How AI Tutoring Can Reshape Teachers’ Days (Opinion)

When ChatGPT and the coming of artificial intelligence aren’t sucking all the oxygen out of the room where educators hang out, the next biggest contender for attention has to be tutoring. In particular, much has been written recently about lessons learned in implementing tutoring at scale. Put AI and tutoring together, and we have the future of education.

Implemented well, tutoring is highly effective compared with other educational approaches. In the past, tutoring has been a special service provided for a very limited number of students. Over the past two years, however, programs have ballooned as districts, sometimes complying with state mandates, spent federal COVID-recovery money on them. In Texas, for example, House Bill 4545 made it mandatory for districts to offer tutoring to students who don’t pass state achievement tests. Given how effective tutoring is, why shouldn’t every student benefit from it?

The problem with tutoring is that good implementation is resource-intensive. It needs trained instructors to provide one-on-one or small-group interactions with students. Right now, we barely have enough qualified teachers to teach classrooms of 25 to 30 students.

If conducted in person but out of school hours, small-group tutoring also needs schools to be open longer and additional transportation arrangements for students. And that’s not all. A school- or districtwide tutoring program requires dedicated planning, scheduling, supervision, and monitoring. It needs a reliable, user-friendly system for documenting which students should receive tutoring in what subjects, which ones show up for their tutoring sessions, and which ones are making adequate progress. Tutors need to be evaluated so that effective ones are retained and ineffective ones are retrained or dismissed.

Well-implemented tutoring is what many of us think of as very good education. Yet, schools just cannot afford this for all students every school day unless the nation makes a concerted effort to increase education spending dramatically on a permanent basis. It won’t.

So, here’s a potentially more feasible solution to work toward: Every student receives at least one to two hours of in-person tutoring each school day from professional teachers while the rest of their time is spent in two ways. They would either be using highly personalized, AI-based tutoring platforms along the lines suggested by entrepreneurial educator Sal Khan or participating in teacher-led academic and nonacademic group activities. Students’ days would be a mix of individualized, small-group, and large-group activities—some delivered by an artificially intelligent platform and the rest by regular teachers.

Like in-person tutoring, AI tutoring platforms personalize learning. By processing large amounts of data on how students interact with content, these programs anticipate what learners are likely to get wrong—and right. They provide real-time assessments, identify knowledge gaps, and suggest personalized learning paths. Students can pursue particular areas of interest using multimedia resources and simulations that go well beyond the classroom, learn at their own pace, and receive tailored feedback. They can revisit materials as often as necessary and ask questions without slowing anyone else down or risking peer judgment. Built-in detectors of affect and engagement challenge the bored student, support the challenged student, gently reprimand the student entering random responses, and surprise the sleepy one via a friendly, customizable avatar.

We know that currently available intelligent tutoring systems work to help students learn a variety of content. Future generations of artificially intelligent tutoring systems are likely to produce even greater effects.

Younger students might spend more time on in-person activities while older students might have more flexible schedules with options for virtual tutoring and completing platform-based work off-site.

For their part, teachers would balance the time they spend teaching between tutoring students and leading or facilitating larger group activities where interactions between students are the goal.

Working closely with students in small tutoring groups, teachers could optimize the role they play in students’ overall development, forming deep and sustained connections with each child. Providing each student with this kind of targeted support would promote teacher-student trust, increase motivation, and help build students’ confidence, leading to improved academic outcomes and social-emotional growth.

Teacher-led, large-group activities offer students opportunities for collaborative learning, discussion, and creative expression. Science labs, debates, physical education, music, dance, or drama teach content but also nurture teamwork, communication, and leadership. In this modality, teachers act as facilitators, mentors, and role models, fostering a safe and supportive learning environment.

Yes, development of artificially intelligent platforms and their content will require large initial investments and ongoing maintenance. But given the evidence behind in-person tutoring and intelligent tutoring systems, the overall approach we propose should be worth the investment at the scale of a large district, a consortium of districts, a state, or, better still, the nation.

Prototypes of the model could initially be tested on a smaller scale, for example, by a handful of charter schools; by a small, innovative district; or by a cluster of schools within a larger district. We only need one or, to avoid a monopoly, a handful of high-quality platforms. These may be developed by commercial publishers; by nonprofits with substantial philanthropic backing; or by well-funded, forward-thinking education agencies. Perhaps this could be task No. 1 for the proposed National Center for Advanced Development in Education. That entity would fund high-risk, high-reward research and development projects within the U.S. Department of Education’s Institute of Education Sciences, like DARPA, the Department of Defense’s Advanced Research Projects Agency.

With savvy scheduling, we should need neither more nor fewer teachers to provide in-person tutoring and to run group activities. The costs may still be greater than before the pandemic, but the benefits to students—and ultimately to America—of combining two of the most powerful strategies we know in education should far outweigh those costs. Our proposed approach can be tailored to provide struggling students with greater in-person support and to allow advanced students to steam ahead at their own pace.

Now that the enabling technology is within reach, district and state education leaders should seriously consider dropping their incremental approaches to change and go all in on an approach based on what works. There are no more excuses for failing to make school effective for everyone.