Short answers, pulled from the story.
Rina Dechter introduced the term deep learning to the machine learning community in 1986. Igor Aizenberg and colleagues later introduced it to the artificial neural networks field in 2000, in the context of Boolean threshold neurons.
The first working deep learning algorithm was the Group Method of Data Handling, published by Alexey Ivakhnenko and Lapa in 1965. It trained arbitrarily deep neural networks layer by layer through regression analysis, and a 1971 paper described a network built with this method that had eight layers.
Yoshua Bengio, Geoffrey Hinton, and Yann LeCun were awarded the 2018 Turing Award for conceptual and engineering breakthroughs that made deep neural networks a critical component of computing.
AlexNet, designed by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton, won the large-scale ImageNet competition in October 2012 by a significant margin over shallow machine learning methods. It helped trigger the broad adoption of deep learning in computer vision.
Deep learning reduced phone error rates on the TIMIT benchmark from roughly 26% with randomly initialized recurrent networks to 16.5% with hierarchical convolutional deep maxout networks. All major commercial speech recognition systems, including Amazon Alexa, Apple Siri, and Google Now, are now built on deep learning.
Deep learning networks can be fooled by small, imperceptible changes to images that cause confident misclassification, a vulnerability documented in 2013. Researchers have also shown that adding stickers to stop signs, using certain sounds to hijack voice command systems, and feeding false data into training sets can all compromise deep learning systems.