>>13546261>Machine learning has been going for like 50-60 years and has been pretty fucking productive for the last 10Horse shite. 50 years and what does it have to show for it? There are no serious real world applications of ML out there, only meme-tier crap:
>image recognitionthank you for censoring tiddies on instagram, such a worthy endeavour. Also thank you for some more dystopian crap like face recognition
>speech recognitionUseless, can barely recognise word by word. You can always tell the messages that have been dictated by the stupid mistakes
>smart assistantshilariously bad, can't do anything more complex than setting an alarm or looking up a song (often not even that)
>refining search engine resultsGoogle is much more crap at getting you what your are actually looking for now than it was 10 years ago
>customer service botsnobody uses them other than to ask for being connected to a human operator because they are fucking clueless
>medical imagingno actual deployment to speak of
>product recommendationuseless crap nobody asked for and everybody could do without. You are far better off following your friends taste/suggestions
Perhaps the most telling example is that those creepy Boston Dynamics robots, which at least work, don't use stuff like reinforcement learning to walk (which would be the obvious guess) but rather good old-fashioned Optimal Control theory
In other words, kys
