> This is not the same as accurately predicting results, if the real-world weather function is determined by the weather of surrounding locations, time of year,

> This is not the same as accurately predicting results, if the real-world weather function is determined by the weather of surrounding locations, time of year, and moon phase. How many have the "human intelligence" to do this? Especially more accurately than a computer (and without using any themselves) training on the same inputs and outputs?