>>990315>knowledgeable enough regarding MTL/Neural Networks/etc.Performance is cheaper per year, quantum computers are inevitable and offer qualitative leap in said performance (when adapted to classical IT tasks).
If you actually knew the difference between traditional Machine Learning approach which is 50+ years old and modern neural network specific AI (like YouTube algorythms), you'd understand that Machine Learning was actually limited by cost of performance for the longest time and this is rapidly changing.
Basically the time will come when you won't need a sophisticated neural network, finely tuned to your specific task. With big, powerful and cheap enough computing power you can feed giant massive of Big Data (translation pairs) to a less sophisticated system and get satisfying results.
Language barrier IS going to be sold by purely throwing computing power at it.
