Has artificial intelligence reached the point where it can assess and provide feedback on student essays?
Part 1 of 2
Beware the paper tsunami!
Almost without fail, if you ask English teachers what the most arduous aspect of their job is, they will tell you it is the slog of reading student essays and providing meaningful feedback. It’s not that they don’t want to teach essay-writing skills to their students. They do. And it’s not that providing feedback is a drag in and of itself. Helping kids become more proficient through coaching providing advice is absolutely rewarding. What makes assessing essays tough is the workload itself.
Now, before you start thinking, “suck it up and do your job” (and I admit that a teacher complaining about reading and assessing student work does at first sound a little precious at first), consider the following:
I have five English classes, each of which has a about 40 students. When I assign I an essay, the average length of the finished product is about three pages. So, every time I assign an essay, I have 600 pages to read, assess and provide meaningful feedback on. If I spend 10 minutes per essay, I’ll spend 30+ hours going through those papers. And that’s just one assignment!
These statistics are what keep English teachers up at night, and this is why writing instruction is being short changed in our schools. Kids should ideally write at least four short-form and two long-form essays a month. That’s not happening. To be honest, I know teachers who simply do not assign longform essays anymore because of the workload it creates.
One very-reputable teacher I know describes the experience as imposing on oneself a self-inflicted paper tsunami without a life preserver. “Each school year, I make an attempt to teach essay skills”, she says, “and everytime I make it about two essays in before I become completely swamped. And it’s not getting any better with ever-increasing class sizes.” Teachers want to do write by their students, but they also want the workload to be reasonable. They want relief.
When frustrated teachers discuss these issues, the subject of artificial intelligence (AI) always comes up. We all know that super-fast, computer processing power and clever, self-learning algorithms are changing the world around us. Computers can beat chessmasters, drive cars (almost without incident) and simulate human conversations. So why can’t AI be developed to grade essays and alleviate teachers of the guilt caused by cutting back on written assignments? Where are the gradebots?
Some point out that AI for teachers is already here.
Rhode Island Education Commissioner, Ken Wagner, certainly seems to think so. He’s claimed that “the research indicates that AI technology can score extended student responses with as much reliability- if not more reliability- than expert trained teacher scores.”
This sounds very encouraging, yes? And, check out the Intelligent Essay Assessor that education mega- giant, Pearson, has come up with:
“The Intelligent Essay Assessor (IEA) is an Internet-based tool for automatically scoring the quality of electronically submitted essays, IEA uses Pearson’s state-of-the-art Knowledge Analysis Technologies™ (KAT) engine, which automatically evaluates the meaning of text, as well as grammar, style, and mechanics. What’s more, IEA can also evaluate short constructed responses. In tests with thousands of constructed responses, the Intelligent Essay Assessor has proven as reliable as professional human scorers.”
Pearson happens to be one of the biggest players in the Common Core assessment market and uses this type of technology in the scoring of high-stakes, end-of-year essay-writing tasks taken by kids in multiple grade levels in 21 states. It’s counterpart, the SBAC, put out by ETS, assesses kids’ writing in 19 states, and it also uses automated (AI) scoring.
So, a vast majority of our nation’s students are not having their writing assessed by human graders, but by smart, algorithmic engines, and, if it’s true, and not science fiction, then its welcome progress in terms of giving teachers that assistance they need to take the pressure off the assessment load.
It seems like the next logical step is to have such tools actually available for frontline, classroom teachers. I can see teachers purchasing this type of software on their own if the price was right, or, even better, schools purchasing licenses from companies like Pearson and ETS for teachers and even students and parents to use. Teachers would assign essays more frequently and the feedback would being given instantly. But does this AI technology really work?
There are skeptics out there and their concerns will be dealt with in part two of this feature.