Machine Learning at Toppr: Is Personalised Learning the Way Forward?
Toppr, online tutoring platform in India is innovating by using data and machine learning to create a personalised learning platform for K-12 students. What are some of the issues they are facing, how have they solved it and what is the way forward?
Spending hours solving a math problem while I prepare for an exam, I was unable to figure out where I am faltering! I wish I had an app that not only told me where I am going wrong, but also prepared a personalized path of learning to get me enough practice to master the concept!
This is exactly what Toppr is doing. The company, based in Mumbai, India is an EdTech company offering online tutoring services to learn, practice, clear doubts and practice papers for K-12 students. It enables supplemental practice and test preparation for competitive exams. The online tutoring market in India is expected to be almost US$9 billion by 2022. The opportunity is particularly huge in markets such as India, where students lack the access to quality education and teachers are often unable to identify the individual learning in a class of 50-60 students. Toppr offers customized leaning to students that covers their syllabus and the entrance exams they aspire to clear. As the company targets to acquire 8 million students on its platform, it is only scratching the surface of the 400 million school going children in India.
The company uses machine learning and data (collected from a millions of data points from every user activity) to continuously improve its algorithm to develop an adaptive learning platform. Machine Learning has made learning more structured as students can attempt questions according to their learning level. It can identify a student’s weakness down to the concept level and pop up questions that help them improve and giving them insights on their performance. The company has evolved its practice module and further improved by collecting more data from new students using the platform. The use of this technology has also granted them an edge over their competitors. The company has been hiring the right talent in its technology and engineering team to build an adaptive product that can evolve with the increase in the customer base. The company also hired experienced individuals from the sector Prabhdeep Bedi from the US (McKinsey), Niranjan Chavan from Canada (Compass Group); and Sudeep Mitra with over 20 years of experience at companies including Hummer, Reliance, Tata, Airtel and IBM. These professionals were hired to help with the next phase of growth and integrate technology in the growth process.
A challenge while using Machine learning technology is the lack of data points from new users. The company is facing problems in converting its users to paid users. The lack of conversion from the freemium model does not allow the company to collect quality data and help improve its algorithm. Therefore, the management team has been working on raising awareness about the product and helping the end consumer understand the benefits of technology in the education sector. Zishaan Hayath, CEO, is also an angel investor and has been very vocal about his opinion on the role of technology in opening transformative avenues in improving education outcomes. The increased internet penetration in the country and the inability of the legacy tuition providers to provide quality services has created the opportunity for the company to establish a strong product-market fit. It was about building trust within the student community by demonstrating that the app was worth their time. Further, learning involves developing a sense of discipline that is critical to self-paced learning. While computers can provide suggestions about what students should do, actual practice that ultimately lead to outcomes can only be executed by the student. The company had thus focused on engaging the students on the platform to enable them to take the test and practice materials and enable the machine learning algorithm to give the most optimum answer. The company has used design to create a platform that improves visibility and discovery. It gives the students insights from the machine learning algorithm that is easy to digest and uses gamification to make learning fun.
Challenges and engagement tools
Machine learning in education allows teaching to be differentiated and personalized. Computers can deliver customized lectures to each student, liberating the instructors to work on a one-on-one basis with students to help with more difficult problems in students’ understanding. It still begs the question on whether one day the machine will be sophisticated enough to completely replace the teacher and the student would be able to learn on their own? Or will we continue to value the human wisdom that has developed and passed down over generations in education?
