For many students, handing in an essay is the precursor to an agonizing spell of uncertainty. Professors, it often seems, take years to grade papers. One might as well be walking down the Lawn toward Cabell Hall, graduation tassels swinging, by the time a professor from first year hails you down clutching an essay flecked with red ink.
So an instant-grading system for essays, such as the software that education-technology company EdX recently unveiled, might seem appealing. But such a system achieves efficiency at the expense of human wisdom. When it comes to essays, in contrast to multiple-choice or true-false tests, we think getting feedback from professors — not computers — is worth the wait.
EdX, a nonprofit founded by Harvard and MIT, is most well-known for its massive open online courses. The company’s essay-grading program is its latest digitally ardent venture, as it and other online-learning firms seek to expand the role technology plays in higher education.
EdX’s automated-grading software, which the company plans to make available for free to any institution that wishes to use it, employs artificial intelligence to evaluate essays. The tool requires live teachers to first grade 100 sample essays, and the program models its grading of further papers on this limited sample. The program scores essays instantly and then lets students rewrite and resubmit to try to improve their grades.
Automated scoring of essays raises several problems. Machines cannot “read” in the way humans can. Computerized essay rating runs the risk of superficiality by focusing primarily on surface features such as word size and vocabulary level. As Les Perelman, the former director of MIT’s writing program, has shown, automated essay-grading software can be duped: A nonsense SAT essay Perelman wrote received a top score from an e-rater. Machine grading can miss errors that occur and find errors that do not exist, and such systems tend to grade minority groups and second-language writers unfairly.
But the larger problem, at least for those of us with a soft spot for humans, is that automated essay grading is a reductive approach to writing and to teaching. There is no one correct way to write an essay. A student can approach a given topic or argument in a variety of ways, all of which might be effective. A computer program cannot adequately take account of the opportunities for creativity and flexibility writing affords. Clear and convincing written communication requires not just precision but also imagination.
And there is more to pedagogy than mere assessment. An essay provides an opportunity for a student and teacher to grapple with the same set of ideas and refine an argument together. EdX’s automated system eliminates the personal element that makes learning extraordinary: the joy of discovery and the thrill of solving a puzzle under a teacher’s guidance.
EdX’s software is a symptom of an impatient time: many of us like responses to be immediate; whether they are thoughtful or not is another matter. But entitlement to instantaneity is not something we should encourage in students; patience, it seems, would be a better value to cultivate.
Other applications of automated grading benefit students by freeing up time for teachers to teach instead of, for example, putting forth the mechanical labor of flicking check marks on a Scantron sheet. EdX’s program, by contrast, diminishes student learning by replacing thoughtful if delayed feedback with sterile scores. Essay grading is a province human teachers must protect.




