23 February 2021 in IEEE Software
We address the question: How can AI help software engineers better do their jobs and advance the state of the practice? Our perspective comes from building and integrating AI-based techniques in Facebook’s developer infrastructure over the past two years. In this article, we describe three productivity tools that we have built that learn patterns from software artifacts: code search using natural language, code recommendation, and automatic bug fixing. We also present a broader picture of how machine learning can bring insights to virtually all stages of the software lifecycle.