My research concerns building communicative AI systems that can communicate with us in natural languages. I believe such systems are necessary for achieving AGI’s that we can trust, and in building such systems, we will inevitably need to better understand intelligence, especially human intelligence. Currently, I am working on (a) enabling natural-language parsers to better parse sentences by using large quantities of commonsense knowledge, and (b) building cognitively inspired systems that learn to bootstrap commonsense knowledge from the physical world and from reading texts.