Futurists and others hope for the day when everything we do is automated, when humans are living a life of leisure while our robot and AI servants do our bidding. This is not my area of expertise, and certainly, I’ve seen computers become capable of helping us do a multitude of tasks. Mapping, facial recognition, search, building cars, driving cars!–these are all things computers and/or robots are currently doing, and in some cases, doing well. People are also using algorithms to create art and poetry, activities we might think of as being uniquely human. So these things seep into areas that we might not think they can or even should.
My good friend, Audrey Watters, often writes about the ways that education relies on computer data-crunching to evaluate student learning or teacher performance. Now, I’m a big fan of data and what it can reveal, but I’m no fan of letting a computer program evaluate something as messy as learning all by itself. For one thing, most of these programs evaluate learning strictly through multiple choice tests, which can only tell you so much about how well a student is doing. Personally, I weigh process a whole lot more than product, which is what those tests evaluate without regard to process. I have no idea how a computer program is going to evaluate a project.
My former career was as a writing teacher and the grading of papers by computer is a fantasy many have longed for to alleviate the long hours of assessing student writing. Essay graders exist, but they aren’t that good, taking long words and more complex sentences as strong signals for good writing. Good writing is more than that, as anyone who writes or reads, much less teaches, knows. Still, there could be a role for computers in the writing process besides being used as a word processor or plagiarism checker. This article from Slate discusses how students respond better to computer feedback than teacher feedback. The reasoning makes sense to this for the same reasons people in general trust any data coming from a computer. Students think computers are more objective, maybe even more accurate than their teachers. And, they’re less anxious about getting feedback from a machine than a judgemental teacher. So maybe this is a reasonable role for computers and algorithms as part of the process, and not as the evaluator of the product.