The off-the shelf-machine learning models and general datasets that most online proctoring companies use don’t work well. Rosalyn’s advanced, human-in-the-loop invigilation system with AI built for online proctoring offers a better solution.
An AI-based online proctoring system is only as good as its dataset. To ensure fair and accurate assessment, that dataset needs to reflect the real-world diversity of test-takers.
The scope of a purpose-built AI model is limited to the kinds of scenes that are encountered in test invigilation. The dataset on which it is trained includes people of all shapes, sizes, ethnicities, and neurotypes using a range of user access scenarios, including various screen resolutions and internet connectivity types and bandwidth. This expands the definition of “normal” for test-taking behavior, increases accessibility, and reduces bias.
While online proctoring programs without purpose-built AI may include interesting—but unnecessary—capabilities, these programs are often not trained to identify different human characteristics, movements, or normal behaviors during an exam. Because of this, the software is much more prone to incorrectly flagging “violations”, increasing student stress and anxiety levels. It also puts the student’s reputation at risk and makes the online exam experience difficult for students with children, disabilities, or certain medical conditions.
Online proctoring with purpose-built AI allows institutions to adapt the online exam experience to students' needs while minimizing the potential for various kinds of bias endemic to off-the-shelf AI and machine learning models. Around this core artificial intelligence technology, Rosalyn has built a proctoring system that maintains the integrity of online assessments while respecting the student experience.
No matter how good artificial intelligence is at detecting potential violations, only a human being can make the final decision about exam integrity. Rosalyn's human-in-the-loop AI proctoring system offers that functionality and a user experience that respects students' right to privacy and fairness.
A customized exam experience. Human proctors can personalize the student’s online exam depending on their specific needs.
AI as the first line of defense. If the AI technology spots a potential violation, it immediately flags the event for human validation and, when necessary, real-time intervention.
Reduced algorithmic bias. AI models are continually updated with new test session data to improve their accuracy, reduce bias and maintain the integrity of assessments.
Decisions made by people. When the system flags potential violations, remote human proctors are alerted and make the final call on required actions.
Everyone in higher education has a stake in fair assessments. Students want to be judged on the quality of their work. Professors want to reward effort. Schools want to hold up the reputation of the degrees they confer and the scholars they graduate. Online proctoring using purpose-built AI allows schools to maintain the integrity of their assessments without alienating students from their instructors or school.
Rosalyn’s innovative and easy-to-use online proctoring solution is designed to ensure that no student takes unfair advantage in an online assessment while respecting the rights and dignity of all.