Richard Socher says replacing manual feature engineering with vectors, neural networks, and end-to-end backpropagation enabled AI systems to work well at scale.

“and where that started you know was part of how we started image yet but anyway so like it was is really like fun to do but when we replace all of that manual feature engineering with vectors and neural nets and just backprop through everything it actually started to work really well at scale and so then everyone started to do architecture engineering and I was like that clearly can't be it”