(Word Count – 737)
1. Toppr, Official Website, https://www.toppr.com/ , Accessed Nov 12, 2018
2. Jibu Elias, “Technology can bridge the gap in our broken education system, Toppr CEO,”, Dec 26, 2017, https://in.pcmag.com/edtech/118190/technology-can-bridge-the-gap-in-our-broken-education-system , Accessed Nov 12, 2018
3. Malik Shaikh, “Our Design Journey In Building The All New Toppr Web App”, Sep 13, 2017, https://blog.toppr.com/redesigning-the-toppr-web-app-9e2dce696fb5 , Accessed Nov 12, 2018
4. Dearton Thomas Hector, “Ed-tech market far from saturation, says Zishaan Hayath of Toppr”, Apr 20, 2017, https://www.vccircle.com/ed-tech-market-is-a-long-way-away-from-saturation-says-zishaan-hayath-of-toppr/ , Accessed Nov 12, 2018
5. Will McGuinness, “The Benefits and the Limitations of Machine Learning in Education”, Feb 5, 2018, https://www.gettingsmart.com/2018/02/the-benefits-and-the-limitations-of-machine-learning-in-education/ , Accessed Nov 12, 2018
6. Anita Shashidhar, “Learning on the go”, Jun 21, 2015, https://www.businesstoday.in/magazine/cover-story/coolest-start-ups-2015-haygot-education-toppr.com-strategy/story/219952.html , Accessed Nov 12, 2018
7. Prachi Verma, “New-age education companies are luring top talent with crore-rupee packages”, Dec 26, 2017, https://economictimes.indiatimes.com/jobs/new-age-education-companies-are-luring-top-talent-with-crore-rupee-packages/articleshow/62244384.cms , Accessed Nov 12, 2018
8. Disha Seth, “Completing a student’s learning cycle using technology”, Nov 5, 2018, https://blog.toppr.com/completing-a-students-learning-cycle-using-technology-9db8c391ddf6 , Accessed Nov 12, 2018
9. Banasree Purkayastha, “Toppr: This edtech platform caters to over 2.5 million registered students”, Feb 28, 2018, https://www.financialexpress.com/industry/toppr-this-edtech-platform-caters-to-over-2-5-million-registered-students/1082250/ , Accessed Nov 12, 2018
10. Disha Seth, “Completing a student’s learning cycle using technology”, Nov 5, 2018, https://blog.toppr.com/completing-a-students-learning-cycle-using-technology-9db8c391ddf6 , Accessed Nov 12, 2018
11. Unknown, “5 Ways Machine Learning Is Transforming Education and the Classroom”, Apr 16, 2018, https://www.lanner-america.com/blog/5-ways-machine-learning-transforming-education-classroom/ , Accessed Nov 12, 2018
Student comments on Machine Learning at Toppr: Is Personalised Learning the Way Forward?
Loved the article and the concept of adapting learning models to individual needs. I do believe humans will play a vital role as operators of machines as well as decision-makers with machine outputs. However, the question to me after reading this article is how could adaptive learning platforms implicitly teach human beings to think like computers? If the differentiated value of human beings is that we think differently than machines, we should be wary of programs that shape of we learn. If human beings lose the ingenuity and eccentricity that makes us valuable, we may play less of a role in a more automated future than w
I think that digital educational platforms in India are an interesting concept. However, I believe that that given the price sensitivity of the Indian consumer and the prevalence of the traditional teaching model (students being taught by a teacher in a classroom), that having only a digital platform will be difficult to monetize. While Toppr’s platform is innovative and it is using machine learning to continuously improve its product, I think it will have a difficult time convincing people to pay for its product. I believe that for Toppr to maximize its value, it needs to partner with physical coaching centers with trained professionals. The combination of a teacher and the platform will, in my opinion, drive more students to trust its product. Once a critical mass of students is built, Toppr can further scale its digital platform.
One question that this raises for me is if there are any fairness implications in the way the algorithm is developed. For instance, what if the computer decides that a student should be at a certain level and never challenges them outside of their comfort zone. There have certainly been times that my test scores would not have suggested I would have done well in a challenging subject that I ultimately did just fine in. I wonder whether certain students will be held back in less rigorous courses if this is the case.
This online platform is an innovative way to leverage machine learning in order to improve both the quality and accessibility of education for students. I believe one of the key advantages here is the immediacy of feedback for students on their progress. Rather than wait for intermittent report cards, they can see real-time improvement in their results. There is also a time savings component for teachers or tutors to reduce the need to mark or grade assignments or tests. That being said, I am curious how Toppr thinks about its target market for this product, which although used by students, is paid for by parents. As such, in order to successfully transition into the paid model, I believe the company should communicate the value of the platform in a way that strongly resonates with parents.
I would be curious to see Toppr’s efficacy metrics. Do students who receive their personalized curriculum actually perform better than student’s who don’t, all else equal? While i agree that AI could be leveraged for personalizing content, i see it as a technology that complements rather than completely replace what teachers do.
Thank you for such an interesting article! Reading this piece made me think of all of the possible ways that the ability, through machine learning, of the platform to personalize the teaching to the specific student could have a real impact, outside of simply being a tutorial tool, for students with learning challenges. For example, if Toppr was able to build out a part of its platform that specifically helped students struggling with dyslexia or ADHD, this would be an incredible value add to these children’s lives, as a huge problem in the world of education (at least in the United States where I have experience) is students with special educational needs being unable to receive the help that they need during the school day given the demands on teacher’s times, especially in the public school systems (with typically larger classroom sizes than the private school systems). Developing exercises and drills to help students with these additional challenges that they could complete at home in order to help them better perform and learn during the school day would be hugely beneficial, and I am hopeful, after reading this article, that the possibility for such a platform to be created is imminently possible.