Creating systems that understand what we say and how we say it, engage in dialogue and respond in an intelligent, human-like manner (known as conversational AI or dialogue systems) has been one of the main goals of Artificial Intelligence. Intelligent conversational personal assistants (e.g., Siri, Cortana, Alexa) are not only widely used examples of such systems, but represent one of the fastest growing areas of AI, expected to affect most aspects of our daily lives in the near future.

Although a lot of progress has been made in natural language processing and understanding, building effective communication systems is far from being a solved problem. One of the main challenges is our perception of reality.  Bombarded with a large amount of information, we filter out the parts that seem unimportant based on our own genetic programming, early imprints, and subsequent conditioning of our learning and experience, thus adding our judgments, inferences, biases, and preferences to every day communication.

Located at the intersection of language – technology – society, Dr. Girju’s research interests are in designing, building, and testing AI systems that use language in order to facilitate successful human–human and human–computer communication. She is also working at the interface between verbal and non-verbal communication in developing novel human language technologies.

Her main broad research questions are:

1) How can language be used to result in better understanding and communication (and how should we address various barriers to effective communication)?

2) What are our linguistic habits and how do these manifest in the way we think as well as in our speaking/listening/writing?

3) How does language facilitate or impede the satisfaction of our emotional needs? (i.e., How should we use emotions to better connect with others - humans or machines?)

Dr. Girju's most recent research projects include behavioral analytics and affective computing (aka emotions) approaches to natural language understanding with applications to health care, multi-modal interfaces, and social media.