When growing up, about 90% of the kids in the world learn to read and write. They learn about using characters to construct words and use words to form sentences. With these sentences, they can tell stories.

These days we are teaching computers to do the same. By giving them lots of text, we can teach them to give answers to questions, interact through chatbots, even interact through voice. Computers can also write news and even (parts of) books.

In this talk, you’ll get an idea of what you need to start your own “Natural Language Processing” experiments. How can you teach a machine to read and write? Without going into the mathematical details, you learn the basics of word2vec, CNN, RNN, LSTM. In the end, I discuss transfer learning using BERT. Of course, the talk contains demonstrations of these techniques, and I’ll show you a reading and writing machine.