Content
@
https://warpcast.com/~/channel/aichannel
0 reply
0 recast
0 reaction
Justin Hunter
@polluterofminds
What are the most interesting crypto + ai projects? Most seem to just be tokenized incentives for compute which is not unique to ai, imo
1 reply
0 recast
3 reactions
π_π
@m-j-r.eth
you would think by some point that interoperability standards and [foo]-augmented generation would converge on an internet-scale knowledge & repertoire base. https://app.primeintellect.ai/intelligence/intellect-2 is great, too, but this is just a secondary structure, tertiary should articulate which mobile-capable specialist can serve a given part of a reproducible outcome.
1 reply
0 recast
2 reactions
Justin Hunter
@polluterofminds
I honestly don't think decentralizing compute resources is the big unlock for crypto + ai. I don't think it's a real problem. I don't know what the real problem is, which is in itself the biggest issue that always faces crypto. Looking for a problem
1 reply
0 recast
1 reaction
π_π
@m-j-r.eth
it's the surest path to zettascale compute. the training is not ideally decentralized, but its usage has to be (just for venture scale return on capex). I think MCP is rudimentary frame, once there are enough versions, it should become self-reproducing to all niches. just like with animal models optimize for watt, the local margin of compute will have to optimize for memory footprint, so the premeditated context is the fulcrum. deep research & computer use is barely explored, imho. yet, we get BitNet (https://huggingface.co/microsoft/bitnet-b1.58-2B-4T). everyone lets the studio bear cost of pretraining, some users risk the extra margin to squeeze 90+% quality on narrow tasks, and then everyone let the brooms carry the water to some ulterior technological leap.
0 reply
0 recast
1 reaction
