fbpx Skip to content

Improving Online Proctoring by Combining AI with Humans

  • in Blog

Many areas of online education and educational technologies benefit from AI and machine learning, including online proctoring. Although AI is advancing every day, it still needs human oversight to maximize its capabilities.

But before we show you how combining AI with humans improves online proctoring, we’ll summarize what it is and how it works. If you’re familiar with online proctoring, skip to the next section.

Online proctoring, sometimes called remote proctoring and proctored testing, protects academic integrity by using AI software and/or humans to monitor behavior during online assessments.

Generally speaking, online proctoring is accomplished in a few ways: 

  • Fully automated proctoring (only uses AI to monitor behavior)
  • Live proctoring (only humans monitor behavior)
  • Blended proctoring (monitors behavior with a combination of AI and humans)

It depends on the type of online proctoring used. For example, automated proctoring can include browser lockdown software, which is limited, but it can cover a few aspects of exam protection. 

In contrast, blended proctoring solutions are more flexible and used for traditional assessment formats and nontraditional alternative assessment formats, like virtual presentations/demonstrations, handwritten essays and math problems, and using industry software to complete activities.

Online proctoring is used to create a fair test environment and protect academic integrity.

While most assume that online proctoring is just a way to prevent cheating, some services can improve the entire testing experience for everyone.

In this article, we’ll show you how online proctoring that combines AI with machine learning and human proctors can improve online assessments by reducing distractions and test anxiety, providing better exam insights, and streamlining review.

Three ways combining AI with human oversight improves online proctoring and the testing experience:

1. AI + human proctors creates a better test experience with fewer distractions Fewer distractions mean that test takers can better focus on their exams. However, some proctoring solutions can trigger unnecessary flags that disrupt test takers because they’re overly sensitive. The downside for admins is that those flags take extra time during review. Many behaviors can trigger flags during an exam, such as an exam taker talking to themselves while working through a test question. In general, most proctoring AI can detect sounds, but some have the ability to intelligently detect voices. The terms “sound detection” and “voice detection” are often used interchangeably, but they impact the test experience in different ways.
  • Sound detection may trigger a flag for basically any sound, such as a doorbell ringing, a dog barking, and coughing.
  • Voice detection flags specific keywords and phrases, such as “OK Google” or “Hey Siri.” Exam admins can add custom lists of words and phrases that the AI listens for, such as, “What’s the answer?”

While AI voice detection is advanced, it can still be tricky in certain situations. What if the test taker says a keyword that triggers a flag, but they aren’t actually cheating?

For example:

  • “OK Google” is used as a keyword that triggers a flag.
  • During the test, the test taker says, “OK, Google was founded in 1998, so it’s older than Microsoft Bing.”

So, the AI does its job here, but review from a live proctor is key.

When AI voice detection triggers for “OK Google,” it alerts the live proctor, who can review the situation in an analysis window and determine that the test taker wasn’t cheating, so intervention isn’t required.

2. Provides flexibility to proctor alternative assessment formats and activities

AI works well for most traditional assessment formats, like multiple-choice and true-or-false questions.

But, what if exam administrations want to assess presentation abilities, which includes a demonstration?

Combining AI test monitoring with human proctors is crucial to protecting alternative assessment activities.

For this example, AI monitors for things like other people or prohibited resources in the room, voices, and the use of restricted browsers.

However, during non-traditional assessment activities, like presentations, exam admins may want to allow specific resources, such as notecards or a whiteboard.

Exam admins can adjust settings before the assessment, decide which proctoring features to use, and provide proctor instructions for further customization. Proctor instructions can provide information, such as resources permitted during the assessment and specific accommodations that provide bathroom breaks and assistive technologies, for example.

Example: The assessment requires test takers to complete math problems using a pen and paper.

The admin asks the proctor to turn on “Scratch Paper Allowed” so that test taker aren’t flagged for looking down at their papers.

Further instructions can require test takers to use specific items (blank white paper and a black or blue pen) and that they must show the paper before and after each problem is completed.

This combination of AI with human oversight is important because it provides the flexibility to proctor complex alternative assessments while reducing interruptions and flags to review.

3. Provides data feedback that makes AI smarter and streamlines reviews

AI can collect data from several areas, such as potential violations and suspicious behavior, feedback from human proctors, and input from exam administrators.

This comprehensive collection of data helps the online proctoring AI continually learn and adapt to provide individual organizations and exam admins with better insights about their online assessments. In other words, the AI makes more accurate correlations between specific behaviors that can indicate academic dishonesty and those that are false flags due to oversensitivity or other factors.

This helps proctors and exam admins focus on important behaviors and situations related to academic dishonesty and streamlines the review rather than being bogged down by reviewing unimportant flags.

Honorlock Online Proctoring

What is Honorlock?

Honorlock is an online proctoring service that monitors online assessments with AI combined with live test proctors. Our approach makes online proctoring simple, easy, and human.

How does Honorlock work?
Honorlock’s online proctoring AI monitors the exam session and alerts a live proctor if it detects any potential suspicious behavior. The live proctor can then review the situation in an analysis window before deciding to intervene.

This blend of AI and human review provides a less intimidating and non-invasive proctored assessment experience because they aren’t constantly watched as they take their exam.

In fact, a survey about test anxiety showed that Honorlock’s approach to online proctoring actually helped reduce test anxiety.

Honorlock features and benefits:

  • No scheduling needed because proctored exams can be taken 24/7/365
  • Live, US-based support is available on-demand 24/7/365
  • Integrates with your LMS in an hour
  • Use Intelligent Voice Detection
  • Can detect cell phones
  • Allows proctoring on third-party exam platforms
  • Provides easy-to-read, time-stamped reports and recordings
  • Finds leaked test content and provides steps to take action
  • Offers a flat-rate cost per exam or per exam taker
  • Secures and protects data
  • Verifies ID in about a minute

Get in touch to see how Honorlock makes online proctoring better.

Sign up for more online learning & higher education resources

Tyler Stike

Tyler has over a decade of experience in ed tech and higher education. He focuses on creating relevant and useful content for educators, institutions, and technologists.