You are here

Feature

AI accelerates in K12

Machine learning will offer teachers deeper insight, but not replace them
Rose Luckin is the chair of learning with digital technologies at University College London’s Knowledge Lab.
Rose Luckin is the chair of learning with digital technologies at University College London’s Knowledge Lab.

Until recently, the quality of classroom instruction relied almost entirely on a teacher’s resourcefulness, motivation and intelligence. Soon, it will also depend on artificial intelligence—with lessons based more on what students need to learn than on traditional methods of instruction.

“Even with PCs, projectors and the internet, the way students are taught hasn’t changed that much in the past 50 years,” says Rose Luckin, the chair of learning with digital technologies at University College London’s Knowledge Lab. “With artificial intelligence helping the teacher, there will be a revolution in education that addresses many of teaching’s shortcomings.”

Sometimes called “the fourth industrial revolution” (after steam, electricity and computing), artificial intelligence is all around us, from beating people at poker to analyzing mortgage applications and replying to requests  on voice-activated phone systems at home or on your phone.


Sidebar: Security: The malware-cloud connection 


In education, AI mimics the human brain by using sophisticated cloud-based computers to learn about students as they work through material. Forget about robots in the classroom (at least for now), because AI will work behind the scenes to tailor curriculum to each student’s needs and learning style.

More advanced personalization

At the moment, AI can answer 40 percent of routine student questions, such as where reading assignments are located. But, as the technology matures it will be used in more and more classroom situations, according to McKinsey Global Institute’s report “Artificial Intelligence: The Next Global Frontier.”

Use of AI in K12 and higher ed could grow 47.5 percent by 2021, according to a recent survey by Research and Markets, “Artificial Intelligence Market in the US Education Sector 2017-2021.”

AI will usher in a more advanced era of individualized teaching. For instance, educators in the Oak Creek-Franklin School District, just south of Milwaukee, use the ALEKS artificial intelligence system for all sixth through ninth grade math classes.

While Oak Creek-Franklin isn’t a 1-to-1 district, 2,000 students use Chromebooks to learn all phases of math—up to algebra—through artificial intelligence. After a initial assessment, the Web-based program “knows what the student knows and challenges them with new material,” says Annalee Bennin, Oak Creek-Franklin’s director of curriculum and assessments.

And far from replacing teachers, AI can make them more efficient by reducing the mundane tasks, such as grading, so they can work with students who need enrichment and intervention.

Painting a complete picture

Though AI has the potential to deliver truly custom content aimed at each individual student, it is not the same as adaptive learning platforms, where correct answers lead to harder questions.

The way AI works is a delicate ballet of data and interaction. It keeps tabs on each student’s needs on a moment-by-moment basis and, compared to periodic assessments, gives teachers a deeper understanding of students’ abilities.

Coppell ISD, which borders Dallas, has deployed IBM’s Watson Element for Educators to create more complete pictures of learners. All pertinent aspects of the student—from test scores and homework to outside activities and discipline reports—are assessed by Watson’s algorithms.

The system then prepares an individualized learning program for the student that the teacher and staff can review and augment.

Virtual Virginia’s online high school uses an AI Chinese language program that aims to engage students in conversation with lifelike avatars who seem to have a mind of their own. Students speak in a conversational style as the program, developed by Alelo, picks out the nuances of their speech and probes knowledge with pointed questions.

Spotting specific needs

Other programs, which stress interaction between learners and educators, combine a sophisticated artificial intelligence system in the cloud with live teachers to help students progress.

A child’s struggles with fractions, for example, could be caused by a number of common problems: deficiencies in multiplication and division, not understanding the least common denominator concept, or confusing the numerator and the denominator.

Jessica Steele, a teacher at Brunswick Acres Elementary School in New Jersey’s South Brunswick School District, recalls how none of the traditional approaches worked with a third-grader who was having trouble with multiplication.

Its AI platform, created by Thinkster, determined the student needed help counting by twos and threes. This specificity could not have been achieved using classic, multiple-choice testing. It required the deeper understanding of the child’s abilities that AI brings to the classroom, Steele says.

AI can go a step further by using its in-depth knowledge of a student’s abilities in order to perform predictive analysis. This can flag students who need special attention in a specific area or who are thriving but might be bored and better served with more challenging material.

Day by day as more student data is included in the AI learning model, the technology’s teaching ability should improve, not unlike a young teacher’s evolving classroom techniques.

“The power of AI lies in its data,” says Luckin, of University College London. “It can rewrite the rules of education.”


Brian Nadel is a freelance writer in Pelham, New York.