“Ankur Goyal asserts that there is a general consensus among practitioners that the internal workings of Large Language Models are not fully understood.”