The risks of robo-marking: AI in Australian schools

Automated grading technologies are common in educational institutions overseas, and will likely arrive in Australian schools over coming years.

While “machine marking” or “automated essay scoring” (AES) could potentially alleviate teacher workloads, experts warn that without robust national guidelines and an independent advisory body, some schools could be left behind.

AES is increasingly used in countries such as the United States, in a range of high-stakes testing contexts where results determine student progression and school funding.

A white paper from the University of Sydney suggests that this kind of technology in Australian schools could cause controversy without careful governance and consultation with the education community.

“We know teachers are already experiencing heavy workloads and this new technology could help ease the pressure, so long as the implementation doesn’t create even more work,” said Professor Kalervo Gulson, from the School of Education and Social Work.

“We don’t recommend automated grading in high-stakes testing where the results might impact the student’s future or the school’s funding. And like all types of artificial intelligence in schools, it is important that the people impacted by it understand what it will do, and have a say in its introduction.”

Co-author Professor Greg Thompson, from Queensland University of Technology, said: “We should continue to be cautious about any technology that relies on opaqueness being used to make consequential decisions in schools. More time and effort is needed to be spent on opening these systems up to improve both scrutiny and understanding.”

White paper recommendations include:

• Establishing uniform, ethical guidelines for schools purchasing AES systems from commercial edtech companies.
• Ensuring all schools — including rural and regional — have the necessary digital and human infrastructure to support implementation.
• Building transparent best-practice guidelines for the use of artificial intelligence in schools.
• Initiating an independent advisory body for large-scale assessment in Australia.
• Establishing a risk-based framework that ranks the potential harm of AES adoption in specific assessment contexts and school settings.
• Learning from large-scale rollouts of AES in the US and other jurisdictions, including ethical, legal and financial implications and the impact on school, department and regional decision-making.
• Utilising stakeholder expertise across multiple locations and decision-making levels, including educators in schools and classrooms, prior to rollout.
• Not using AES in high-stakes testing where outcomes are consequential to individuals or to schools.
• If AES is to be widely adopted, Australia must lead the way in the collaborative development of AES guidance, policy and regulation.
   

The white paper was developed by the Education Futures Studio and the Education Innovations research program, funded by Sydney Social Sciences and Humanities Advanced Research Centre.

Image credit: iStock.com/metamorworks