EdX, the consortium of universities, including Georgetown, that offers Massive Open Online Courses for free, introduced experimental software last week that uses artificial intelligence to grade and provide feedback on essays.

“The goal of edX’s open-ended assessment tool kit is to help students learn through relevant feedback systems,” edX Director of Communications Nancy Moss wrote in an email. “While often the preferred way to help students learn is to have one-on-one or small group interaction with a faculty member, this is not always possible in a MOOC or classroom environment.”

Moss said that software was still in the pilot phase and added that the technology is not intended to replace the role of an instructor but rather to help teachers handle the workload of a MOOC, which can include thousands of students and assist them in their other roles.

Moss added that she believes that students in MOOCs will benefit from this instant feedback, since they will not have to wait for hundreds of submissions to be graded individually by a human.

“Open-ended AI assessment is less about grading a paper or answer and more about offering ways to generate feedback for students in an accelerated way,” Moss wrote. “The current technology provides ongoing, real-time feedback to students, allowing learners to iteratively improve their answers, with the help of an instructor-defined rubric.”

English professor John Glavin said the technology could be used to help students understand the basics of a given subject before moving on to more complex topics.

“A lot of what I spend my time doing could be better done if I were to create a MOOC in which I would say, ‘Here is a very basic dramatic structure and here’s how it works. Did you get it? If not, let’s go back and do it again,’” Glavin said. “I’d love to have all of that taken off my back so I could spend more of the time reading and critiquing individual submissions.”

He added that the essay-grading software would allow students to revise and resubmit their work until they get the basics right, freeing him to focus on grading the advanced work and final product.

Glavin also noted that technology can make the classroom and learning process more comfortable and accessible but warned that tools such as essay-grading software have their limits, especially when evaluating more technical or specific essays.

“The key is to really decide which subjects and which skills are really suited to this kind of training and which aren’t,” Glavin said.

Other professors were less ready to consider using computers to grade submissions. Government professor Fr. Matthew Carnes, S.J., said that he is skeptical of a machine’s ability to evaluate the complexities of an essay.

“I have major reservations about this. I think there’s something to really understanding an essay that computers just can’t copy,” he said. “The computer can’t capture all the subtleties. Even with rubrics, they can’t capture all of the pieces.”

Classics professor Douglas Boin (COL ’99) said that though he frequently uses technology in the classroom, he prefers to personally interact with students in grading.

“I’m a big fan of using technology in the classroom,” Boin said. “But in terms of handing over the power to a computer to do grading, I’m not ready to do that yet simply because I have my own high quality-control standards already in class. I think many professors might feel that way.”

Boin noted that technology should be focused on creating the best learning environment on campus before branching out and that schools would succeed online only when they had streamlined their existing technological resources.

“I think there’s a lot of in-house technology building that has to be done before we start outsourcing large lecture classes, things like linking grades on Blackboard to the registrar or to MyAccess,” Boinsaid.

Boin also said that schools should focus on tools like the Corp’s Classy, a class registration system released April 2, before moving to online learning models. Initiatives like edX, according to Boin, work best when they complement a strong technological infrastructure.

“One is window-dressing and the other one is much more long-term that must be done,” he said.

Leave a Reply

Your email address will not be published. Required fields are marked *