The Chronicle of Higher Education recently profiled a writing instructor, Les Perelman, and his team of Massachusetts Institute of Technology (MIT) and Harvard University students who have created their own electronic essay generator named “Babel.” As a critic of automated grading software, Perelman and his team designed Babel to prove that grading software currently used by testing companies cannot “tell gibberish from lucid writing” by checking facts or reading for meaning.
Babel generates original essays using as little as three key words, producing sentences that are grammatically correct yet meaningless. They consistently earn high marks when graded by software such as “MY Access!,” the same program the Graduate Management Admission Test uses as a second reader. Despite these alarming results, Perelman remains cautiously open-minded about the potential for new software.
A nonprofit cofounder of Massive Open Online Courses (MOOCs) known as edX has developed its own system with MIT called the Enhanced AI Scoring Engine (EASE). EASE attempts to mimic professors’ grading styles when scoring essays. A professor grades several essays using his or her own criteria and scans them into EASE. The program then reviews the professor’s marks for patterns and assimilates them.
Some professors who instruct MOOCs have experimented with EASE and received satisfactory results. At times, they found the software overly generous or stingy with grades, but many agreed that it could potentially be a useful grading tool for MOOCs as a supplement to peer grading.
