Although online proctoring was commonplace in Higher Education before the onset of COVID-19 in 2020, the adoption of online proctoring accelerated dramatically when institutions were faced with the stark decision to move testing online, adopt alternative forms of assessment, or stop testing altogether. As institutions began converting to online proctoring, an interesting dialogue from detractors emerged with questions and comments like, “How are students safeguarded?” “It’s not a level playing field,” “It’s not comparable to classroom testing,” and “It doesn’t stop cheating.” From the perspectives of students, similar concerns emerged such as: “How is my data being used?” “Why don’t I have a choice?” “It’s creepy,” “I’m worried it will accuse me of cheating when I’m not?” “It’s invasive,” and much more.
Online proctoring offers flexibility, convenience and availability for the student, and — if done properly — can also be easier to both administer and take as opposed to scheduling and securing an exam at a physical location.
There now exists a variety of online proctoring modalities. Some providers utilize AI technology, with the intention of making the online proctoring process more efficient. Similar to using a fingerprint or facial detection login with your cell phone, AI technology has been used in online proctoring for ID validation. AI can pick up on repetitive actions just like humans can and can also flag suspicious behavior when a student is taking an exam, such as changes in lighting or repetitive movement that may indicate a student looking at unauthorized material.
With the onset of COVID-19 and the impact of the pandemic, we have noted an overreliance on AI in online proctoring. Many institutions quickly adopted AI in online proctoring in order to implement the lowest-cost solution as quickly as possible. However, while AI is efficient in flagging behaviors, those flags are not always indicative of integrity violations. Replacing a human proctor with AI will no doubt create efficiency and cost saving, however it’s at the detriment of accuracy, fairness and reliability.
AI proctoring often treats common human behavior as misconduct, which poses a real risk of unfairly penalizing honest students and creating unnecessary anxiety. Or worse, discriminating against certain groups – for instance parents, people in a crowded living situation, or individuals with a neurological disorder or severe test taking anxiety. AI can’t interpret behaviors as being either meaningless or dishonest. AI just says “this behavior doesn’t match the data set of norms I have been programmed to identify as acceptable.” AI-based proctoring that assumes all students take tests the same way—for example, in rooms that they can control, their eyes straight ahead, fingers typing at a routine pace, mouse clicking in a normal cadence —puts a black mark on the record of students who operate outside that norm.
Whether it’s powered by human intelligence or artificial intelligence, the purpose of any exam security measure is to maintain fairness, create a level playing field and protect the integrity of an academic qualification. Institutions that use Canvas may have access to a multitude of tools to help them achieve this objective. Faculty must return to this purpose before introducing any new technology that may impact the fairness and reliability of the measures they are implementing. PSI has always operated under the principle that proctoring student’s exams remotely is a human-centric process that can be assisted by technology, but never wholly facilitated by technology. The reliance on AI and machine learning algorithms alone, without human objectivity, for such a task like making determinations on human behavior, is a risky proposition. Many students have attested to this point over the past year and have demanded more transparency and accountability from their institutions and the technology vendors.
Online proctoring can be an incredibly useful tool when implemented properly and for the right reasons. Our aim has always been to work with institutions to create a level playing field and replicate as closely as possible the experience of taking an exam in-person. Part of that experience includes exam security being overseen by a human proctor who is trained to know the difference between out of the ordinary or unusual behavior and an individual who is trying to cheat.
Join us for this session that will address some key points faculty using Canvas should consider when assessed against the measures of fairness and reliability.