This Israeli Startup Developed AI to Build the Future of Education

Share on linkedin
Share on facebook
Share on twitter

Massive Open Online Courses, or MOOCs, were created in 2012 and have been heralded as the future of education. Led by Coursera and edX, they were created and pioneered by many of the best universities, like Harvard, MIT, and Stanford. Today, more than 900 more universities around the world have followed suit, including schools in Israel such as Bar-Ilan University, Technion, and Tel Aviv University. In fact, a national consortium of higher education institutions in Israel, IsraelX, has brought together many of Israel’s MOOCs into one central location.

Illustrative university class. Credit: Sense Education

For many, MOOCs have held the promise of extending access of higher education to millions more learners. But they’re lacking a critical component to the learning process: personal and instant feedback, a proven method for robust learning. That method is nearly impossible for MOOCs, where class sizes often top thousands of students.

Sense Education, an Israeli and US-based startup, is solving this seemingly intractable problem by scaling education with their AI-based smart grading solution to ensure the future of education maintains its personal touch.

Last year, more than 100 million learners studied through a MOOC, according to the research group Class Central. And while the number of enrollments are rising for MOOCs at large, completion rates are not improving. Among all MOOC students, 3.13 percent completed their courses in 2017-18, down from about 4 percent the two previous years and nearly 6 percent in 2014-15. More people are certainly finding their educational content and even their degrees online, with more than 10,000 MOOC courses available today. But MOOCs have not succeeded in truly disrupting education, especially when benchmarking MOOCs to traditional education methods. “But there doesn’t need to be a tradeoff between quality and scale,” says Sense CEO, Seth Haberman.

Sense Education was founded in 2016 by CTO Ronen Tal-Botzer, serial AI entrepreneur having founded two AI startups prior (Cognitical and Correlor Technologies), and concurrently lecturing at Bar-Ilan University. The startup is led by Seth Haberman, former founder of programmatic TV advertising company VisibleWorld (acquired by Comcast), SVP at Comcast, and Emmy Award winner.

Sense’s platform provides rapid custom feedback on students’ online submissions and suggests recommendations to improve their problem solving approach. “You cannot scale education if you can’t scale feedback on open-ended assignments,” stated Tal-Botzer. Sense uses unsupervised machine learning to cluster student submissions together according to a certain pattern that represents an archetype of a given student’s problem-solving approach.

Sense uses unsupervised machine learning techniques, inspired by bioinformatics and genomics to cluster and classify students’ learning archetypes. Credit: Sense Education.

“We took inspiration from bioinformatic and genomic algorithms that are used to analyze DNA and genes, which we applied to education” explained VP R&D Nati Ghatan. “We take the student submissions and determine the pseudo-genetic sequences, as does genetic sequencing. We use bioinformatic algorithms to analyze a part of the strand, like genes in the DNA sequences of student submissions. We’re not looking for exact matches in a student’s problem solving approach, but rather identifying differences in patterns. And we use machine learning to identify which differences matter by using an evolutionary algorithm that makes those genes compete for survival against each other, leaving us with clusters, where each cluster represents a single unique ‘genetic’ makeup where all the submissions that fall into that cluster share the same genes and patterns. Now, an educator can see those homogenous clusters – each cluster represents a single approach to a problem – allowing the professor or teaching assistant to review only several of the submissions per group, get the gist of that particular group, and provide high quality feedback to be distributed to the students.”

Sense’s dashboard view of automated grading and clustering of submissions. Photo: Sense

By allowing course staff to provide the same feedback to similar submissions, Sense helps teachers reduce repetitive work. Sense’s platform subsequently switches to a classification method to enable students to solve a problem in the proper problem-solving technique. For example, they’ll classify personalized feedback for a given student, like clever hints for their submissions. As a student submits multiple submissions for a problem, Sense can track their change from one cluster to another, and the professor can utilize this data to monitor and identify students that may need help early on in their studies, thereby reducing attrition rates.

The startup’s AI isn’t a panacea to our education woes, but does significantly accelerate a bottleneck in the online education system. “It’s a very much hybrid model. It’s a like a level 2-3 self driving car,” explained Sense CBO, Akiba Covitz, former Associate Dean of Harvard University and founding VP of university relations at leading MOOC provider edX. “The goal is to give both professors and students the tools they need to succeed. Sense gives you a faster and a fairer way to grade submissions, especially in the growing world of online learning. It also brings teachers back into the center of the students’ education even as class sizes grow.”

Sense has won a number of competitions, in Israel, Europe and the US, and was a participant in Y Combinator’s 2016 cohort. They’re already being used in many universities across the US, including Georgia Institute of Technology, Arizona State University and Southern New Hampshire University. Sense has raised a total of approximately $4.5 million in private funding from investors including OurCrowd. They have offices in both Tel Aviv (R&D) and New York City (Executive).

The massive, multi-trillion dollar education industry currently doesn’t have an AI solution to provide feedback to students at scale. Some of the largest tech players in education are the learning management systems providers, including BlackBoard ($700 million in revenue for 2018), Moodle ($6 million private funding and open source), and Canvas (part of a publicly traded company Instructure, with a market cap of $1.5B).

Market forecasts by research group HolonIQ show rapidly increasing revenue in the EdTech sector, nearly tripling since 2015. And what may best be called “e-learning” is projected to grow to over $325 billion by 2025. With that kind of growth in sight, educational institutions are looking for tech partners to help them compete in an increasingly competitive space. Sense is positioning itself to lead the way in that surge, and accelerate a deepening partnership between education and AI.