Facebook Developing AI Team to Better Understand Users

According to MIT’s Technology Review,  Mike Schroepfer, CTO of Facebook, claims that a need for better understanding the type of content users are searching for has lead to the creation of a new team at the social media giant, known as the “AI Team.”  The team of eight is working toward creating an artificial intelligence model focused on Deep Learning in order to simulate actual brain synapses when they are processing data. This can help Facebook better serve users with content and information that they are interested in, based on their behavior and profile information on the site.

Deep Learning

Deep Learning is not a new term, especially to Facebook competitors Microsoft and Google. Both have teams that are working on building artificial intelligence systems that can help users find information, organize their data, and be more effective online. In addition, Google announced a new AI tool called word2vec in August 2013, which figures out the patterns and connections in language without human guidance, according to GigaOM. This software is built for understanding and creating natural language patterns without the assistance of a human controller.

Unassisted Help

It is crucial for this type of Deep Learning technology to be effective with little human assistance in order to process the large amounts of raw data that comes with millions of users that are on Facebook each day (as of May 2013, Facebook reports over 1.11 billion people are using the site daily).

In order for Facebook’s AI team to succeed, they will have to create a product that runs with minimal human interference and doesn’t take up large amounts of processing space– a heavy feat.

Schroepfer stated to MIT that some data from their efforts may be released to the public, which could help other social media and technology networks present better data and uses for the end user, according to Engadget.


photo credit: Saad Faruque via photopin cc