Mettl’s assessments have been the biggest filter in our recruitment process. Their platform has helped us reach out to a higher volume our applicant numbers. Mettl constantly keeps innovating on their products and tries to introduce a new aspect to everything.
The evolution from the pen-and-paper approach to the online model and eventually to AI-based (Artificial Intelligence-based) exams has acted as a cornerstone of education. The ease of use on-the-go, access to resources, exposure to social media platforms and credible outcomes are boosting online education in an unprecedented manner. AI’s infusion in the education ecosystem has revolutionized higher education and the change has been wholeheartedly embraced by universities that leverage AI-based exam technology to advance their curriculum and offer students the flexibility to take the exams.
AI enables machines to learn from experiences, adjust to the environment, thereby perform human-like activities. Machines can be trained to accomplish specific tasks by processing vast data and recognizing patterns, using deep learning algorithms and neural networks. Machine learning’s increasing adoption in AI-based online exams is based on the following interesting features:
The steep surge in the adoption of AI in exam automation has led universities to scale online education. Center-based tests are exorbitant for students and take a toll on a university’s budget in hiring proctors, finalizing a venue and making other logistical arrangements. Meanwhile, the availability of proctors for live proctoring has become a limitation – add to that the additional costs involved. However, with AI, universities are increasingly embracing exam automation, and auto proctored tests as they are the most financially viable solution and maintain integrity.
Auto proctoring is a boon for universities as it offers credible, time and cost-effective solutions via AI-based exams. Auto proctored exams reduce the cost by a third, making it an affordable technology. The auto proctor acts like a human invigilator in exams taken by students from any part of the world. The technology has become reliable due to its interactive features that help create a cheat-proof environment.
AI-based exams are now equipped with advanced facial recognition features, owing to intensive research by engineers and neuroscientists in the recent past. Universities use AI proctoring units that rely on facial recognition to ensure students don’t employ unfair means to ace an exam. The accuracy and precision of the detection are directly proportional to the volume of images fed to the system – the system red flags cheating instances. Authentication via government authorized photo ID further lowers the probability of impersonation.
The GaussianFace algorithm developed in 2014 by researchers at Hong Kong University achieved facial identification scores of 98.52% compared with the 97.53% achieved by humans.
Researchers have long been working on voice recognition to enable systems to understand human speech. A voice-enabled auto proctor identifies audio for authentication and matches it to any background noise to eliminate cheating instances by providing a graphical representation of the speech patterns.
As of May 2017, Google’s machine achieved a 95% word accuracy rate for the English language due to its machine learning algorithms that can detect speech and respond with meaningful results. This accounts for nearly a 20 percent improvement since 2013.
AI accelerators are driving digital transformation outcomes for enterprises. Multiple technological advancements have led AI to take the next step by eliminating the need for human invigilators for creating student report cards. Universities are sent system-generated reports based on students’ performance after the exam is over. Since technology has improved the integrity and credibility of the exams, universities view AI-based exams as a one-stop solution for providing an efficient, affordable and scalable solution for online education.
With the increasing demand to provide quality education to a vast number of students, the dependence of technology multiplies to share the human burden, eliminating errors, using exam automation. The pace at which AI’s adoption has gained momentum, manual and live proctored tests may be completely phased out in the next five years, paving the way for a fully automated examination model.
Students can cheat on a proctored exam. The different methods used for cheating include impersonation, suspicious behavior such as gesturing, external help and easy navigation.
This happens mostly in high-stakes exams when students employ an impersonator to appear for the exam, especially when institutions are not well equipped to deal with online authentication at remote locations.
Exam automation from homes has bolstered students’ motivation for cheating, where they can seek their friends’ or family’s help to find the correct answers. Gesturing to seek answers is a common problem.
Unlike the traditional examination where students aren’t allowed to carry electronic devices to the examination center, exam automation has allowed students to use smartphones or wireless Bluetooth devices for cheating.
Students misuse social media and other websites or applications to copy-paste answers. They go to great lengths by even connecting external devices.
Online exams can be conducted without cheating by employing AI-based exams and AI proctoring features, such as Mettl Secure Browser (MSB) and Mettl Proctoring as a Service (MPaaS). While the secure browser disables test-takers from screen-sharing and navigating outside of the test window, MPaas provides a secure virtual environment with world-class anti-cheating features.
Mercer | Mettl’s online examination system, with AI, enables educational institutions to conduct online exams with ease. ‘Examine’ is a scalable and easy-to-use examination platform that allows you to create and host exams and certifications. It offers a complete suite of Mercer | Mettl’s pre-configured proctoring technologies, including a robust security mechanism. It offers you an end-to-end secure examination solution. Mettl Proctoring as a Service can be added easily onto any client platform through API. You can use it if you have an existing examination or LMS and need a seamless layer of proctoring technology to create a highly secure and well-invigilated examination experience.
MPaas’ three-point candidate authentication ensures the utmost reliability and authenticity of the candidate. The steps include candidates clicking their picture, showing their ID proof and proctors identifying the ID against the registration details.
Our exam automation technology is hands-on at identifying any suspicious behavior. It auto-flags various doubtful cases with 95%+ accuracy, including the non-visibility of face or presence, detection of a mobile phone, and additional person, negating any possibility of malpractice during an online proctored exam.
Mettl Secure Browser pre-defines the applications and websites to be blocked or whitelisted. The browser lockdown feature restricts candidates from opening any unwanted software or app, disables screen recording and screen projection.
Easily record and review entire test sessions, complete with annotations and automatically flagged activities such as smartphones, distracted eye movement, another person’s presence, etc. - an excellent exam automation feature.
Mercer | Mettl’s Examine effortlessly integrates its proctoring services with over 250 LTI compliant systems instantaneously, such as Canvas, Moodle, D2L, Blackboard and more.
Originally published September 11 2018, Updated December 16 2020
A writer at heart, Megha has been in the content industry for 4 years. Starting her career from print, her journey spans across IT, legal and consulting industries. She has been associated with Mercer | Mettl as Assistant Manager, Content Marketing for 2 years.
Online examination, also known as virtual examination, is conducted remotely on a computer with high-speed internet. Like a classroom exam, it is time-bound and usually supervised through a webcam and proctor, making it cheating-free, secure and easily scalable.