"Assessment security begins where academic integrity ends"
You don’t have to secure all assessments - just those that must be valid measures of the students’ mastery of learning outcomes (e.g., summative assessments). Where you can, revise learning outcomes, redesign assessments, and create cultures of integrity to make "cheating the exception and integrity the norm" (Bertram Gallant). But where you must, secure assessments so that you know they’re being completed in the way you intended (reliability) and they are measuring what you designed them to measure (validity).
All instructors should work to create classroom cultures of integrity. Emphasizing integrity and establishing shared norms and expectations can't hurt and could positively impact some student choices. However, for those assessments that really matter - that determine if a student has sufficiently achieved the course learning outcomes or is ready to move on to the next course in the sequence - stopping at promoting integrity is insufficient, as is simply trusting students to do the right thing even when no one is watching. After all, we generally trust our neighbors, but we still lock our doors to raise the barrier of entering our homes without our permission. We might like to trust ourselves from eating junk food, but to really prevent ourselves from doing so, we don't store junk food in our homes. We need to understand that students - humans - need help sticking to their goals or keeping their promises, whether it comes to cheating, or workout plans, or study plans, or really, anything!
And trusting students not to cheat isn’t fair to the students who won't cheat no matter what.
It's also not fair to the students who won't actually be able to resist the temptations and opportunities to chat that come with unsecure assessments. Nothing like putting an 18 year old in front of a computerized high stakes test, unproctored, at home, and then say "but don't click on that button that will give you all of the answers to this test!".
Sometimes students - humans - need help to refrain from engaging in behaviors that won’t serve them well in the long run.
No. You may have heard that "authentic assessments" are the way to prevent cheating. It's true that authentic assessments can enhance students' instrinsic motivations to learn, and also sometimes make it difficult for GenAI to satisfactorily complete. But, authentic assessments aren't always secure assessments. If they are still completed unproctored or unobserved, then they won't be secure. Having said that, secure assessments don't have to be inauthentic! Oral assessments can be authentic, and traditional exams can even include authentic scenarios/questions.
If the assessment is unproctored or unobserved, no matter the nature or quality of the assessment, it is not considered secure. There is no way to design a "cheat proof" or "AI proof" out-of-class assessment.
The assessments that need to be secure are those that:
evaluate, rather than facilitate learning;
determine if a student is ready to progress into the next course in the sequence;
predict a student’s ability to succeed in a particular genre/discipline (e.g., math placement tests);
require a restriction of cognitive offloading (because lower level knowledge is being assessed);
have to be reused term after term with few changes; or
result in degrees being granted (e.g., theses, culminating exams, capstone projects, dissertations).
The good news is, that you do not have to go back to in-person, paper-based, proctored exams to secure assessments. Certainly, that is one option. But what students want to write for 1-3 hours and which instructors actually want to have to read handwritten work?
There are many other alternatives including:
you could experiment with Oral exams. A team of UC San Diego faculty found that short oral follow-ups to written exams (remote or in-person) has a lot of benefits, including enhancing academic integrity
or maybe Interactive oral assessments
what about Two-stage exams?
However, if you want or need to stick to more traditional assessment format, like quizzes or exams, then they need to be invigilated, preferably in-person with trained proctors. Be careful in assuming that simply walking around the room will be sufficient, especially if students are taking the test on their own computer device. Lock-down browsers should be used, as should assigned seating and ID checking (remember: contract cheating - humans taking exams for other humans - is still a thing).
At UC San Diego, and at a growing number of other colleges and universities (e.g., University of Illinois, Urbana-Champaign; University of British Columbia; and UC Riverside), instructors who want to do computer-based testing can send their students to take their assessments in the Triton Testing Center's Computer-Based Testing Facility (CBTF). CBTFs are not only great for securing assessments, but they are also good for enabling mastery-based testing - at least if the tests are administered on an platform that enables the individualization of assessments (like PrairieLearn does) - which has been found to reduce equity gaps. As Craig Zilles and colleagues argue - "every university should have a computer-based testing facility".