US-Israeli startup wins prize for AI software that helps teachers grade tests
Sense Education’s technology, which makes it easier to give students personalized feedback, is honored at EdTech summit in Spain
Sense Education, a US-Israeli startup whose artificial intelligence software helps teachers give personalized feedback on their students’ open-ended assignments, has been named the number one EdTech startup at South Summit’s 2018 enlightED competition.
The event, which saw more than 600 global startups in the field of digital education participate, was held in Madrid, Spain, on October 3-5. Sense Education won in the category for “Best Technology Solution for Learning and Education,” according to a statement issued by OurCrowd, an investor in the startup.
“EdTech is on fire and our technology will ultimately play a major role in providing better and more scalable education for millions of learners,” Sense Education founder & chief technology officer Ronen Tal-Botzer said in the statement.
Providing personalized feedback is particularly important because “teaching is not just giving information,” Sense Education’s president and CEO Seth Haberman told The Times of Israel in a phone interview. It is rather about a dialogue between students and instructors, he explained.
The company takes advantage of Learning Management Systems, online platforms where students can electronically submit their works and receive feedback, to show how such systems can be used to give personalized comments even if there are hundreds of submissions.
The software developed by the firm basically produces AI-generated personalized feedback suggestions about the students’ work, and also suggests a grade for students in the fields of finance, engineering, accounting, science, and technology. The papers and tests need to be open-ended, meaning that they are not restrained by preset answers, and allow for multiple solutions to the questions.
Haberman said that the company used the same principles that regulate genetics — of finding patterns — for the engine of its software, which was first developed at Bar-Ilan University in Israel. The software determines how students have answered the questions, allowing instructors to provide them with “faster, fairer, and more personalized” feedback, he said.
“Genetics is about understanding patterns, and we use the same genetics (method) to understand common patterns of solutions, which are not 100 percent unique,” he stated.
When the company’s AI engine receives electronically submitted assignments, it discovers the common response patterns “naturally shared among students’ answers.”
These response-patterns are then divided into several clusters, based on the type of solution chosen by students.
Such a system produces several clusters, or groups of patterns. In most cases, said Haberman, the first five to six clusters gather patterns widely shared among the group of scholars under examination, accounting for the answers of 70 to 80 percent of all students, he said. This means that most of the feedback that needs to be given will be similar, because feedback is given per solution group instead of per individual submission.
This enables instructors to give grades to about 80 percent of the class, saving time to dedicate to more unique solutions given in the submissions, he said.
From the analysis of patterns, it is also possible to know how students approached the question. Instructors can thus see if the students did extra work to get to their result or conversely if they did not understand the question, leaving room for educators to improve their questions for the next time, Haberman explained.
Patterns that occurred in the vast majority of submissions also allow instructors to investigate social phenomena and families of behavior and shared ideas within the group of students in question, he said.
Without such technology, it would be virtually impossible for instructors, schools and e-learning companies to devote attention to each student’s words, especially for courses with high enrollment or online courses with thousands of scholars, the release said.
The solution developed by the company does not imply that AI can substitute for human educators, Haberman said. “Instructors are still better at giving feedback,” he said. But the software works like an extension, making the instructors’ work better by helping them with tedious procedures like calculations, especially if carried out on a large scale. The use of AI is aimed at enhancing and personalizing the feedback given to each student as the result of a hybrid approach, he said.
Personalized feedback improves the learning experience in various settings like university and high school, as well as e-textbooks, professional training and prep tests, the company said.
Sense represents “what we are all striving for: exceptional technology in the service of better education for all,” said María Benjumea, founder and CEO of South Summit, said in the release. South Summit is Southern Europe’s annual conference focused on business opportunities and innovation, which gathers startups, entrepreneurs, and investors from around the world.
Sense Education is a graduate of accelerator Y-Combinator/ImagineK12’s 2016 program. The startup is a US firm, with business development based in New York. However, the research and development team is in Tel Aviv, with 6 engineers and product specialists.
Sense Education has raised a total of $2.5 million from its largest investors: OurCrowd, Mindset Ventures, a Brazilian VC that invests in Israeli and US technology companies, and a variety of angel investors in the US, OurCrowd told The Times of Israel.
Sense Education’s software is being used in Israel at Bar Ilan and Tel Aviv universities, while a successful pilot study has been conducted at one of the five best engineering schools in the US, Haberman said. Also, the company is developing new products, including one based on Excel and addressing the area of accounting.
According to Haberman, at this time Sense Education has no clear competitors but rather other companies following its lead.