\
Block chains are way too slow handle modern financial transactions. They are interesting, but limited use cases in low transaction rate scenarios.
They have similarities with LLM token evaluation and a lot of the same problems.
Well they are consensus based, but the original vision of that was regular people all over the world, using regular hardware, would be determining that consensus.
As opposed to what we really got which is a relatively small number of corporate miners, concentrated in some jurisdictions, that grossly outweighs the individual miners, and may be at a point where some of the security assumptions of crypto are compromised. I.e. decentralization.
Regular hardware from around when the original bitcoin paper was released is pretty useless for mining when the computation required keeps rising exponentially. Some of the more successful Crypto miners (aka "crypto bros") are now into building AI data centers. Which gives you an idea of the scale needed.
LLMs remind me of blockchain where you add a new token to the end of the prompt and then have to traverse the entire prompt again to produce the next token. Wash, rinse, repeat. That's really inefficient, similar to blockchain.
Building AI models in hardware *may be* the best solution for Local AI.
No one should want any of the frontier model AI companies getting copies of everything you feed to their AIs.
See: https://chatjimmy.ai
This AI model and training is from 2021... so kinda of dumb... but ask it anything and see how fast it is. It will tell you how many tokens/sec it took for your answer.
AMD now owns this tech.
