to support products they are selling to that locality, the us government should tax the us company the avg wage difference for every hire made in a foreign country. Then we can see what is about filling gaps in talent vs exploiting global flexibility of a company vs individual laborers.

If it's been going on since 2019, i'd assume an investigation has happened in the last 6 years. So a comment on why it's still allowed seems reasonable.

Rather than a lack of algorithmic definition. Math has a funny way of describing more than just our reality, perhaps they aren't working within the correct restrictions to the mathematical parameters that went into their logical proof. Until they can state exactly what is undefinable by physics, and we are unable to define it... I'll just keep ansuming this is less a proof of simulation being impossible, and more that, we lack the current understanding of how the simulation might work.

You dont need one equation to run a simulation, you can work with many. And the simulation only has to apply what your 'players' are observing at a given time for the given things they are observing. If the simulation known how complex it needs to simulate something, and it only needs to do it for 'players' in the simulation to higher complexities, the room for optimization seems pretty large to make the physical requirements much less than impossibly high.

in any case, the whole non algorithmic stuff sounds like a matter of ignorance rather than an impossibliity to define rules for something. Maybe an ignorance we are incapable of overcoming because the simulation won't create something that can, to stop the recursive problem (or we've reached the recursive level where simulating in a simulation is incapable of being good enough to offer that level of understanding due to it's accumulated approximations of true reality)

The dollhouse is the one that used human brains for compute power

Youtube music in getting absolutely filled with ai slop songs. I'm constantly having to thumbs down them when i hear their very obvious formula.
Though, i doubt youtube understands i'm thumbing down garbage ai songs and not anything to do with the metadata about the type of music. My feeds are likely very odd now to their algo considering what i have liked vs thumbed down.

Ai music really needs to have identification mandatory tagged and filterable. It's tedious and tiring to keep having to deal with it.

Pretty sure this sounds like direct die liquid cooling with the added post processing of etching some channels that increase surface area and direct flow.

Seems pretty logical for an improvement to direct die cooling (which has been around forever). Just takes the work done to water blocks and implements them in micro scale on the silicon itself.

My concern would be with such tiny channels, how frequently they would get clogged by inevitable partially dissolved minerals picked up over time from the radiator and other metal in the system.

My experience with water blocks has been, the smaller the channels, the more frequent the maintenance is needed to retain 'like new' performance. Granted, silicon isn't metal, so maybe it doesn't attract deposits.

Well, less sad and more better than expected. Given the behavior of the bcache maintainer/creator, i fully expected it to be torn out entirely. It's experimental after all, so removing it was totally on the table.