EssayGrader promises that you can ”upload all your students’ essays, then come back a few minutes later and find them all perfectly graded” and ”95 per cent time savings.” Similar programs currently on the market include Writable, Crowdmark and Gradescope. The technology magazine Axios writes that some teachers in the USA have already started using these tools.
Its examples come from high school level, but Gradescope markets itself primarily to the higher education sector. Meanwhile, in December, a research group and at the University of Surrey in the UK launched the Key Evaluation and Assessment Tools Hub (KEATH), a program they claim is able to match human teachers’ previous grading and comments. The researchers claim that KEATH will “free up time for both student contact and research.”
“I know about the programs, but have not tested them myself,” says Ingrid Forsler, who conducts research on academia’s use of AI at Södertörn University. “There are probably contexts where they could make reasonable assessments, such as short-answer questions where you have to define a concept. I am more sceptical when it comes to longer, essay-type questions.”
Ingrid Forsler
Södertörn University
She points out that AI-generated texts tend to look correct, without necessarily being so. “One danger is that AI marking similarly prioritises answers that look good and use the ”right words”, but do not really show understanding. At worst, that penalises originality; the system does not recognise good answers that deviate from the typical. But of course you can imagine some kind of combination where a person has the last word. We will see whether it then saves time. We might see teachers getting more work, training the AI system instead.”
In Sweden, Karlstad University introduced an AI tool in March which should be able to summarise course literature and look for language errors in texts that have already been written. However, one student interviewed by Swedish television wondered whether both students and teachers would now ”become very lazy”.
Erik Svensson is a professor at the biology department at Lund University.
“We already use AI,” he says, “for example to identify species in photos. My initial reaction is that this sounds quite attractive. Marking exams is one of the most boring parts of our job. But if a student questions the assessment, then we can’t hide behind an algorithm.”
”My initial reaction is that this sounds quite attractive. Marking exams is one of the most boring parts of our job.”
Erik Svensson
With big promises of time savings, will teachers soon even be expected to hand over marking to a digital assistant?
“I think the universities will eventually integrate such programs into digital examination systems,” says Ingrid Forsler, “but whether it will be mandatory or optional to use them, I dare not say.”
Erik Svensson makes a comparison with digitalisation in general.
“We were also told that would save time, but instead it became a way to shift more administration onto teachers, so I now take promises that new technology will free up more time for research and the like with a pinch of salt.”
