Harish Agrawal is Director and Global Head of Products of KEA, a Magic EdTech product that focuses on the education and training sectors to deliver interactive digital solutions for publishing and deploying learning content.
In this interview, Mr. Agrawal shares his experience of building an AI-driven learning platform that can benefit learners of all ages, as well as educators and educational publishers, from K-12 education to corporate training.
This interview has been
featured in the AI for Education Initiative 2019.
What is your background and how did you get involved with KEA?
With an educational background in advanced computing, I have over 20 years of experience in spearheading the development of large-scale enterprise software products, right from the inception stage to the final rollout and marketing. Now, as the head of products at KEA, I have been managing the development, strategy, roadmap for digital learning products that redefine education and corporate training. Magic has been in the Learning and Education space for almost 30 years, we understand the needs of today’s learners and the challenges faced by corporates.
We have developed an innovative, next-gen, AI-driven learning experience platform (LXP), KEA. We built KEA off the success of MagicBox™, our Cloud-based SaaS and mobility platform focused on K-12 education. MagicBox™ already has over 3 million users worldwide, having logged in more than 5 million learning hours. We believe that KEA takes digital learning into the next generation, becoming the world’s first-ever AI-powered learning platform.
Who are KEA’s customers and how do you create value for them? How is AI leveraged in your product?
Kea is targeted towards Fintech and Healthcare companies. Most companies are launching new products and changes to existing products in a truly agile manner, Today’s LMSs and L&D teams are not geared for changing content at the speed of product launches and development. Kea allows Product managers, SMEs and other people who have knowledge about the product to quickly create content and distribute it. AI behind Kea learns from this content and make it easily discoverable.
Our intensive research revealed that when education and corporate training are delivered in a format that addresses the learning styles and preferences of the learner, the maximum benefits of education can be gained.
Our aim, while developing KEA, was to address the needs of the current and future generations of learners, who are digital natives and are very different from their preceding generations, not just in the way they seek learning, but also in the type of learning experiences that benefit them the most.
Keeping this in mind, our intensive research revealed that when education and corporate training are delivered in a format that addresses the learning styles and preferences of the learner, the maximum benefits of education can be gained. In addition, our research also showed that while digital natives crave technology solutions and prefer a self-directed approach to learning, they also value relationships and real-life communication and connections.
Tell us about your product KEA.
We believe that KEA will revolutionize the way in which learning content is delivered and consumed. Being the first-ever AI-powered LXP, it makes learning hugely engaging, intuitive and personalized. AI also helps the platform use deep learning to predict learner behavior, assess learning needs and pace of learning to present the most relevant content to the user. It also infers the preferred learning time, based on user activity, which allows it to schedule notifications and gives “nudges” to the user to complete a course or specific segment of training.
For learning content providers, it offers a faster way to create engaging content, using existing content to create multimedia courses. In addition, KEA enhances the ability to provide seamless learning experiences by allowing integration with most standard LMSes. It also allows the conversion of documents into “trackable” content so that the learning progress can be assessed. In addition to all this, it captures learner responses, such that course efficacy, preferred content formats and even learning patterns can be analyzed and/or predicted.
With the use of NLP, the platform auto-generates meta-tags for all content formats, including text, images, videos, presentations, etc., to improve discoverability and recommendations. Most importantly, with an attractive, easily navigable and intuitive user interface, it makes the entire process satisfying for all parties.
We believe that the greatest opportunity it offers is enhancing learning outcomes manifold, while helping publishers better monetize content, reducing costs and speeding up time to market.
The biggest advantage perhaps that AI has brought to learning is personalization. An AI-driven learning platform is able to customize content recommendations, based on the specific needs or knowledge gaps of each learner while personalizing content recommendations based on topics that the particular individual might not have mastered yet.
What are the major opportunities brought by AI for education today?
AI is changing the way people learn, easing continued learning for life. In addition, it is redefining how learners can be supported while allowing organizations to streamline their learning programs. For instance, AI can allow the automation routine administrative tasks that are not just tedious but time-consuming, such as course scheduling and even assessment of learners.
The biggest advantage perhaps that AI has brought to learning is personalization. An AI-driven learning platform is able to customize content recommendations, based on the specific needs or knowledge gaps of each learner while personalizing content recommendations based on topics that the particular individual might not have mastered yet. Most importantly, it allows each learner to proceed at their own pace, making the most of the learning content.
