Trends Of Artificial Intelligence (AI) For Online Exams

Contrary to the common belief, teachers also experience exam anxiety; exams are logistically challenging and demand extensive planning. Setting up a question paper, or rather several different sets of question papers, printing those question papers and answer sheets, securely delivering those question papers to exam centers, gathering those papers, sending them to teachers for correction, and then publishing the results are all steps in the process.

There is a pressing need for innovative exam administration methods since teachers must spend a lot of time planning, administering, and arranging exams rather than using that time to better support their students and classrooms.

Simply said, the Indian educational system is enormous. Out of the three hundred and fifty million people enrolled, seventy million have entered higher education alone in the last two decades. Now, it is a very difficult undertaking for many state and central education boards to plan and perform tests for millions of pupils. Paper leaks have occurred during exams administered by several state boards and the Central Boards for a variety of reasons. A lot of time, money, and labor are needed. When one exam is finished and its results are out, it’s time to begin planning for a new exam.

Many universities grapple with paper checking and publication issues in addition to paper leaks. The lengthy process of reviewing thousands of papers might cause findings to be months late due to the continuous instructor shortage. For instance, the Mumbai University results were not released for five long months, which caused a sense of apprehension among students and their parents. Similar issues exist with college-level exams, which makes administering them a difficult undertaking.

Advertisement

What’s next?

Online tests largely depend on artificial intelligence (AI) and machine learning (ML) to ensure security and stop cheating. The use of facial and speech recognition technology is one way that AI and ML are deployed in online exams.

•     How can we make conducting exams easier? 
•     How AI is bettering online exams
•     Onscreen Marking System
•     AI-powered tools to evaluate answer sheets

The AI-based exams commonly alluded to as AI-proctored tests, do away with the requirement for placing onsite invigilators and offer a setting free from plagiarism. The webcam stream from the candidate is continuously monitored by the AI proctoring program, which flags or reports any suspicious activity. Exams powered by AI make it simple to keep track of lots of applicants.

Universities have expanded online education as a result of the sharp increase in the deployment of AI in exam automation. Center-based exams are expensive for students and drain a university’s finances because proctors must be hired, a location must be chosen, and other logistical considerations must be made. However, as auto-proctored exams and exam automation become more commonplace thanks to AI, colleges are doing so because these methods are the most financially sound and maintain academic integrity.

❖    Extraordinary Features of AI-Based Exams

Universities benefit from auto proctoring because it provides reliable, efficient, and economical solutions through AI-based tests. Exams that are automatically proctored cost a third less, making this a more economical technology. When students from all around the world take exams, the auto proctor functions like a human invigilator. Due to its interactive characteristics that aid in producing an environment free from cheating, the system has acquired credibility.

⮚    Image Recognition

In light of recent rigorous research by engineers and neuroscientists, AI-based tests now come with sophisticated facial recognition characteristics. Universities deploy facial recognition-based AI proctoring devices to make sure students don’t cheat on exams. The volume of photos provided to the system directly correlates to the detection’s accuracy and precision; the system highlights instances of cheating. Impersonation is also less likely when a government-approved photo ID is used for authentication. Compared to the 97.53% attained by humans, the Gaussian Face algorithm created in 2014 by researchers at Hong Kong University achieved facial identification scores of 98.52%.

Advertisement

⮚    Voice Recognition

To assist systems to comprehend human speech, researchers have long worked on voice recognition technology. By showing a graphical representation of the speech patterns, a voice-enabled auto proctor may recognize audio for verification and match it to any background noise to eliminate instances of cheating. Accelerators for AI are helping businesses transform digitally. As a result of numerous technical developments, AI has advanced by doing away with the necessity for human invigilators to create student report cards. After the exam, system-generated reports based on the performance of the students are delivered to the universities. Universities regard AI-based tests as a one-stop solution for delivering an effective, economical, and scalable option for online education since technology has increased the integrity and legitimacy of the exams. The dependence on technology grows as a result of the rising requirement to deliver high-quality instruction to a sizable student body, eliminating errors and adopting exam automation. Given how rapidly AI is being adopted, manual and live proctored exams may be phased out over the next five years, opening the door for a completely automated test procedure.

Whether it’s for educational institutions or corporate hiring, integrating AI into the assessment process offers certain special advantages that are not possible with conventional methods. A startup using artificial intelligence to power exams enables organizations to administer the evaluation at scale. This was born out of the necessity to ensure that every student had a personalized learning experience and to establish a system where each student is routinely evaluated based on their abilities. AI is used in many psychometric tests for job candidates and employees, even in the corporate world. In situational judgment tests (SJT), this can take the form of convincing algorithm-based judgments derived by examining test-taker replies, or it can take the form of genuine chatbot-style interactions with applicants. HR and talent choices are increasingly frequently influenced by the implementation of AI in assessment.

• Re-evaluating the assessment process

When it involves hiring in corporations or educational institutions, AI integration offers certain special advantages that cannot be obtained through conventional methods.

The first on the list is that AI can analyze enormous amounts of data with improved efficiency and precision—far more so than any human can. Today’s computers are more powerful, thus more candidate data can be precisely processed in less time.

