--

Jordi, since you are obviously interested in this topic, I thought you may appreciate a few recommendations - in case you wish to evaluate some different opinions out there, compared to the "DL optimists" you mention:

https://spectrum.ieee.org/deep-learning-computational-cost

https://nautil.us/deep-learning-is-hitting-a-wall-14467/

https://thegradient.pub/machine-learning-wont-solve-the-natural-language-understanding-challenge/

https://spectrum.ieee.org/deep-learning-cant-be-trusted

https://medium.com/ontologik

Even Jeffrey Hinton, the other member of the triad of the Deep Learning idealists with Lecun and Bengio, doesn't believe it anymore.

Of course it is hard to admit something like this if you have based your entire career, your persona and your success on DL. And that is why even some respectable scientists keep pushing. It is a positive thing for science. They should be pushing.

Whether I should believe that they will succeed, though, that is another matter :)

--

--

A Grumpy Computer Scientist

UK-based AI professor interested in AI, mind, science, rationality, digital culture and innovation. Hobbies: incessantly fighting nonsense.