The Center for Data Innovation spoke with Oliver Kocsis, product manager at Novakid, a San Francisco-based startup that uses AI to power its English as a Second Language (ESL) platform. Kocsis explained how Novakid uses machine learning to deliver a customized curriculum for each student.
Gillian Diebold: How does AI increase student engagement in the learning process?
Oliver Kocsis: Using AI, we monitor and assess the progress of our students very closely and customize their curriculum to their needs. The assessment is completely seamless for our students. We do not want them to experience stress during our lessons as they would typically experience during traditional school exams. Our sophisticated system of competencies ensures the proper balance of one-on-one lessons, homework activities, and exciting mini-games to achieve the most efficient language learning pace tailored to individual students.
Recently, we have started to learn more about and measure the speaking behaviors of our students. We do this using deep learning machine learning algorithms. We are at the beginning of this journey, but the early data suggests a huge potential for AI to improve our students’ speaking skills.
Diebold: What differentiates Novakid from other ESL platforms?
Kocsis: Novakid is the leading ESL platform in Europe for children between the ages of 4 and 12. We were able to gain the trust of our students and parents via our professionally curated curriculum. This curriculum is not only tailored to different ages, focusing on the typical cognitive needs of children of different ages, but is also customized to every individual student.
Our platform lets our customers choose from thousands of teachers, based on their preferences. Our customers are not tied to a single teacher, they can switch from one to another whenever they’d like. Thanks to our education platform, a new teacher can continue exactly where the previous person left off.
Diebold: How do you mitigate issues of fairness when using AI for things like assessments?
Kocsis: We use a combination of well-defined algorithms (including machine learning) and human judgment to assess our students. This combination has provided us with the best results so far.
Diebold: Privacy is a major concern with education technology. What strategies or technologies do you use to alleviate those concerns?
Kocsis: Positive, safe content for children on online learning platforms is important. Some of the key elements are the usability of the content itself, the appropriateness for certain ages that navigate the platforms, and data usage and data protection we employ to protect students. We try to minimize the personal and sensitive information we collect and always cross-check every change we make.
Child-friendly content is crucial. All of the learning content at Novakid is maintained and updated on a regular basis whereas the process of its creation has several review stages to make sure it is positive for a young student.
An example of accurate content creation is the Novakid VR-based English course for children, Virtual Explorer. It contains over 200 carefully chosen and designed 360-degree interactive panoramic slides from the world capitals, airports, museums, and tourist places. Each slide passes through at least three stages of review by Novakid’s Content department to make sure that only child-friendly places are featured.
Taking into consideration the age range and child’s development abilities, we opted to create separate accounts for students and their parents. The Kid’s Area is a gamified learning environment for students where they can access the English lesson with their teacher, do homework, play games, get rewards, choose an avatar, and even watch animated comic stories. It is completely free of any commercial elements.
Moreover, the progress evaluation method at Novakid is based on over 1,500 parameters, A/B testing, and other components of a data-driven approach. We are planning to implement speech-to-text solutions and emotion recognition software to advance personalization even more. In this respect, the student data is crucial to create the educational experience in online English learning tailored to the student’s needs. As such, we only use certified cloud technologies to store the limited data we collect.
Overall, the privacy of a child is a paramount consideration for us. We implement industry best practices with regard to data management and access control. All of our data is encrypted in transit and at rest and we leverage AWS and Google Cloud services to ensure secure storage and processing. We also work with world-class partners such as Intercom and Stripe who have a strong reputation for the security of their services.
Novakid strictly protects the student data (attendance rate, grades, records of the lessons) as well as students’ and their parents’ personal information. Access to all the information is limited to Novakid staff and the student’s parents or guardians. As children’s personal data is processed by Novakid’s teachers as well, Novakid concludes a data processing contract with the teachers in order to prevent fraudulent or inappropriate activities.
Diebold: What trends do you anticipate taking hold in the EdTech space in the next five years?
Kocsis: The pandemic has fundamentally changed the perception of the EdTech industry. It was seen as a playground for technology fanatics and adults and is now an open space that welcomes all personalities and age groups. I believe augmented and virtual reality will take a bigger share in the industry to get students even closer to the topics they learn about. Ultimately, I hope this paradigm shift will enable high-quality and equal education in every corner of our world so that the next generation can live in a world without borders.