Artificial intelligence bridges the gap between human and machine

"Whether we're developing open-source machine learning for Twitter or creating a smart robotic limb, it is all artificial intelligence," says Kory Mathewson, PhD student with the Faculty of Science.

Katie Willis - 12 October 2016

From research on robotic limbs to a tool for pushing relevant content to Twitter users, PhD student Kory Mathewson's passion lies at the intersection between human and machine.

"My fascination with computing science arises from the exchange between two intelligences-one human and one artificial," explains Mathewson.

Under the supervision of Richard Sutton and Patrick Pilarski with the University of Alberta's Faculty of Science, Mathewson is exploring the ways in which interactive machine learning can make our lives easier, while also keeping our privacy safe.

"Whether we're developing open-source machine learning for Twitter or creating a smart robotic limb, it is all artificial intelligence," says Mathewson. "There are many inputs that produce many outputs. It's all numbers to a computer system."

Interning at Twitter

During the summer of 2016, Mathewson interned at Twitter, using his expertise in artificial intelligence to develop torch-twrl. Inspired by the way humans learn torch-twrl is a reinforcement learning tool helps to develop algorithms to provide Twitter users with relevant, interesting content based on both their stated preferences and their behaviour on Twitter.

At the heart of Mathewson's work is teaching artificial intelligence to model human intelligence and understand how human behaviour changes over time. So, what does it mean for an artificial intelligence to understand you? How is your personal data shared? These questions, explains Mathewson, are top of mind for not only industry leaders like Twitter but also artificial intelligence researchers and developers.

"My fascination with computing science arises from the exchange between two intelligences-one human and one artificial." -Kory Mathewson

The future of artificial intelligence

"As we move into an increasingly modern and technologically advanced society, we have more and more autonomous technology," explains Mathewson. "For example, self-driving cars, automated calendars, apps that measure our food intake and exercise are all commonplace. The more we rely on technology, the more we give up some of our autonomy to these artificial intelligences. So, we want to be able to control how this sharing happens."

As the internet gets exponentially bigger, artificial intelligence will be used to help filter and sift through the masses of information in order to find what we want.

"For instance, without machine learning-based spam filters, email would be unusable."

Twitter has open sourced torch-twrl so that the technology is available for anyone, anywhere in the world to tweak, change, and use.

"It is in our nature to ask questions," explains Mathewson. "Science, much like machine learning, is about following your curiosity, finding answers to those questions, asking more, and continuing to learn about the world."