j4k3

::: spoiler We are at a phase where AI is like the first microprocessors; think Apple II or Commodore 64 era hardware. These showed potential, but it was only truly useful with lots of peripheral systems and an enormous amount of additional complexity. Most of the time, advanced systems beyond the cheap consumer toys of this era used several of the processors and other systems together.

Similarly, now AI as we have access to it, is capable, but has a narrow scope. Making it useful requires a ton of specialized peripherals. These are called RAG and agents. RAG is augmented retrieval of information from a database. Agents are collections of multiple AI's to do a given task where they have different jobs and complement each other.

It is currently possible to make a very highly specialized AI agent for a niche task and have it perform okay within the publicly available and well documented tool chains, but it is still hard to realize. Such a system must use info that was already present in the base training. Then there are ways to improve access to this information through further training.

With RAG, it is super difficult to subdivide a reference source into chunks that will allow the AI to find the relevant information in complex ways. Generally this takes a ton of tuning to get it right.

The AI tools available publicly are extremely oversimplified to make them accessible. All are based around the Transformers library. Go read the first page of Transformers documentation on Hugging Face's website. It clearly states that it is only a basic example implementation that prioritizes accessibility over completeness. In truth, if the real complexity of these systems was made the default interface we all see, no one would play with AI at all. Most people, myself included, struggle with sed and complex regular expressions. AI in its present LLM form is basically turning all of human language into a solvable math problem using regular expressions and equations. This is the ultimate nerd battle between English teachers and Math teachers where the math teachers have won the war; all language is now math too.

I've been trying to learn this stuff for over a year and barely scratched the surface of what is possible just in the model loader code that preprocess the input. There is a ton going on under the surface. All errors are anything but if you get into the weeds. Models do not hallucinate in the sense that most people see errors. The errors are due to the massive oversimplifications made to make the models accessible in a general context. The AI alignment problem is a thing and models do hallucinate but the scientific meaning is far more nuanced and specific than the common errors from generalized use.

What is the deal with the cage thingies?

Those are not citizens. This is not a democracy.

The unfried crust and/or lack of a cheese like petroleum derivative ring in the periphery is unamerican

Every government office has maintained the stance as it has been. It is still the ROC not the ROT. https://en.wikipedia.org/wiki/Taiwan

Why should I care when Taiwan's official stance is also that it is the one legitimate government over all of China? Seriously asking. I could be more supportive of independence if it had been independent all of this time. Instead, this is a stalemated civil war where neither side has changed. How can one discuss the ethics without the full scope of the divide?

It is not super common to impregnate on first offense, especially if you were her first child. You can count the days backwards from your birthday to see when it happened. If you were the first child, you may have been a day or few late.

Growing up, I found it funny how many of my friends happened to be born in the first week of September... Happy New Years. There is often, not always, but often some correlated reason why they were free to screw around too much.

Multi threading is parallelism and is poised to scale to a similar factor, the primary issue is simply getting tensors in and out of the ALU. Good enough is the engineering game. Having massive chunks of silicon laying around without use are a mach more serious problem. At the present, the choke point is not the parallelism of the math but actually the L2 to L1 bus width and cycle timing. The ALU can handle the issue. The AVX instruction set is capable of loading 512 bit wide words in a single instruction, the problem is just getting these in and out in larger volume.

I speculate that the only reason this has not been done already is because pretty much because of the marketability of single thread speeds. Present thread speeds are insane and well into the radio realm of black magic bearded nude virgins wizardry. I don't think it is possible to make these bus widths wider and maintain the thread speeds because it has too many LCR consequences. I mean, at around 5 GHz the concept of wire connections and gaps as insulators is a fallacy when capacitive coupling can make connections across all small gaps.

Personally, I think this is a problem that will take on a whole new architectural solution. It is anyone's game unlike any other time since the late 1970's. It will likely be the beginning of the real RISC-V age and the death of x86. We are presently at the age of the 20+ thread CPU. If a redesign can make a 50-500 logical core CPU slower for single thread speeds but capable of all workloads, I think it will dominate easily. Choosing the appropriate CPU model will become much more relevant.

Mainstream is about to collapse. The exploitation nonsense is faltering. Open source is emerging as the only legitimate player.

Nvidia is just playing conservative because it was massively overvalued by the market. The GPU use for AI is a stopover hack until hardware can be developed from scratch. The real life cycle of hardware is 10 years from initial idea to first consumer availability. The issue with the CPU in AI is quite simple. It will be solved in a future iteration, and this means the GPU will get relegated back to graphics or it might even become redundant entirely. Once upon a time the CPU needed a math coprocessor to handle floating point precision. That experiment failed. It proved that a general monolithic solution is far more successful. No data center operator wants two types of processors for dedicated workloads when one type can accomplish nearly the same task. The CPU must be restructured for a wider bandwidth memory cache. This will likely require slower thread speeds overall, but it is the most likely solution in the long term. Solving this issue is likely to accompany more threading parallelism and therefore has the potential to render the GPU redundant in favor of a broader range of CPU scaling.

Human persistence of vision is not capable of matching higher speeds that are ultimately only marketing. The hardware will likely never support this stuff because no billionaire is putting up the funding to back up the marketing with tangible hardware investments. .. IMO.

Neo Feudalism is well worth abandoning. Most of us are entirely uninterested in this business model. I have zero faith in the present market. I have AAA capable hardware for AI. I play and mod open source games. I could easily be a customer in this space, but there are no game manufacturers. I do not make compromises in ownership. If I buy a product, my terms of purchase are full ownership with no strings attached whatsoever. I don't care about what everyone else does. I am not for sale and I will not sell myself for anyone's legalise nonsense or pay ownership costs to rent from some neo feudal overlord.

Stupidity is everywhere. It is universally human and in every race and culture.