71 - Is intelligence just an algorithm? With Jade Abbott

We chat to Jade Abbott from Retro Rabbit about artificial intelligence, broadly and more specifically about NLP and what that means for us.

Chantal, Kenneth & Len talk to Jade about natural language processing, commonly referred to as NLP. What does it take to get a machine to understand what we're saying as people?

Jade has always had a fascination with smart machines, from trying to build robots in school and now teaching machines to understand what we're saying. Jade takes a fairly complex topic and helps us come to terms with it. We question whether people, or intelligence, is algorithmic and what that means.

Processing natural language is not without challenges and Jade walks us through the maze of terminology and some tools to get started with, and we have several resources below to help as well. What would happen if AI tries to right a movie? What happens if the movie was made?

Importantly, neural nets are not the whole of AI. We wander around expert systems, random forests, and other great statistical models that very useful and predictable.

Is intelligence just an algorithm? What do you think? Let us know!

Find and follow Jade online:

• https://twitter.com/alienelf
• http://github.com/jaderabbit
• https://twitter.com/fmfyband
• https://fmfy.bandcamp.com/

Jade has some repos with sample projects on GitHub:

• https://github.com/jaderabbit/botcon2016
• https://github.com/jaderabbit/deepdreamsofelectricsheep
• https://www.kaggle.com/jaderabbit/training-an-lstm-to-write-songs

Jade offers some great resources not specifically covered in the show.

For people looking to get into AI:

• https://www.coursera.org/learn/machine-learning - Andrew Ng's Coursera Machine Learning course
• Kaggle - http://www.kaggle.com/
• https://medium.freecodecamp.org/the-best-data-science-courses-on-the-internet-ranked-by-your-reviews-6dc5b910ea40

How Neural Networks Really Work:

• https://www.youtube.com/watch?v=EInQoVLg_UY&t=78s How Neural Network's Really Work by Geoffrey Hinton

Using Natural Language for AI

• The key blog post on using deep learning for Natural Language Processing: http://karpathy.github.io/2015/05/21/rnn-effectiveness/
• Beautiful tutorial on Word2Vec http://mccormickml.com/2016/04/19/word2vec-tutorial-the-skip-gram-model/
• My kaggle notebook for training a neural network to generate songs: https://www.kaggle.com/jaderabbit/training-an-lstm-to-write-songs

Here are some resources mentioned in the show:

• Ex Machina - https://en.wikipedia.org/wiki/Ex_Machina_(film)
• Alien Covenant - https://en.wikipedia.org/wiki/Alien:_Covenant
• Marvin from Hitchhikers Guide - https://en.wikipedia.org/wiki/Marvin_(character)
• Sherlock - https://en.wikipedia.org/wiki/Sherlock_(TV_series)
• Natural Language Processing - https://en.wikipedia.org/wiki/Natural_language_processing
• word2vec - https://en.wikipedia.org/wiki/Word2vec
• Convolutional neural network - https://en.wikipedia.org/wiki/Convolutional_neural_network
• Recurrent neural network - https://en.wikipedia.org/wiki/Recurrent_neural_network
• Kaggle - https://www.kaggle.com/
• Sunspring | A Sci-Fi Short Film - https://www.youtube.com/watch?v=LY7x2Ihqjmc
• Rabbiteer - https://rabbiteer.io/

Picks

• Jade
  • Kaggle
  • Sunspring | A Sci-Fi Short Film
  • Creativity: how is AI impacting this human skill?
• Chantal
  • For Computers, Too, It's Hard to Learn to Speak Chinese
• Kenneth
  • Westworld
• Len
  • Instaparse

Thanks for listening!

See all past picks on our picks page.