Second, AI confronts the prejudices and stereotypes held by people that frequently manifest during appraisal. It results in arbitrary marking in the educational setting, but it might result in bad hiring decisions for businesses. Nevertheless building this trust requires that the AI system’s programming be done impartially. The algorithm will always produce biased results if the input data is biased, to begin with. Finally, AI provides socially-distant yet trustworthy evaluation methods without compromising the sanctity of the assessment process in a time when being near to one another is considered the greatest of all crimes. Artificial intelligence (AI) has a significant role to play in evaluating enormous amounts of candidate data by merging many features, such as robotic process automation, machine learning, pattern matching, natural language processing, etc.

Examiners can use the resources provided by automated proctoring programs to stop cheating. The software can log system data, restrict online access, and track keystrokes. To record test takers and their surroundings, they can also seize control of computer cameras and microphones.

Advertisement

AI is used by certain programs to “flag” dubious behavior. Facial recognition algorithms verify that the student is still seated and that nobody else has entered the room before proceeding. The algorithms also pick up on other behaviors that can point to cheating, such as whispering, weird typing, strange motions, and so on.

Examiners can expand their investigation by checking previously recorded video and audio and questioning the student once the program “flags” an occurrence. A necessity during the pandemic, automated proctoring software claims to lower exam cheating when given remotely. Fair tests safeguard the value of credentials and convey the importance of academic integrity. They play a significant role in the certification criteria for professions like law and medicine. Honest students are wronged by cheating. If left unchecked, it provides these pupils more motivation to cheat.

• Security

Simple technical approaches can get around a lot of the anti-cheating safeguards, according to our evaluation of the software. This result indicates that the tools might only offer modest advantages. It poses a security concern to make pupils install software that has such extensive computer control. In some circumstances, even after students uninstall the software, it slyly persists.

• Access

Some pupils might not have access to the required hardware or the quick internet connections the software needs. This results in technological problems that are stressful and detrimental. 41% of the pupils in one incident had technical glitches.

• Privacy

Online proctoring raises privacy concerns. Examiners can peer into students’ houses through video capture and study their faces covertly. It differs from conventional in-person test supervision in that it is conducted with such close observation and is videotaped for potential later viewings.

Truth and bias Significant fairness issues are raised by proctoring software. The software we studied uses facial recognition algorithms, but they aren’t always reliable. The algorithms utilized by the big US-based manufacturers do not distinguish darker-skinned faces as correctly as those with lighter skin tones, according to forthcoming research by one of us. The resulting covert discrimination may exacerbate existing biases in society. Similar issues with proctoring software and facial recognition technology, in general, have been identified by others.

Concerningly, the proctoring algorithms can mistakenly identify test-takers who exhibit unusual eye or head movements. This could give rise to unfounded assumptions about students who are neurotypically different or who have unusual exam-taking habits. Exams are already stressful experiences that have an impact on our behavior, even without automated proctoring.

• Investigating baseless suspicions

Educational institutions frequently have the option to accept or reject certain automated functions. The proctoring corporations may assert that AI-generated “flags” are just grounds for the school to look into possible academic dishonesty and not evidence of it.

However, when based on erroneous machine-generated suspicions, looking solely into and questioning a kid can be unfair and distressing in and of itself.

Advertisement

• Surveillance culture

Automated exam monitoring may also set a wider standard. The general public is becoming more concerned about surveillance and automated decision-making. When introducing potentially dangerous technologies, especially when they are imposed without our genuine agreement, we should exercise caution.

It is essential to devise methods for remotely grading exams fairly. Exams won’t always be able to be substituted by other forms of testing. Nevertheless, organizations utilizing automated proctoring software must be responsible. This entails being open and honest with pupils about how technology operates and potential outcomes for student data. Examiners should also provide useful alternatives, such as opportunities for taking the exam in person. Providing alternatives is fundamental to informed consent.

In China, one out of every four schools uses AI to assess student homework. This system, which uses machine learning, can automatically grade students’ work and, in some contexts, even make recommendations.

Some online grading tools can even read and comprehend the handwriting of the candidate while scoring the test papers nearly as well as teachers! These tools make it simple to discern letters, numbers, and other symbols.

These AI-powered gadgets also have the benefit of learning, just like people do. So, thanks to machine learning, if the system makes any missteps and those mistakes are reviewed and rectified by teachers, the system won’t repeat them.

These tools are also considerably quicker than teachers; some of them can analyze answer sheets in as little as 90% less time. As a result, results might be released shortly after the exam is over. This can save a tonne of time and work while also addressing the university’s lack of evaluators. The use of AI for the automatic assessment of descriptive responses in online tests has several benefits. Among the advantages are:

Advertisement

• AI algorithms can efficiently and accurately assess a high volume of exam responses, which reduces the time and effort needed to mark exams and frees up teachers to concentrate on other elements of instruction.
• Increased accuracy and efficiency when marking examinations.
• Improved fairness and objectivity in grading. All students receive the same degree of evaluation thanks to the objectivity and consistency of AI algorithms, which also reduces the possibility of bias or human error in grading.
• For students, detailed and personalized comments. The strengths and flaws of each exam answer are highlighted, along with recommendations for improvement, by AI algorithms. This can aid pupils in bettering their abilities and comprehension of their performance.
• enhanced information reporting and analytical capabilities. Exam answers can be used by AI algorithms to create comprehensive data sets, which give teachers the ability to monitor academic achievement, spot trends, and patterns, and decide on teaching and learning strategies.

Overall, the automatic evaluation of evocative answers in online tests using AI can have many beneficial impacts on both students and teachers, strengthening the efficiency, timeliness, and equity of the exam process.

Author – Kamaljeet Yadav, Principal, Subodh Public School, Jaipur