No, the information cost grows exponentially going from random letters to English sentences. We can measure this with crossentropy to get a probability of gene

No, the information cost grows exponentially going from random letters to English sentences. We can measure this with crossentropy to get a probability of generating a well formed English sentence from a uniform distribution over the letters. To generate just a few hundred words we end up needing more trials than are available in the lifespan of our universe.