Content
@
0 reply
0 recast
0 reaction
Raunaq
@friendly
I am developing SuperFast with the goal of eliminating the learning curve in crypto, making the move to on-chain activities a matter of seconds rather than a long process. https://frames.neynar.com/f/c7e3ce16/817f52fc
3 replies
1 recast
4 reactions
Chinmay 🕹️🍿
@chinmay.eth
LLMs are probabilistic and it can make a mistake. Who would be responsible if money is lost in a transaction?
2 replies
0 recast
0 reaction
Anil
@korgy
Inspite of all of this if something were to go wrong, on the inference end (not the Daaps being used under the hood for e.g. Uniswap, Curve), probably the AI assistant should insure the transaction.
0 reply
0 recast
0 reaction
Anil
@korgy
LLMs can hallucinate. Fine-tuning with domain-specific data helps reduce this. Additional processes can validate user intent, like a maker-checker protocol, and user confirmation of actions, e.g., “You are about to swap 0.001ETH for 100 Degen tokens.”
0 reply
0 recast
0 reaction