Should AI Grade Your Work?
Human grading isn't neutral either. Neither answer is the easy one.
Should universities let AI grade student work? The honest answer isn't "yes" or "no" — it's about who's actually accountable when it gets something wrong.
A workable case for using AI grading responsibly, not a blanket yes or no.
Human grading gets treated as automatically fair, but tiredness, workload and unconscious bias all creep in. AI grading is more consistent, but it comes with its own problems — it can be confidently wrong, hard to see inside, and it can't actually be held responsible for a decision.
Research on bias in algorithms, how often AI tools get things confidently wrong, how transparent (or not) these systems really are, and how universities currently handle accountability.
The real question isn't which system makes fewer mistakes — it's which system's mistakes are easier to catch and fix. AI bias can actually be tested and audited in a way human bias can't.
Put both systems side by side against the exact same standard — fairness, transparency, accountability, the ability to correct a mistake — instead of treating them as two separate debates.
The whole project is about this — a proposed setup where a human always has final say, but the system itself is constantly checked and open to challenge.
Four working principles: a human stays accountable, the system stays visible, it's monitored continuously, and students keep a genuine right to appeal.
"The most interesting question about adopting AI is rarely about accuracy — it's about who's actually on the hook when it goes wrong."
Read the complete report submitted for this project, including the research, analysis, strategy, and final recommendations.