Chatbots are a huge application of artificial intelligence that can talk to
you, just like a real person. I recently created my own chatbot, check it out here!
An explanation of how I built my own chatbot
to talk like a real person.
Tensorflow is a huge library that allows anyone to create projects. I used Tensorflow's built-in sequence to sequence model, which allows chatbots to take in long sequences of human speech as inputs, and output human speech.
Recurrent neural networks, or RNNs, are a special type of neural network. This is because they have the ability to basically remember their inputs. This property works really well for something like a conversation, so they are mainly used to create chatbots.
An encoder-decoder model consists of two RNNs. The first, the encoder, changes the inputted sequence (or words that you type) into something a computer can understand. The decoder then changes the chatbot's response into human speech.