By Dyane O’Leary
Associate Professor of Legal Writing
Co-Director, Legal Innovation & Technology Concentration
Robots (insert: fun term for computer algorithms) vacuum our homes, tell us the weather, tweet out football scores, play our favorite songs, and help us avoid last minute traffic jams. But they can’t write “like a lawyer,” right?
Wrong. Algorithms can and do create written content. The Associated Press was ahead of the pack in 2014 when it paired with a technology company focused on language generation software. Today, machine journalism—aka robot reporters—produces articles about baseball games, financial news, and a host of other topics. Every day we hear about new sophisticated projects using artificial intelligence technology, like the MIT project in which a computer was “trained” to write horror by digesting 140,000 horror stories, then asked to produce its own (probably not a bestseller, but still impressive!).
Algorithms already help with legal writing. At least they have for any attorney who has accepted a spellcheck or grammar suggestion while working on a document in Microsoft Word. Of course, spelling fixes and wording suggestions are only a small piece of the traditional legal writing puzzle along with the core pieces of creativity, analysis, style, use of authority, citation, organization, and tone. Tools to enhance legal writing are exploding in popularity. WordRake is one piece of editing software I’ve used with my own writing and introduced to the students I teach. The Suffolk Law Legal Practice Skills Program is collaborating this year to introduce WordRake as an option for all first-year law students. In short, it is an add-in to Microsoft Word (an Outlook version is available, too, for email) that “rakes” your document in a track changes-like format and provides suggestions to re-word or delete text. The user then considers, accepts, or rejects those suggestions on an individual basis. The tool was built after WordRake founder Gary Kinder identified hundreds of signals that often indicated wordy, cluttered, and unclear or unnecessary writing – the opposite of what every lawyer wants, yet what we all seem to still do.
One thread weaving through discussions of “machine writing” involves training data. Any algorithm needs to be trained because machines can’t create something out of nothing. Instead, they need to be fed significant quantities of data (in the case of a writing tool, lots of documents and examples) and the system “learns” over time. Such is the case with legal writing “bots.” Tools like WordRake can’t make poor legal writing good with a wave of some magic computer wand, but in my experience can make legal writing just a little bit better.
A final word of caution. Legal writing editing tools are only as good as their users. WordRake forces me to consider a change – does it jibe with my desired tone? Does it clash with my personal style? The algorithm might prompt me to use two words instead of five but maybe I like those five. That’s ok. It’s still my work, after all. Isn’t it? Some question whether new writing tools clash with existing academic integrity rules. Perhaps the better question is whether old academic integrity rules need to change to keep pace with new technologies that help lawyers serve clients in a more efficient and effective way. Law students need to be exposed to such tools, not shielded from them.
Lawyers should embrace tools like WordRake, not fear them. If a robot lawyer can “take” a small piece of my job and improve the clarity of my writing, freeing me up to focus more time on strategy and case development and client counseling, then in my book that’s a win.