LLMs at t=0 are predictable (will produce the same output starting from the same input) and do a much better merge work than any other non ML based algorithm.

LLMs at t=0 are predictable (will produce the same output starting from the same input) and do a much better merge work than any other non ML based algorithm. Cut & paste all the examples you can come up to Claude and tell me if LLMs are not able to do this kind of merging.