Using AI and biometrics to enhance exam proctoring
This is a guest post by Oksana Mikhalchuk, a technology writer at Oxagile.
While a number of students of all degrees — from elementary school to Doctoral — are shifting to online and remote exams, proctoring providers face two challenges. First is to increase exam capacity, second is to keep the quality bar of proctoring results high.
To overcome both, a livestream camera is not enough. With thousands of students on board, no examiner can address identity, cheating, and privacy concerns efficiently at the same time. In the era of artificial intelligence and machine learning, biometric solutions seem to be the most suitable go-to, and there are good reasons for it.
Face recognition
Underpinned by ML-driven biometrics, face recognition can serve multiple purposes — from registering students to the exam and confirming their identity to securing their presence during the test.
Thanks to machine learning, face recognition can do more than validate the exam participants and make sure they are human. The technology is resistant to any type of identity mischief, and soon will be able to distinguish facial expressions. Glasses, new hairstyles and pictures instead of a human won’t confuse the system. That cuts off some popular cheating methods — like asking another person to take the test.
Student identification
Since identifying persons using their face raises privacy concerns and might be considered unacceptable in some cultures, education providers are searching for other ways — like behavioral biometrics.
Less invasive than the old-fashioned ‘ID-to-the-camera’ and less effort-consuming, behavioral biometrics analyzes unique human activities like typing and keystroke patterns. AI-enabled recognition compares a short phrase (written or typed) to a credible sample, and confirms the student’s identity at every stage of the examination.
The demand for visual and behavioral methods will grow because they are interchangeable and can suit special needs: disabled and limited mobility students may prefer face recognition instead of typing. During fundamental examinations like finals or accreditations, several methods can be combined to ensure double security. Finally, behavioral biometrics are easier to submit and feel less intrusive than an ID scan.
Action monitoring
An AI-enabled proctor views an exam taker as a set of behavior scenarios — body position, eyes and head movements, speech, etc. The red flags are deviations from a ‘normal’ pattern: unusual posture, hands under the desk, shifty eyes. Some students consider such an approach straight-up offensive — since it reduces a person to a number of algorithms.
The biometrics factor makes computer vision less mechanic-centered and more adaptive by tracking reactions and environment rather than movements — for example, a change in lighting.
Computer vision can already scan a student for hidden devices like a camera lens in a button or radio waves from any transmitting device. Each abnormal behavior incident means a penalty point, with no need to rewatch the whole recording — but the final review is up to the proctor.
Wearable technology
Trying to make supervised examinations a less intimidating experience, biometric proctoring leverages wearable hardware. Smartwatches and fitness bracelets can track movements, pulse and temperature to indicate abnormalities, while machine learning-enabled analyses it in real-time. Students are free to stay physically comfortable during an exam.
Wearable biometric-enabled technology combined with action tracking solves one more problem. It saves time for both proctors and students, eliminating the need to conduct visual inspections of student’s desks and rooms via a webcam.
But, with all the automation there is still a human point you can’t get around. There is no better way to reduce the stress of proctoring ‘invasiveness’ than to explain exactly what is going to happen — how their identity will be checked, and what parameters will be monitored.
Key takeaways
Biometric-enabled proctoring can cover more students with less human effort and presence involved — and make sure all participants feel comfortable and deliver the best results.
Enhanced with AI and biometrics, proctoring doesn’t interfere with a student’s experience, and becomes seamless. In particular, it eliminates identity confirmation struggles, reduces the number of fake cheating alarms, and makes remote and on-site proctored exams less stressful and more accessible.
About the author
Oksana Mikhalchuk is a Technology Writer at Oxagile, a New York-based provider of next-gen software engineering solutions around AI, computer vision, biometrics, and more. Oksana creates content about state-of-art tech opportunities in healthcare, education, entertainment, and more.
DISCLAIMER: BiometricUpdate.com blogs are submitted content. The views expressed in this blog are that of the author, and don’t necessarily reflect the views of BiometricUpdate.com.
Article Topics
artificial intelligence | biometrics | education | facial recognition | identity verification | machine learning | monitoring | remote proctoring | wearables
Comments