REMOTE PROCTORING USING AI – ENABLING SEAMLESS MANAGEMENT OF ONLINE EXAMINATIONS

Since the education industry is experiencing an enormous transformation with emerging technologies, educational institutes are looking to conduct semester-end and entrance examinations remotely. While many schools are shut down amid the COVID-19 outbreak, many universities have replaced examinations with an assignment that students can copy and paste from the internet. ETS that conducts GRE and TOEFL, among others, is allowing students to give exams from home where a proctor monitors them for the entire duration of the exam. Adopting this technique is not viable on a large scale.

Educational institutions need a solution that does not let them compromise with the security, experience, convenience and feasibility of educators and students. With numerous exams being postponed or canceled amid COVID-19 risk to students, remote proctoring can facilitate them to give exams in the safety of their homes.

With AI-based remote proctoring and invigilation technologies, it can be ensured that students do not get indulged in cheating or unfair means during the examination.

With a combination of manual and AI-based technologies, remote proctoring offers various benefits. While it allows students to take a test from any location with specific technical prerequisites, it also removes the need for physical examination centers.

“AI-proctored exams are a great way to ensure that the lockdown doesn’t mar the aspirations of students and that the academic calendar is followed in due time,” says Randhir Kumar, founder of BasicFirst Learning. Adds Shweta Doshi, co-founder of career-focused edtech platform GreyAtom, “It’s an exciting time for educators to challenge the status quo on how we assess our learners.”

What is Remote Proctoring?

Remote proctoring is the process of authenticating, authorizing and controlling the online examination process in a scalable manner. It is a technology that allows organizations to enable assessment anywhere and anytime, ensuring full security standards.

In other words, candidates don’t need to come to a specific place as they can give examination from their homes.

In the traditional exam process, an invigilator has to be present at the exam center to check candidates appearing for the exam. To examine 30-40 candidates, you require one invigilator. However, to conduct an exam of 1000+ candidates, you would need more than 25 invigilators controlling the exam process.

Online proctoring can be conducted through the internet via the web camera of the candidate. It can record every single examination session from beginning to end, not just via video, but also captures desktop screens, chat logs and images.

Proctoring can be classified into different types which are as follows:

• Video Proctoring
  Video Proctoring is useful for high stake examinations where a candidate is monitored via a continuous video streaming activity. A candidate’s video during the entire exam is recorded and the assessment controller checks if the student got involved in cheating or unfair means by analyzing their behavior from the video.
• Image Proctoring
  Image proctoring is suitable where internet connectivity is not good. This proctoring type aims at verifying remote candidates multiple times randomly. The system would capture pictures of the candidate at specific time intervals, for example, exam start, end, questions attempted, after every 30 or 45 seconds.
  Educational institutes can validate those images to ensure that an actual candidate has conducted an online exam and no malpractice has happened. Image proctoring is cost-effective as compared to video streaming quality.
• Auto Proctoring
  If you want to do monitoring and analysis activity automatically for remote candidates, you can perform auto proctoring. It is used to conduct the continuous streaming activity of candidates sitting at remote locations for online assessment.
  It performs the analysis of videos and images to identify if the candidate is indulged in cheating, such as the use of the mobile phone during the exam, someone assisting the candidate or candidate is using notes or reference books.
• Candidate Identity Verification
  In this proctoring, the identity of the candidate is verified before the start of the online exam. The candidate is supposed to show an identity card and exam hall ticket in front of the camera
  The proctor sitting at a remote location verifies the identity card of the candidate and approves or rejects them based on submitted records.

As compared to the above methods of proctoring, automated proctoring is one of the best ways that use a webcam and screen-sharing program. In this proctoring, a human proctor is replaced with computer algorithms that can flag suspicious behavior.

Using ML, AI-enabled remote proctoring systems can continuously learn, adapt and get smarter. The aim to introduce AI into proctoring is not to replace humans but to increase the accuracy of proctoring by helping humans in identifying details like low sound levels, whispers, reflections, shadows, etc.

How AI-based Remote Proctoring works?

• Detect Identity Fraud
• Analyze cheating behavior
• Discover Content Theft

The AI-based remote proctoring process is repeated thousands of times to develop, train and refine every event defined in the system. An event can be a single behavior or indicative of identity fraud, content theft and cheating behavior.

For example, if someone is found looking off-screen to the left, it can be considered as a single data point and that specific portion of the video is segmented and labeled as unfair means. Once the number of such data points of the same behavior goes beyond the limit, the continuous event of building, training and refining is initiated.

Each of the thousands of events that execute through the process is categorized as potential fraud, theft or cheating. All the events would result in whether or not the session should be marked with a suspected breach of integrity.

A variety of AI technologies can be used to enhance remote proctoring services and provide an efficient way of organizing examinations to institutions.

AI technologies used for Remote Proctoring

• Pattern Recognition
  Every act of cheating depends on the pattern of specific behaviors. AI can detect patterns within data. By looking at the available data, it tries to identify whether there are any regularities within it.
  Pattern recognition is defined as the classification of data based on already-gained knowledge or statistical information collected from patterns and/or their representation
• Voice Recognition
  Voice Recognition technology helps pick up sounds and match it with the background noise to remove instances of cheating by recognizing speech patterns.
  It is useful for finding out differences between sounds that should and should not be in a testing environment.
• Facial Recognition
  From identity verification to detecting new faces within a testing environment, facial recognition can be used in multiple ways. It can recognize various faces at a time and identify if anyone assists the candidate during the online examination.
• Eye Movement Detection
  AI-integrated Eye Movement Detection can help recognize eye movement patterns to identify if the candidate is looking straight towards the screen or is looking into any object, book or mobile phone. Eye movements indicating misconduct are gathered to identify if the candidate is involved in any unfair means.
• Plane Detection
  Using Plane Detection, the remote proctoring software can discover the spatial definition of the physical environment being used by the candidate. When it is combined with object recognition, the system will gain a better understanding of everything in a given space.
• Mouth Detection
  It is similar to eye detection where facial keypoints are used and test-takers are required to sit straight. The distance between the lip key points can be noted for 100s of frames. The infringement is reported if the distance between the lip points of users increases than a specific value.

How can AI improve Remote Proctoring Services?

• Improved Accuracy
  Artificial Intelligence is like a second set of eyes for human proctors as it can generate an alert for anything it sees and catch things that humans may not have time to recognize in real-time.
• Additional Scalability
  Efficiencies obtained through improved accuracy in recognition of an unusual behavior allows remote proctoring software to proctor online assessments without compromising the additional security capability of human intervention.
• Unmatched Security
  If the system catches a negative behavior or unpermitted resource that a human proctor may miss, the remote proctoring system can take immediate action against the candidate. It provides an additional level of confidence in high-stake exams where the prevention or remediation of misconduct is the preferred outcome.
• Mimic human behavior
  AI can constantly help alert humans to aberrant behaviors. Once the system becomes as accurate as humans, you won’t need humans to look after every single candidate.
• Creating a smarter AI system
  An AI model only works perfectly when it is accurate. The system requires hundreds of data points to become accurate. Therefore, you will have a smarter AI system the more you have the data to work on.

Conclusion

Remote proctoring using AI can transform the education sector and has made everything possible virtually. AI-integrated computer systems can ensure the authenticity of the test by preventing the candidate from cheating and indulging in unfair means during the assessment. With Remote Proctoring, educational institutes don’t need to delay or postpone examinations amid the COVID-19 outbreak.

If you are looking to build a remote proctoring software for your institution, consult our edtech experts who can help you create a custom proctoring solution to conduct online examinations.