The computing power of AI is like a fierce horse. It can't be restrained on-chain. Lagrange suggests using a co-processor as a groom, leading it off-chain, and after the calculation, handing over a "certificate of good conduct." The groom has great strength and claims to save 90% of the fodder (Gas), but if the horse stumbles and makes a mistake in the calculations, can the groom hold it down? Once this certificate is handed over on-chain, will the gentlemen dare to trust it?
