Can We Replace Programming Languages by Natural Instruction?

Abstract

Now that computers are finally able to have simple conversations, it is time to explore the potential for replacing programming languages with natural language instruction. For example, less than 1% of phone users can program their phone to do new things for them, but if this line of research succeeds we might change that to 99%. This talk will describe our recent research exploring how we might enable phone users to teach their phones to perform new commands, using natural language interaction together with demonstrations. This line of research represents a paradigm of ``conversational machine learning'' that complements current data-intensive statistical approaches. If successful, it has implications for many types of computer interfaces, from giving users more control over their smart phones, to a providing a new generation of teachable web browsers. This talk covers joint work with Igor Labutov, Forough Arabshahi, Brad Meyers, Shashank Srivastava, Toby Li, Jennifer Lee, Antoine Bosselut, and Yeijin Choi.