It's not exactly the same problem, in that you can parallelize usage of an LLM and copy it over to another computer, but cannot do the same things with a brain. Put it another way, humans do not have the processing power needed to answer hundreds of millions of queries per day, while LLMs do.