4 ms·
Here is my perspective working in industrial automation field (think of these for 4 points as matrix): 1) Things that are easy to automate are already automate
by _fizz_buzz_ 6y ago
Here is my perspective working in industrial automation field (think of these for 4 points as matrix):
1) Things that are easy to automate are already automated.
2) Many things that are difficult to automate but increase productivity a lot are already automated.
3) Many things that easy to automate and increase productivity only a little bit are already automated.
4) Things that are difficult to automate and increase productivity only a little bit NOT automated.
We constantly make advances in automating things, but the low hanging fruits and high productivity gain things are already gone. It's not so clear to me, that our advances are keeping pace with the things that are left to automate. I feel the point about exponential growth is completely non-obvious. What if the tasks we are automating also get exponentially more difficult and/or the productivity gains decrease exponentially.
- divbzero 6y agoOP’s thesis is that AI sets us on a new exponential curve that dramatically increases the ease of automation and the range of things worth automating. At the same time, your point about other exponential effects that offset the gains is thought-provoking. As other comments point out, the Moore’s Law in computing hardware has not translated to a Moore’s Law for software usability or productivity. It’s not guaranteed that a Moore’s Law in AI will translate into a Moore’s Law for everything.