If a computer scientist and a Shakespeare expert walking into an auditorium sounds like the opening line of a bad joke to you, you are not alone. Vinton Cerf and Dr. Michael Witmore themselves see the potential humor behind their dialogue hosted by the UT Humanities Center.
The unlikely pair took to the stage of the Student Union Auditorium Thursday night to discuss how computers can expand our understanding and access to literary works, such as Hamlet or Romeo and Juliet.
There may not be two men more qualified to have this discussion, entitled Machine Reading in the Digital Age. Cerf is often called one of the “fathers of the Internet.” He is a past Vice President of Google and has received countless awards and honorary degrees for his groundbreaking work in programming. Dr. Witmore is a professor of English who currently serves as director of the Folger Shakespeare Library in Washington, D.C., the largest collection of Shakespeare’s printed works in the world.
The dialogue quickly centered on the power of machines to increase our knowledge of classic texts. Dr. Witmore recalled his time teaching at Carnegie Mellon, when a computer sorted the plays of Shakespeare into the categories of comedy, tragedy and history in a matter of minutes using textual analysis, much to Dr. Witmore’s surprise. The computer had shown that the frequency of simple words such as “if” or “the” could predict the genre in which Shakespeare was writing.
“The computer drew my attention to a layer of the language that I had never paid attention to. And it showed me a constraint that either consciously or unconsciously Shakespeare was reacting to,” Witmore said. “It’s a different way of describing and understanding what Shakespeare had to get done as a writer. And that’s useful.”
Dr. Witmore and Cerf framed the goals of such textual computation around the expansion of access to critical information. If new algorithms can be made to identify texts in simpler ways, then even someone with no previous knowledge of Shakespeare can search for and discover his works.
While this technology is powerful, it is certainly not perfect. As Cerf was quick to remind the audience, it can be influenced by the bias of those that program it.
“Machines do you what you tell them to do. They don’t always do what you wanted them to do,” Cerf said. “I would be super cautious about relying on these machines without some additional consideration.”
The dialogue, while a volley of technical terms such as “multi-layered neural networks” and “successive approximations,” was ultimately about the power of different areas of study working together, as exemplified by Cerf and Dr. Witmore themselves.
Dr. Witmore summed up the unity between the two seemingly different experts.
“We’re not so far apart,” Witmore said.