Using machine learning to enhance human efficiency, learning and creativity
The intelligent interactive systems currently designed are typically too inaccurate and cumbersome to be useful; as a result people avoid them in favor of performing tasks manually. Dr. Krzysztof Gajos, Associate Professor of Computer Science at Harvard University, is designing more user-friendly systems that enable novel ways of interacting with computation. Using machine-learning algorithms and artificial intelligence, Dr. Gajos is revolutionizing the way humans and computers interact with one another. The development of useful and predictable interactive systems will benefit education, enhance creative development, and improve computer usability for people with disabilities.
There is difficulty in developing intelligent interactive systems, or systems that combine user interaction and machine learning, that are controllable, useful, and predictable despite the underlying technology being somewhat unpredictable and sometimes incorrect. Harvard University's Dr. Krzysztof Gajos is an Associate Professor of Computer Science working with these technologies to develop useful and efficient interactive systems that are helpful, rather than difficult, for the user.
Dr. Gajos' research is centered around making intelligent interactive systems more predictable and controllable, and therefore more useful, across a variety of disciplines and applications.
One of the newest projects in Dr. Gajos' lab is the development of tools that allow communities to share in the collaborative creative experience. Using current methods, a group of 1000 individuals instructed to brainstorm solutions to a problem will yield many duplicate solutions, with the remaining individuals sharing in a widespread variety of possibilities. Dr. Gajos' tool will reduce duplication and assist newcomers in building ideas off of those already brainstormed. This system will reason about the quality and diversity of past solutions to isolate those that may optimally inspire new users and direct them towards creating novel and valuable solutions of their own. This system uses machine learning to help the members of the community get the most inspiration out of each others' ideas.
"Lab in the Wild" is a live online platform for engaging the public in experiments whose results can inform the design of future interactive systems. With over two million visitors and one million participants, the platform attracts participants who require neither pay nor supervision, but are incentivized by the personalized feedback provided at the end of each experiment. A simple yet carefully designed system of checks and balances ensures that data collected is as good as data from a traditional supervised experiment. With participants from more than 200 countries, all age groups, and a wide variety of backgrounds, this platform allows collection of data that is representative of the great diversity of cultures, preferences and abilities.
In much the same way that a teacher may explain gravity by dropping an object in front of a classroom, social phenomena, cognition, perception and behavior can be explained to students using tools designed by Dr. Gajos. Real-time behavioral experiments showing how people act, reason, and behave are being brought to classroom settings. Having students participate in real-time social experiments and simultaneously analyzing the results offers insight into human cognition, perception and behavior that would otherwise be too counterintuitive or subtle to notice.
Dr. Gajos has developed a system, called Supple, that automatically generates user interfaces for individuals with motor impairments. The system first measures each person's individual abilities and then automatically generates new interfaces for existing applications that best match what that person is able to do. Between work, school, commerce, and social activities, people spend eight or more hours a day in front of a computer. With the modern day demand of using computers at such a high frequency, it is not enough to merely make it possible for people with impairments to use computers -- the access also has to be efficient. The Supple system enables individuals with motor impairments to use computers more easily and efficiently than existing accessibility solutions.
A natural curiosity for learning and discovering solutions to problems was an early indicator that Dr. Gajos would have a career as a researcher, though the path that would lead him to human-computer interaction was less obvious. While biology or mechanical engineering could have been viable outlets due to Dr. Gajos' passion for finding solutions that directly impact people's lives, shared family interests proved dominant in directing Dr. Gajos' research. With an electrical engineer for a father, Dr. Gajos was drawn to electronics early on, beginning with basic electronic systems and transitioning to computer science. Today, Dr. Gajos uses his knowledge of machine learning and artificial intelligence to create systems that will better people's lives and enhance human-computer interactions.
Dr. Gajos is an associate professor of Computer Science in the School of Engineering and Applied Sciences at Harvard University. His research interests are in human-computer interaction, artificial intelligence and applied machine learning. The phrase "intelligent interactive systems" describes well many of his interests: he is interested in how intelligent technologies can enable novel ways of interacting with computation, and in the new challenges that human abilities, limitations and preferences create for machine learning algorithms embedded in interactive systems. Together with several students, Dr. Gajos has started the Intelligent Interactive Systems Group at Harvard. The main themes in his current research are personalized adaptive accessibility, creativity support tools, interactive machine learning, methodologies for conducting large-scale experiments with online volunteers, and crowdsourcing.
Dr. Gajos is also a co-editor-in-chief of the ACM Transactions on Interactive Intelligent Systems.
In June 2008, Dr. Gajos graduated from University of Washington and subsequently joined the Adaptive Systems and Interaction group at Microsoft Research for a one year postdoc.
While at the University of Washington, Dr. Gajos built the SUPPLE system for automatically generating personalized user interfaces.
In the Fall of 2005, Dr. Gajos was visiting faculty at the Ashesi University in Accra, Ghana, where he taught Introduction to Artificial Intelligence.
Before coming to the University of Washington, Dr. Gajos spent seven years at MIT where he earned his Bachelors and Masters degrees, and where he also worked for two years as a research scientist managing the operations of the Intelligent Room Project and coordinating some of the activities related to Project Oxygen at the MIT AI Lab (currently part of CSAIL).