The article discusses the potential and limitations of using AI, specifically ChatGPT, to grade student essays. Early research shows that ChatGPT's grading is almost as accurate as a busy teacher, but not yet reliable enough for high-stakes grading. Researchers, like Tamara Tate from UC Irvine, found that while ChatGPT can grade essays instantly without prior training, its scores often cluster in the middle range and can miss nuances that human teachers catch. Though AI could reduce teachers' grading workload and encourage more writing assignments, there's concern it might harm writing instruction if teachers rely too much on it. Future improvements in AI accuracy and further studies on its impact on student writing are anticipated.
Link to full article: PROOF POINTS: AI essay grading is already as ‘good as an overburdened’ teacher, but researchers say it needs more work