With several successful product launches under its belt, Learnosity’s AI Labs team is making it easier for product teams to integrate AI into their assessment workflows. Here, team member Georgios Kanellopoulos shares some motivations, current projects, and a glimpse into the future.
What do you do at Learnosity?
I am a Software Engineer and member of Learnosity’s AI Labs team, which leverages AI to improve education experience and assessment quality. I am responsible for building software that will allow educators to interact with large language models in an efficient way. I work on the whole software stack, from the database layer to the user interface.
How did you first get into AI? What was it that most appealed to you?
I was introduced to the world of AI during my postgraduate studies. I was immediately fascinated by the idea that machines could learn to solve problems that cannot be described with explicit rules. For example, the basic problem of optical character recognition showed me that AI models could learn patterns directly from data, without the need to program them step by step.
"It's impressive how these models can learn the language structure from vast amounts of text and complete challenging tasks related to text processing."
I was motivated to dive deeper into AI systems, and this led me to large language models. It’s impressive how these models can learn the language structure from vast amounts of text and complete challenging tasks related to text processing.
What I find exciting about my current role is the opportunity to leverage the capabilities of these powerful tools to improve assessment quality.
The AI Labs team is leading an exciting new era of innovation for Learnosity, and things are moving fast. What’s your experience been like so far?
My experience has been very positive! Being part of the AI Labs from its early steps has been both challenging and rewarding. To rapidly prototype new products my team got the chance to test new tools and libraries in practice. We have all gained much technical knowledge and learned how to integrate features in a coherent and user-friendly product.
"To rapidly prototype new products my team got the chance to test new tools and libraries in practice. We have all gained much technical knowledge and learned how to integrate features in a coherent and user-friendly product."
Working at this pace would not have been possible without a strong team culture. My team’s commitment to face-to-face collaboration has helped ideas move quickly from concept to reality. Management has been supportive and has successfully guided us through engineering challenges. This recipe allows Learnosity to become a forerunner in AI innovation for assessment.
Can you tell us a little about something you’re working on now and why it’s exciting?
I am currently working on the Item Bank Health Check project. The goal is to improve the quality of questions used in assessments. AI is used to perform diverse tasks on collections of exam questions, such as detecting bias, translating exams and improving accessibility by adding alternative text to pictures and transcriptions to videos. Users can access these features through a web application.
Item Bank Health Check is a product that helps educators deliver high-quality assessments. Using AI, they can quickly and efficiently process a large number of questions. The web application incorporates a “human-in-loop” approach by design, to ensure that all AI changes and additions are reviewed by educators before being presented to students.
Give us one big change you believe AI will have on assessment in the near-ish future?
AI could transform assessments in the near future through personalization. Currently, educators create assessments for all members of their class, without taking into consideration individual performance. AI could fill this gap by generating questions, or selecting the most appropriate from a pool, based on individual past performance.
To illustrate, let’s think that a student has taken an assessment designed by their teacher. AI could analyse the student’s answers and automatically create a follow-up assessment that focuses on the specific student’s weaknesses.
