The rise of remote learning, even before 2020, created a void. Where there was once a human teacher or proctor monitoring students in person, there was now none. Remote proctoring services became as valuable and in-demand as ever, and the increasing adoption of artificial intelligence (AI) has permeated these remote-proctoring programs.
Yet, despite its obvious necessity, remote proctoring has always had flaws. One of these services’ most glaring shortcomings was the penchant for poor internet connections, blurry webcams, misconstrued student movements, and other issues that trigger false accusations of misconduct by the learner.
There are many AI-integrated proctoring platforms. The common thread is their imperfection in evaluating students’ intent and behavior. If you or your student faces an AI-triggered false allegation of wrongdoing, call the LLF National Law Firm today at 888-535-3686 or contact us online. Our Student Defense Team handles these issues regularly, and our commitment to protecting students’ rights and reputations is steadfast.
More and More Remote Proctoring Services Rely on Artificial Intelligence
We should not lose sight of the word “artificial” in the phrase “artificial intelligence.” While humans are undoubtedly imperfect, a human proctor might have advantages that an AI proctor lacks, including:
- A familiarity with a particular student, including their academic performance, personality, and history of abiding by rules and ethical expectations
- Nuance in interpreting humans’ actions—while an AI proctor might flag every single prohibited action on a checklist, a human proctor might be able to recognize and respond appropriately to an obvious internet connectivity issue or a student who is merely adjusting their laptop screen
- The ability to communicate productively with the student in real time, whereas AI chatbots have a limited capacity to have in-depth, human-like discussions with a student
It’s understandable that schools would view artificial intelligence as a viable alternative to human proctors. Human proctors require compensation and are in short supply. That said, schools that trust these AI-powered alternatives implicitly without accounting for the technology’s shortcomings will place students at risk.
Technological Issues Are a Minefield for Students Forced to Use AI-Powered Proctoring Services
Remote proctoring services have long attempted to combine objective data with a more nuanced, human-like interpretation of the learner’s behaviors and micro-expressions. While it is easy enough to see if a test-taker opens a tab and conducts a Google search (the former), it’s far more challenging to make conclusions based on a test-taker’s eye movements (the latter).
We must consider how effective artificial intelligence can be at detecting “suspicious activity” among the students it monitors. Some of the common scenarios that can lead to students being falsely accused of cheating by remote proctoring services include:
- Poor internet connection: Because remote proctoring typically relies on tracking the test-taker’s eye movements, a lagging internet connection might contribute to false cheating triggers. These technologies can also require substantial computing power, so the very use of the remote proctor might contribute to such issues, issues for which the test-taker might ultimately suffer serious consequences.
- Internet and power outages: If a test taker’s internet or power cuts out while using a remote proctoring software, the software could prevent them from logging back in. It’s also possible that the software registers the disruption of the test-taking as an indicator of cheating.
- Lighting issues: Most remote-proctoring services require a clear view of the test-taker’s face, which requires adequate lighting. Proctoring technology might also interpret flashing lights on the test-taker’s face to indicate that the test-taker is accessing a screen on another device (such as an adjacent laptop). Yet, some test-takers have limited control over the lighting around them, especially when they live in someone else’s home and have nowhere else to take an exam.
- Human environmental factors: Most remote-proctoring services consider audio inputs to evaluate whether a test-taker is cheating. Therefore, noises beyond the test-taker’s control, like students talking in the background in a library, may be used to falsely accuse the test-taker of cheating.
- Non-human environmental factors: Pets, weather sounds, traffic outside the testing room, and countless other environmental factors (again, beyond the test-taker’s control) may lead an AI-powered proctoring service to conclude unethical behavior by the individual being monitored.
- The test-takers’ subconscious behaviors: Many of us have tics to which we are largely oblivious. Whether it is an eye movement that AI would consider “unnatural” or mumbling to oneself as they consider which answer is correct, these innocent actions can be flagged as potential or certain signs of misconduct.
While some remote proctoring services are designed to continue working during an internet outage or slow connection, it remains possible that a choppy video connection, for instance, could lead the technology to falsely detect dishonest behavior by the test-taker. As you can see, an unreliable connection is just one of several issues that expose AI proctoring’s limitations.
The Primary Issue with AI-Powered Remote Proctoring: Humans Assign It Too Much Weight
Every proctor, human or robotic, is imperfect. Therefore, it’s not as if AI-assisted proctoring services are too unreliable to be useful. The primary problem is that too many educators and administrators regard a (blatantly) fallible technology as infallible.
Consider that at least one AI-integrated proctoring service generates a “Trust Score” of a test-taker’s performance. Among the ways it generates this score is through the audio cues in the test-taker’s environment, an input we’ve already exposed as unreliable.
Say an educator knows a student well, and that the student has a well-established history of honesty. What is that educator to do if an AI-powered proctoring service tells that teacher that a student’s performance only warrants 15% of their trust?
This is the sort of conundrum that has landed countless students in disciplinary proceedings, facing the tarnishing of their reputation and other tangible sanctions from their school. If you or your student finds themselves in such a circumstance, call the LLF National Law Firm today at 888-535-3686 or contact us online about how we can help.