Initial Steps Toward Automating Legal Document Editing

Drafting successful legal documents requires experienced lawyers, careful research, and nuanced reasoning. Nonetheless, drafting also requires many mundane and time intensive tasks that delay the process and increase its cost. One consequence of this is that law firms hire a large number of support employees to assist with these low-level tasks. According to US Bureau of Labor Statistics data released in May 2016, approximately 40% of workers in US legal services are in support roles (e.g. paralegals and legal assistants) and these workers are paid $12bn annually. Starting by reducing time spent on editing, one of the most common low-level tasks, machine learning has the potential to significantly improve writing and reviewing legal documents. To reach this potential, many challenges need to be further understood and subsequently addressed, including how machine learning can comprehend the key points in a legal document and reflect lawyers’ negotiations on those key points in its recommended edits.