Making Computers Think Like a Journalist

Kristian Hammond, left, speaks to students and colleagues at Northwestern University. CREDIT: Kristian Hammond

Kristian Hammond, co-director of the Intelligent Information Laboratory, is leading a group of scientists in developing artificial intelligence that can think in a manner similar to the way a journalist thinks. The Info Lab studies the flow of information viewed online, and then tries to develop mechanisms to tailor information for people based on each individual's interests and activities on the web. One primary goal for Hammond and his team is to provide information to people before they realize they need or want it. The artificial intelligence effort, called "News at Seven" is a computerized newscast. Every aspect of the newscast is determined by a computer, from graphics to stories to the expressions of the animated hosts. Hammond has even created a program capable of creating opinions; the program takes bits and pieces of reviews from different sites and pieces them together to create a rating of a movie or TV show. For more information on the research, watch a demonstration of the virtual newscast featuring journalist Miles O’Brien and visit the laboratory website. Hammond answers the ScienceLives 10 Questions below.

Name: Kristian Hammond
Age: 52
Institution: Northwestern University
Field of Study: Computer Science

What inspired you to choose this field of study?
I read far too much science fiction as a child and decided it would be fun to try to build computers that reasoned like people. Once I began the work, I realized that what I really wanted was to use my understanding of how the human mind works to help bring the machines to people, instead of forcing people to go to a machine.

What is the best piece of advice you ever received?
One of my early mentors pointed out that when someone couldn't understand what you were saying, it wasn't their fault. It was yours. This one comment has made me focus on communication and making my work intelligible.

What was your first scientific experiment as a child?
I tried to figure out how to make hot light bulbs shatter with drops of cold liquid.

What is your favorite thing about being a researcher?
I get to have ideas in the morning that get made into reality in the afternoon.

What is the most important characteristic a researcher must demonstrate in order to be an effective researcher?
You have to believe in the data more than you believe in your own ideas. I am constantly struck by how many researchers have turned their theories into dogma, making it impossible for them to move forward. I think it is crucial to work on real problems rather than those generated by the desire to prove a point.

What are the societal benefits of your research?
I really do believe we can build devices that can track your needs and respond to them without having to ask for anything. I want a world of information that has no friction at all, so that no matter where you are, no matter what you are doing, no matter what you are thinking, the machine will get you what you want and what you need.

Who has had the most influence on your thinking as a researcher?
Roger Schank and Mick Napier. The former is my Ph.D. advisor. The latter is the creative director of the Annoyance Theater. He taught me more than anyone else that whenever you work on anything, you have to focus on the spine and not be distracted by anything else.

What about your field or being a researcher do you think would surprise people the most?
The thing that tends to surprise people the most is my focus on human reasoning. Most people think that a computer scientist is going to be thinking about the machine all the time. I think about people all the time.

If you could only rescue one thing from your burning office or lab, what would it be?
A story/poster that my son Chance made for me for my birthday a few years ago.

What music do you play most often in your lab or car?
Glam rock and its successors.