AI and Machine Learning Are Transforming How We Learn

Take your seats. School is now in session. And artificial intelligence (AI) is here to help you learn better! From taking care of grading to tailoring coursework and helping students prepare for tests, AI is revamping every aspect of education. Let’s explore how.

Re-engineering Learning

At Gloucester County Christian School in Sewell, New Jersey, something special is happening. Not long ago, Jennifer Turner, and algebra teacher at the institution, used to struggle to keep her students engaged with the coursework. It wasn’t uncommon to find one or two students bored to snores by the time class ended.

Today, the situation couldn’t be more different. “The grades for homework have been much better this year,” Mrs. Turner says. “Students are excited to be in my room, they’re telling me they love math, and those are things that I don’t normally hear.”

Turner chalks up her students’ renewed interest in algebra to AI. With an AI-powered platform called Bakpax, Turner can put schoolwork grading on auto-pilot. She simply takes a picture of an assignment and uploads it. From there, Bakpax reads the student’s handwriting and grades the work.

Alongside Bakpax, Turner also assigns lectures for the students to watch as homework. Together, AI and the Internet enable her to spend more class time on interactive exercises that keep the students involved and interested.

Thanks to the development and refinement of machine learning applications over the past seven years, algorithms have finally found their way into the classroom. As a result, they’ve taken over repetitive tasks such as grading. They’re also tailoring coursework to fit the needs of each individual learner. And they’re even helping prepare students for big exams like the SAT.

But in spite of all these advancements, this AI-fueled revolution of education is actually just beginning. AI is allowing researchers to understand more about how the human brain learns. Applying these insights can not only make education easier but also more enjoyable. And soon, machine learning systems could have human-like interfaces that allow them to converse with students and act as a teacher’s aide.

The Killer App for AI?

Terrence Sejnowski is the president of the Neural Information Processing Systems Foundation and runs the Salk Institute for Biological Studies’ Computational Neurobiology Laboratory in La Jolla, California. He believes education represents a big opportunity for AI: “Education, I think, is going to be the killer app for deep learning.”

One-on-one sessions are widely regarded as the best form of teaching. But this method can never be scaled to educate large populations. And for a while, AI didn’t have any hope of solving this problem, either.

The 1960s saw the introduction of the first computer tutoring systems, which would ask students questions as it went through a lesson. But since computers were costly, this technology remained relegated to research institutes. The 1970s and 1980s saw the small rise of rule-based AI systems that would lead students through each step of a problem by giving hints. But like their predecessors, these systems never became ubiquitous due to their expensive price tags and domain expertise required.

After both of these paradigms came decision tree computer teaching systems. While resembling AI from the outside, these systems are essentially preprogrammed learning paths. If a student answers a question correctly, they go down one branch. If they answer it incorrectly, they go down another predetermined branch.

Today, machine learning is able to make the education process more dynamic by analyzing mountains of data on student performance and elucidating insights that can optimize teaching strategies. With more student interaction, machine learning systems can adapt accordingly. Some studies have shown that these systems can improve student comprehension and performance drastically — perhaps even better than students receiving one-on-one instruction.

Tailored to Help Teachers and Test-Takers

Knewton, a company founded by Jose Ferreira, a former executive of Kaplan, was one of the first organizations to employ machine learning in education. “After a few questions we could very quickly figure out what level you are at and the optimal piece of content for teaching,” Ferreira explains. “The more you worked with the system, the better our profile of you got and the more we could give you better and better content.”

After encountering some financial difficulties, Knewton was sold to Wiley, an education publisher. it was then that Ferreira left to go start Bakpax. Headquartered in NYC, Bakpax leverages a computer vision system to convert students’ handwriting into interpretable text. From there, a machine learning system that has taught itself how to score takes over the grading process.

Bakpax’s platform only takes a few seconds to grade assignments. And students seem to love the immediacy. Besides this, Bakpax’s system also analyzes data over time to identify any areas that the class is having trouble with.

With these types of capabilities, it’s no surprise that machine learning is also making its way into the multibillion-dollar test preparation market. Riiid is one such organization using the technology to enter this arena. The Korean startup uses reinforcement learning to maximize a student’s chances of achieving a target test score and claims that just 20 hours of study on its platform can lead to a score increase of 20% or more. Riiid hopes to enter the U.S. market this year.

A New Era of Education

If it wasn’t clear already, we’re just at the beginning of this AI-powered education revolution. Numerous startups and entrepreneurs around the world are working on new concepts, some of which will see their debut in the next few years.

What are your predictions for how AI will impact education in the near future? Have you taken any courses in which AI played a role as a teacher or grader? As always, please let us know your thoughts and experiences in the comments below!