It is intriguing to see how AI is changing the way we look for and interact with information. We don’t even realize the many ways in which we interact with AI, from Google basing its search results on our location to Amazon cross-selling products based on our shopping history. Siri and Alexa have become ubiquitous in our everyday lives, so why not in learning?
In fact, when learners are faced with an AI-powered learning facilitator, they feel less intimidated about making mistakes. There is no fear of being judged if you don’t know the answer. When stress is low, learning outcomes improve.
Then there is the facility of learning and content usage analytics. Trainers might not always realize that some parts of their learning resources leave gaps in understanding for some learners or for specific concepts. However, when robust analytics are added to the picture, both trainers and learners can receive useful feedback. This means that while organizations can identify those that might need more training support. Usage analytics can also help determine which content formats work best for specific types of learning.
With AI, the entire process of aligning training to the needs and goals of each learner becomes not just possible but simple.
We aren’t looking at a science fiction future where a robot replaces human trainers. What we are aiming for is the perfect balance of the human touch and the advantages of AI and machine learning.
Which AI technology do you think will have the biggest impact on learning in the coming years, and why?
There is no denying that the future of learning for life will be shaped by AI technologies. What will have the maximum impact are natural language processing (NLP), machine learning and deep learning. These technologies will not just personalize learning like never before but make it universally accessible. KEA uses NLP effectively to learn from content and make it easily discoverable.
We aren’t looking at a science fiction future where a robot replaces human trainers. What we are aiming for is the perfect balance of the human touch and the advantages of AI and machine learning. Intelligent learning platforms can alter the entire learning experience while removing human bias and human error from analytics. Accurate, objective analytics also lend predictive capabilities. This could even help learners identify their own skill/knowledge gaps and choose to learn and grow their careers.
How should learning management software adapt for the advent of AI?
Standard learning management systems were created keeping in mind the needs of L&D professionals and content providers. This is why many of the existing LMSes are complex and cumbersome. The focus now has to shift to the needs of the modern learner, who seeks knowledge but prefers to do so at their preferred time and pace and on their preferred device. This doesn’t necessarily mean having a mobile app for LMS, what it also means is a deeper integration in the organization’s ecosystem i.e. Learning platform cannot be just another app in the organization. It needs to provide answers to the user when they need it as they are doing their job.
To make the most of what AI can offer for LMSes, such software also needs to move from the top-down approach they’ve been following, which often also makes them top-heavy. The end result is poor navigation and the lack of an intuitive user interface. They don’t really give control in the hands of the learner.
In fact, the platform itself moves into the background. What this means is that learners have direct access to content, without needing to cope with complex software or interfaces. Take the AI-powered chatbot, for instance. Rather than being a passive reservoir of information, the platform becomes an active learning assistant, using NLP and machine learning to answer questions and recommend content. This means that as the learner continues to use the platform, the assistant gets better and better at interacting with and engaging the learner.
The way forward will be a process of learning for both humans and machines. We will acquire greater skill in being able to present accurate data sets for AI to learn from and provide results. Strategies will get developed around such AI implementation.
What are the current technological limitations of AI which, once overcome, could bring massive improvements to the learning system?
I believe that AI is only limited by the data sets that we present to it. It has famously been said that “the machine doesn’t know that it doesn’t know.” So, it is limited by our algorithms. It again comes down to human bias and error, which can also creep into the data that we present for machine and deep learning.
So, what we need to first understand is the purpose for which we wish to use AI. There is no point is just using artificial intelligence for the sake of saying that something is powered by AI technology. The implementation of the technology has to be based on an end-goal. For instance, when we considered including an AI-powered personal learning assistant on KEA, our aim was to provide a means for learning to be made incredibly engaging and personalized, such that it was tailor-made for the needs of each learner. We wanted the AI assistant to be responsive to the learner, with machine learning allowing it to refine its interactions and content recommendations with use. The end goal was to achieve better learning outcomes while encouraging continued learning.
The way forward will be a process of learning for both humans and machines. We will acquire greater skill in being able to present accurate data sets for AI to learn from and provide results. Strategies will get developed around such AI implementation.
AI technologies of the future will make it easy for people to continue on a path of growth through life. The world and organizations are changing rapidly and we need to constantly upgrade our skills and knowledge about our own products to stay relevant. This is where AI will help us with.
