This one is funny. While cryptocurrency gets most of the brickbats around here on HN, I think Deep Learning is also in the same boat.
Executives want to apply Deep Learning to everything and anything they can get their hands on. The results most of the time are underwhelming and could be had with a much simpler set of tools. Just wondering how long before Deep Learning goes into trough of disillusionment.
On the job I have to constantly shoot down big data / data mining / intelligent nn analysis crapware showeled to my bosses, and 99% of the time the answer is “you got NO data to speak of”
One day a consultant or another yesman will be able to check in and syphon precious resources with their pipe dream and I’ll watch six figures fly out of the window while my dev team struggles debugging simulated ipads with mac minis.
> One day a consultant or another yesman will be able to check in and syphon precious resources with their pipe dream and I’ll watch six figures fly out of the window while my dev team struggles debugging simulated ipads with mac minis.
Six figures? Oh sweet summer child. :)
First you'll need to burn seven figures to have a Big 3 advise your leadership that you need this tech. Then you burn the six figures implementing it with a lower tier firm.
We're in an industry ten years behind so we're just getting into Big Data. Obviously the issue that we can't write working normal software hasn't been addressed but we'll see how it goes. I'm actually relatively optimistic.
It feels like lately bots and AI are the biggest hype-train. Everybody wants bots. Bots for this, and bots for that. With natural language processing bolted on somehow.
Got a simple, defined flow that could just be a decision tree? Nope, we need natural language processing in a chatbot, we can't possibly present a menu.
I feel like I'm fighting the Infocom parser again.
brick·bat
ˈbrikˌbat
noun
plural noun: brickbats
a piece of brick, typically when used as a weapon.
a remark or comment which is highly critical and typically insulting.
"the plaudits were beginning to outnumber the brickbats"
Executives want to apply Deep Learning to everything and anything they can get their hands on. The results most of the time are underwhelming and could be had with a much simpler set of tools. Just wondering how long before Deep Learning goes into trough of disillusionment.