i get it's sexy but can we plz be honest - the assumptions for zkml is extremely challenging and highly nontrivial

- "there will be a breakthrough in accelerator tech + mathematics enabling ai operations to be encoded in a proof-generating language"

- "any computation can be efficiently verified via SNARKs/STARKs"

will probably end up looking a lot like mobile/lite models today

reduced size/complexity, subset of operations, slightly lower accuracy

mostly applicable to a specific category of use cases where the need (privacy, trustless verification, etc.) justifies the limitations