By Sergey Karayev
Artificial Intelligence is coming for education.
But don’t panic.
It’s not going to replace college faculty or teaching as we know it. It’s not a slippery slope. Instead, AI is going to give faculty superpowers, extending their reach and expanding their time.
A good teacher is a role model, a sage, able to become what the student needs. Teaching is too personal, too human, to be turned over to AI.
That’s not just my opinion. Three years ago, McKinsey, the global consulting firm, issued a report on how and where AI and automation was most likely to replace jobs and job functions. They listed “Educational Services” as the sector least likely to undergo that type of technology-dependent displacement saying, “… the essence of teaching is deep expertise and complex interactions with other people.”
Consider also Dr. David Weiss, a psychology professor at the University of Minnesota. Weiss was probably the first person in the world to use computers to give and grade assessments, work he was doing as early as 1969. “As far back as the 1970s people said we could have computers deliver instruction, we won’t need teachers anymore. And I’m hearing that again now because so much is on computer,” he said recently. “But that’s never been realistic. There are things computers can do well and things they can’t,” he said.
That’s all true and unlikely to change. Teachers teach. They are good at it. No one wants to change that.
So, the dawn of AI in teaching does not mean we’re on a path to robot instructors. Computers and algorithms are highly unlikely to come between faculty and students anytime in our foreseeable future.
Where AI can help today is outside the classroom, making many non-instructional responsibilities of teaching easier and faster.
As an example, the area I’m working on is AI-assisted grading. When fully tested and deployed, it will be able to do things such as group student answers by their content, and batch feedback to all essentially similar responses in the blink of an eye. So instead of a teacher writing “forgot to mention the Krebs cycle” 50 times, they can identify the error once and write their feedback once and the AI in the tool will propagate it to other responses with the same error.
AI assessment tools can also help faculty spot sticky subject areas for subsets of students and even make student-by-student recommendations for areas of extra attention. It can spot when an unusually high percentage of students struggled with a particular question, flagging that either the specific question or the whole topic needs teacher review.
Make no mistake. This won’t replace grading—teachers will still decide what’s correct and what isn’t. Teachers will still approve the results. They just won’t need to spend as long doing it, and they will be more accurate to boot.
Used correctly, it could turn the rote process of grading into a faster, less repetitive exercise in much the same way the Scantron or optical mark recognition made scoring multiple-choice assessments faster. Neither innovation replaced teaching, they made being a teacher easier.
Think of it as the difference between using Microsoft Word or a typewriter. Computer-based typing tools such as spellcheckers and cut-and-paste did not replace writing or displace writers, they made writers better, faster, more powerful.
My point is not that automated grading tools and other AI advancements will be mundane improvements. I am confident they will be tremendously important advancements in education. What I’m saying is that the AI that is coming to education will be in the support systems, freeing faculty to do more of what they love, the things computers can’t do: mentor students, make intellectual connections, and inspire curious minds. Giving teachers significantly more time and energy to do those things has the potential to be a game-changer for learning.
AI can do that, and not just in grading but in
other areas too, streamlining the tasks and chores of faculty that exist largely outside and apart from person-to-person, teacher-to-student engagement. The point of AI is to make those moments more frequent and more powerful—to be a teaching superpower.
Sergey Karayev has a PhD in Computer Science from the University of California at Berkeley, is co-founder of Gradescope, and head of AI for STEM at Turnitin. He is also a co-organizer of Full Stack Deep Learning Bootcamp, which delves into best practices of all components of deep learning.