The landscape of online learning is changing fast, and its latest innovation might be both its most promising and riskiest yet.
EdX, the massive open online course platform created by Harvard University and the Massachusetts Institute of Technology, recently unveiled a new automated grading system that evaluates a student’s essay in seconds.
While this is not the first time such software has been created and mobilized in the education community, it is the first time that it’s been implemented in the MOOC world.
Additionally, the new software requires humans to first grade 100 essays or questions before the machine starts grading. The machine also learns from the data and continues to improve its grading techniques.
Coursera, the platform on which Penn professors offer free online courses, does not yet utilize machine grading for student essays. Instead, it employs a peer evaluation model that has students act as each other’s graders.
Economics senior lecturer Rebecca Stein, who teaches a microeconomics course on Coursera, thinks that “something like this [automated grading software] could work for Coursera.”
In Stein’s Coursera class, she uses multiple choice quizzes and peer assessments to evaluate her students’ work.
She appreciates the peer evaluation model and thinks “students who do the grading are actually going to get a lot out of it.”
Engineering freshman and PennApps Lab Technical Lead-elect Steven Krouse agrees. “People always say you only truly learn something when you can teach it to someone else,” he said.
As for the machine grading technology, Krouse is skeptical but hopeful. He said it reminded him of similar software that evaluated some of his work in middle school. “Students figured out in thirty minutes that if you had really long essays with big words, they’d give you good scores even if it didn’t make sense,” Krouse explained.
Stein also mentioned that in her course, “at this point, there is no technology that can do what I want to be done.”
“I want students to face a blank page to answer questions using a model graphically,” Stein said, “What I need is a technology that can interpret a hand-drawn graph by students, and that does not exist yet.”
According to Computer and Information Science professor Ani Nenkova, who works in natural language processing, “computers are helping in many things, and this is just one thing that they can support us [in].”
Nenkova believes that machine grading, despite its shortcomings, has the advantage of “consistency and patience.”
“I really care about writing and teaching my students, and I don’t think in any time in my lifespan, a machine will be able to substitute [for] me,” Nenkova said, “but I think having a machine patiently point out things … can help people be better writers.”
Stein echoes this sentiment. “As much as I appreciate the expertise of my colleagues and myself in guiding students on their educational path, I would not be surprised if technology would help us and support us in the next 10 years,” she said.