Email
Password
Remember meForgot password?
    Log in with Twitter

article imageArtificial intelligence used to mark exam papers

By Tim Sandle     May 29, 2018 in Technology
Trials in China have shown how artificial intelligence can at least match, and sometimes surpass, the ability of teachers to mark exam papers. One in every four schools is evaluating machine-led technology.
Whereas artificial intelligence is being used in some countries to mark multiple choice questions, China is experimenting with machine intelligence to mark essays. According to the South China Morning Post, technology has been developed to interpret the general logic and meaning of the text. The platform can then undertake human-like judgment into an essay’s overall quality. The platform can then assign a grade to the essay and also provide recommended for improvement, selecting categories such as writing style, sentence structure and overall theme.
At present the application of artificial intelligence in Chinese schools is assisting an assessment by a teacher and not removing the teacher from the equation. However, the technology is helping to cut down the amount of time a teacher spends grading an essay and is aiding teachers in ironing out student-to-student marking inconsistencies. The current rate of agreement for grade outcomes between the platform and human teachers is 92 percent. This is based on trials in 60,000 schools involving over 120 million students.
The software is designed to keep on improving, according to Venture Beat. The grading software utilizes deep learning algorithms to self- compare its outcomes with human teachers’ scores. It can also take into account suggestions and comments from human teachers.
The U.S. has also been using artificial intelligence to assess essays, although to a more limited degree. A system designed to help teachers mark essays faster from the University of California Berkeley, called Gradescope, is one such example. A second example is with a platform, which functions as an automated reader, developed by the Educational Testing Service and called e-Rater, can grade 16,000 essays in 20 seconds.
The e-Rater platform was recently assessed at the University of Akron. Here software-generated ratings against 22,000 short essays from junior high school and high school sophomore students compared favorably to the ratings given to the same essays by trained human readers, suggesting a future role for such technology within the education system.
More about Artificial intelligence, Exam Papers, Teachers, Teacher
More news from