The real question on AI and Assessment: What did the student learn?

October 16, 2026 4 minute read

It has been fascinating to watch the conversation around AI evolve in recent years. In August, I posted a reflection from Building Blackboard Together ’26, detailing how our awesome institutional partners have moved beyond AI hype to adopt a more discerning approach to its role in the learning process.

It was with much excitement, then, that I took to the stage at the EDUCAUSE Annual Conference in Denver – flanked by the Cursive guru, Joe Thibault – to discuss AI and Assessment. Our lunch and learn session attracted a full house, but I was most energized by the follow up, with many of the attendees swinging by the Bb booth to continue the discussion in more detail.

Could we have been asking the wrong question?

For much of the past couple of years, the conversation about generative AI and student work has been dominated by one question: Did the student actually write this?

It’s a reasonable question to ask; however, I would propose an amendment. My version: Did the student actually learn while completing the task? This question encapsulates everything in the original but also leaves open the door for learning to be expressed and understood differently in an AI world.

This was the central thesis of our lunch and learn, and the follow up conversations with attendees indicated that it has been something they’ve been thinking about as well.

More than the final submission

AI has challenged some of the assumptions that have long shaped traditional assessment. For decades, we have relied heavily on the final artifact as evidence of learning. A student submits an essay, a project, a presentation, or other finished product, and it becomes the exclusive focus of our evaluation of their performance.

However, consider all the work that happened prior to that submission. The student had to decide where to begin. Ideas were developed and discarded. Sources were considered. Feedback may have led to revision. An argument changed. Something did not work. The student tried again.

By the time the finished work reaches the instructor, much of that evidence is gone. And let’s be honest, AI did not create that problem. It simply made it much harder to ignore.

How we’re providing visibility into the learning process

This brings us to the opportunity of the moment: to embrace assessment approaches that provide instructors with greater insight into students’ processes to complete tasks. An added benefit here is that this allows AI to be used where appropriate within assessments, meaning artificial and human intelligence can be combined for a desired outcome, as is now common in the workplace.

That is why I was so interested in the response when Joe presented Cursive by Blackboard at EDUCAUSE. Cursive isn't built around blocking AI or trying to detect whether it was used. Instead, it gives instructors something that has traditionally been difficult to digitally access: visibility into how written work develops.

This helps with important challenges around academic integrity and has generated a lot of excitement from partner institutions accordingly. However, I think there is an interesting instructional conversation here as well. An intersection between assessment design and academic integrity, bound by a common focus on advancing learning outcomes for students.

Introducing the Transform Evidence Arc

This is where the Transform Evidence Arc (TEA), a new model I recently devised and published provides structure for educators. TEA looks at learning across six parts of the process: Intent, Exploration, Judgment, Application, Reflection, and Iteration. It isn't designed to turn those six stages into six assignments or six additional things to grade. Rather, it provides a way to think about where meaningful evidence of learning might appear throughout the process.

Now think about that in the context of writing. An instructor might see an idea change direction, evidence reconsidered, an argument reorganized, or substantial revision after feedback. Those moments may provide useful context around judgment, application, or iteration. Cursive makes that process visible with a shared record for students and instructors.

Finally, a point that must be emphasized: the evaluation of where learning has occurred should always remain with the instructor. No Blackboard solution works on the basis of passing that off to an algorithm. Our approach is to focus on helping instructors make informed decisions.

Where to from here?

The interest we saw at EDUCAUSE tells me this is a conversation worth continuing. The discussions that transpired after our session confirmed my suspicion that we have an opportunity to genuinely evolve assessment; to rethink what we look for in student learning, what the learning process can tell us, and how we can better support the instructors responsible for evaluating it. I look forward to working with you all to seize the moment.

Connect with Lisa on LinkedIn to keep the conversation going!

Lisa Clark, Ed.D.

AVP Academic Transformation

Dr. Lisa A. Clark is a higher education executive and recognized thought leader specializing in artificial intelligence, academic transformation, and digital learning innovation. As Associate Vice President of Academic Transformation at Blackboard, she advises institutional leaders on AI strategy, learning innovation, and technology-enabled transformation initiatives.