Toggle open
Learning Library Blog Is AI Ready To Meet the Needs of Learning Recovery?
Expand breadcrumbs

Is AI Ready To Meet the Needs of Learning Recovery?

By Errol St.Clair Smith
December 12, 2022
AI Tutors Twitter

Millions of school children walked into their classrooms at the start of the academic year lacking one crucial element that should have set them up for success: the prior knowledge they need in order to take on a new grade level. 

Because of COVID-related learning interruptions, the number of students who started the school year already behind skyrocketed this year. On top of that, a teacher shortage and a tutor drought have compounded the impact. 

Even as the U.S. Department of Education works to recruit 250,000 new tutors, and global companies invest millions to deliver 24/7 online tutoring to students, school districts, nonprofits and commercial tutoring companies are feverishly scrambling to find enough tutors to meet the demand. 

"We don't have enough human tutors to go around to everyone who needs tutoring," says Danielle McNamara, Ph.D., director of the Science of Learning and Educational Technology Lab at Arizona State University. 

Even if it were possible to meet the DOE's target of a quarter million new tutors, McNamara sees another hurdle: Not all tutors will be qualified to meet the complex needs of students who have fallen behind during COVID. 

Enter intelligent tutoring systems, software that uses artificial intelligence focused on pedagogical principles to help students catch up on everything from math and science to reading and writing. 

How intelligent tutoring works

An intelligent or AI tutor is a computer program that simulates the interaction between a human tutor and a student. AI tutors use data to determine which specific concepts or problems are tripping up a student and which ones the student has grasped. It analyzes that information to target instruction and practice so each individual student gets exactly what they need when they need it.

It’s likely that many teachers are already using intelligent tutoring and don't know it. Take Jon Harper, a veteran teacher who provides one-on-one tutoring to elementary students during regular school hours. Every day he roams a classroom of 20 students as the lead teacher covers the course material. 

Harper’s job is to come alongside the third and fourth graders he's charged with supporting using a digital reading tutor to do the heavy lifting. But Harper doesn't think he's using an intelligent tutor; he says it's just a "reading app" called Lexia.

Harper’s students access Lexia on their school-provided laptops while other students work on assigned exercises. 

The relationship between the students and their digital tutor goes back a couple of years. Lexia has tracked each student's progress through the personalized assignments the app created – along with test scores on standardized tests. Lexia knows as much or more than Harper about what each student needs to work on each day. Students can open their laptops and access the app when prompted, and personalized tutoring begins without human input. 

Harper's primary role is to keep his students on task and identify problems that require additional help. 

Results are favorable

Research on the effectiveness of these systems is promising. According to a meta-study by Arizona State University researcher Kurt VanLehn, intelligent tutoring systems are generally as effective as human tutors and, under certain conditions, can exceed the results of human tutors.  

The catch is that intelligent tutoring systems can only match skilled teachers in very narrow classroom scenarios. Most intelligent tutoring systems primarily shine when doing the tasks they were specifically designed to do. 

For example, there are intelligent math tutors that focus only on solving word problems, while others center on fractions. Some intelligent tutors specialize in teaching reading to language learners while others focus on emerging readers. 

Pushing the boundaries, some schools find ways to use intelligent tutoring systems beyond their designed limitations – with mixed results. Harper uses Lexia to teach phonics to general ed students and those with special needs. 

He loves the app but sees its drawbacks. "One of my students is doing really well with Lexia, but she's still struggling with sounding out the word the." Harper has to go beyond Lexia to help this student. It turns out that Harper's experience with Lexia is typical.

"They definitely have their limits," says Jon Nesbit, associate professor on the faculty of education at Simon Fraser University in Burnaby, British Columbia. At the moment, Nesbit says he knows "no intelligent tutoring system that you could just hand over this problem of pandemic learning loss." 

That’s one reason why intelligent tutoring systems for learning recovery are difficult to implement across a school or district. Educators looking for a one-size-fits-all approach will not get it with intelligent tutoring systems. 

Artificial Intelligence - ISTE U edtech PD

Concerns about AI tutoring

Some educators are wary of AI-driven tutoring systems, said Michelle Zimmerman, author of Teaching AI: Exploring New Frontiers for Learning.

The most common concern she hears? "Is the machine going to replace me as the teacher?" 
Her answer is no. 

Computers can’t assess everything that’s going on with a student.  “If a child is really fatigued, if they had a bad day, if there's a social and emotional challenge going on, they may not be able to focus,” Zimmerman said. “And the type of data that's collected through these types of baseline assessments or formative assessments in the programs is only as good as the input that goes in.”

McNamara agrees, “Humans are not robots. Learning is not just a digestion of information; it involves emotions, it's a social interaction. Children need people, they need feedback from a human.” 

The bigger concern is educators depending too heavily on the data these systems provide. "A teacher over-trusting or over-relying on data may think it's the sum total of what we know about a child's learning," says Zimmerman.

Despite the concerns and challenges, intelligent tutors are likely here to stay and will only improve with time. Here are three recommendations for making the most out of intelligent tutors to meet learning recovery needs of students:

1. Know what an intelligent tutoring system can and can't do

Intelligent tutoring systems are highly specialized, says Deb Norton, who teaches ISTE U's course on artificial intelligence. School districts should be wary of buying into a single system and expecting it to meet the needs of all children. "Just like teaching, no one size fits all," she said.

2. Determine what matters most and focus on that

Harper believes it's essential to be very clear about what matters in learning recovery and what doesn't. "I've got a narrow set of things that I'm responsible for working on with each student. Classroom teachers have a huge amount that they're responsible for covering,” he says. “I think a big thing is determining what is really necessary for third graders to be able to do when they get to fourth.” 

3. Create a global index of tools

Educators need to be able select the right intelligent tutoring system at the right time for the right student, McNamara says. She imagines a system that would allow a teacher to access a library of intelligent tutoring systems and quickly identify the content, pedagogical assumptions, characteristics and features. "We need the means to combine a library of automated assessments, intelligent assessments and intelligent tutoring systems that is searchable by parents and teachers." 

Intelligent tutoring systems are here to stay

As artificial intelligence becomes more entrenched in our machines and processes, intelligent tutoring systems are expected to proliferate. Here are a handful of examples of what’s out there now:  

ALEKS: This is an online learning platform that helps educators and parents understand students’ knowledge and learning progress in depth.

Cognitive Tutor: This math program uses intelligent software that monitors the status of the student’s knowledge on a moment-by-moment basis and tailors course material, based on continual assessments.

DreamBox: This platform offers personalized math and reading programs that use adaptive technology to enable teachers to address the individual needs of students in the moment.

Lexia: This is a web-based, individualized reading curriculum for PK-5 students of all levels and abilities, where students learn at their own pace and on their own level. 

Writing Pal: This web-based software tool analyzes essays in the same way a teacher might while also providing writing-strategy instruction, game-based practice and individualized formative feedback to help students improve their writing proficiency.

Freeing up quality teacher time

Harper was pleased with his experience using the digital tutor. It saved him time while building confidence in emerging readers. "I've been able to sit and watch this student build confidence with something as simple as ending sounds, and I've got another student I work with on the same things, and that student is more confident, too." 

Harper also marvels at how intelligent tutoring systems can lighten teacher workloads and reduce teacher overwhelm. 

But perhaps the most important benefit is that digital tutors can give educators more time to do what they do best — develop meaningful relationships with students.

AI is already here. Are you ready to teach the future? Learn how when you read Teaching AI.

Errol St.Clair Smith is an Emmy-winning reporter and executive producer at BAM Education Radio Network.