Machines used to grade essays

Computers are used to write papers and now they can grade them too.

The Educational Testing Services e-rater essay grading machine was first deployed in 1999 and has been used in the past few years for standardized tests such as the GRE.

But now, through years of development, the software has become accurate enough to outperform humans grading standardized essay tests.

“For the kinds of essays that are on tests, the automated score will typically agree with a human score as often as or more often than human scores agree with each other,” said David Williamson, a senior development scientist in the Research & Development division at ETS.

“Essentially the way e-rater works is a conglomeration of many different separate scoring systems that work together,” Williamson said.

“There are several aspects that they do quite well on and several aspects that humans do better on,” he said.

Williamson said an online version of the e-rater essay evaluation program called Criterion is available and used by teachers to help college students with their writing.

The online program can help improve students’ writing after they submit samples for revision purposes, he said.

William Barney, a history professor at UNC, said he has compromised feelings toward machine essay grading instead of human grading.

“I don’t think anything can take the place of an instructor that carefully reads over a student’s paper and discusses it with him or her.” Barney said.

“On the other hand, if you are faced with many students, and it can help students improve their writing, it may be worth a try.”

Barney said he wasn’t sure a machine would be able to interpret creativity in a paper.

“Writing is as much an art as a science and my thought would be that there are nuances and rhythms and balancing that a machine wouldn’t know how to evaluate,” Barney said.

“However, if it was a choice between no editing at all, and the editing of the machine, then I suppose it would be OK.”