Are Deep Learning and GPUs changing the game?

Question

"Essentially, all models are wrong, but some are useful."

--- Box, George E. P.; Norman R. Draper (1987). Empirical Model-Building and Response Surfaces, p. 424, Wiley. ISBN 0471810339.

There is a great post on this topic in a previous post What is the meaning of "All models are wrong, but some are useful"

I just wanted to have some opinions regarding whether this holds true for modern Deep Learning models as well ? I mean arent all models useful nowadays ? I mean look at Alexa, Siri, Google maps, all of these have made our lives incredibly smoother. Isint it ?

Answer 1

Just because some models are useful doesn't mean all models are useful. Even if "deep learning" is useful in many cases that doesn't mean that every single time someone uses it for something the result is automatically useful. It's a tool. Screwdrivers are useful but it's not true that every time someone tries to use a screwdriver they are doing something useful with it.