Building Aeriem: Adaptive learning for K-12 education

Sai Krishna V. K
28 min readJan 22, 2021


What do you do when you are disgruntled about your textbook? You go play cricket. But we decided to build an AI recommendation engine for learning material instead.

The undergrad engineering system in India

Undergrad Engineering in India is a peculiar beast.
Often, it is easy to exhaust 4 years through it all, only to realize that you might not have learnt as much as you thought you would. But to make it to undergrad engineering, you go through a walk of fire better known as engineering entrance examinations. It is competitive, brutal and forces students to begin preparing for it from as early as 5 years prior to the exams.

Unsurprisingly, India has emerged as a strong potential market for investments in training and education sector, due to its favorable demographics (young population) and being a services-driven economy. Further, India’s expanding role in sectors such as software development, generic pharmaceuticals and healthcare, would require the country to invest into learning and training segment as well.

The education sector of India is divided into two main segments; the core segment comprises of schools and higher education, while the non-core comprises of coaching classes, pre-schools and vocational trainings.

Competitive Exams define high-school and the drive to excel creates a fast-paced scenario where every mark counts and every method helps. With lakhs of students appearing across the country for exams such as the JEE, it is imperative that their training organization equip them with the best possible facilities in those crucial learning stages for them to hold the edge and ace their exams.

Back in 2012, Ed-Tech was far from the spotlight it currently has, and almost all of the activity around the education space was not digital. The classroom was the only meaningful source of teacher — student interaction, and thus, learning. In its race to be counted as a developed nation, it is important for India to educate its population and make them contributing citizens. India has the advantage of youth on its side with more than 50% of the population below 25 years of age, but this can become a potent force only if shaped properly. The increasing penetration of internet and mobile devices in this demographic, and the growth of online education market in India is indubitable too.

The iPad’s effect on learning

When Apple launched the iPhone back in 2007, not many could have predicted how massive the shift in the technology landscape it would end up causing. When the iPad was launched a few years later, the expectation was not any lesser.

At that time, the iPad promised to tackle the concept of productivity from a completely different perspective. If your worldview of productivity is limited to what can be done on a PC — documents, spreadsheets, presentations, coding — then of course you will produce a product that is like a PC, but worse for having tablet features. Of course you’ll produce a laptop.

If, though, your worldview of productivity is defined not by the PC, but rather by people — then you will produce a product that is nothing like a PC, but rather an intimate, responsive object that invites people in, and transforms itself into whatever you need it to be. You’ll produce an iPad.

Watching Steve Jobs at D8 in 2010, it’s clear he came up with the now famous “iPads are cars, PCs are trucks” analogy on the spur of the moment. The entire exchange, is worth a rewatch:

Steve Jobs: You know, uhm, [long pause], uh, I’m trying to think of a good analogy.

When we were an agrarian nation all cars were trucks because that’s what you needed on the farm. But as vehicles started to be used in the urban centers, and America started to move into those urban and then suburban centers, cars got more popular and innovations like automatic transmissions and power steering and things you didn’t care about in a truck as much started to become paramount in cars. And now, probably, I don’t know what the statistics, maybe 1 out of every 25 vehicles, 30 vehicles is a truck, where it used to be 100%.

PCs are going to be like trucks. They’re still going to be around. They’re still going to have a lot of value. But they’re going to be used like 1 out of X people.

The iPad is a truly disruptive product. It is inferior to a PC on the attributes that matter to PC-users, even as it excels on orthogonal attributes that appeal to a new type of customer. Those orthogonal attributes certainly include things like portability and battery life, features highlighted in last week’s event. But the more essential attributes are those that make the iPad very much not a PC. The immersiveness of apps, and responsiveness of touch. The safety of iOS, and the discoverability of the app store.

There could not be a better proposition for the iPad than the classroom. The past years have seen a sea change in the use of technology in special education. The introduction of the iPad, followed by numerous other tablets, has put technology into the hands of students in a way unprecedented in the years before. But this transition was bumpy, and the curve of adoption falls right off the cliff outside USA, and outside the territory of the iPad. Why necessarily did Android tablets in India fail to replicate the success of the iPad in USA’s classrooms? The answer starts with a clone of the iPad, or an attempted one, at about a tenth of the cost.

India’s moment with tablets beginning with Aakash

Imagine for a moment you’re an entrepreneur looking at a stubborn problem — in this case, a mass of a few hundred million poor, uneducated people. To lift them out of poverty, friends who study economics tell you the first thing you ought to do is offer them access to affordable education. And that if you can, you’ll achieve three things. Create a better world; create an incredibly compelling business; and perhaps get a stab at a completely different country.

There are two ways to go about the problem. The first, you reckon you ought to think through the problem. That means look at the world around you, tinker with ideas, figure what works best, and build a cost-effective solution that eventually helps achieve the objectives stated above.

The second is a pig-headed one. Look at how others around the world are attempting to crack the problem; call in the global media; tell them a tablet-like device with a touch screen can be built and sold at $35; another matter altogether you’ve got no clue how to go about it or why; and then try your darndest best for a stab at glory. In any case, as long as the problem is cracked, who gives a damn? With the benefit of hindsight, it is obvious the Ministry of Human Resource Development (MHRD) chose the second option. What else explains the fact that back in 2011, the ministry announced it is in the middle of developing a low-cost computing device for students that would cost just $35? And that when complete, a global tender for five million units of the device would be floated? The blitz that accompanied the announcement had the world in a tizzy. Including me.

A little less than a year later, in February 2011, the Indian Institute of Technology (IIT), Jodhpur, which had taken upon itself the onus to decide what specifications this animal would run on, put out a global tender to build the first 10,000 units. In return for these services, the institution received Rs 47 crore from the government. DataWind, a 12-year-old Canadian company with subsidiaries in the UK and India won the contract, produced a prototype built to spec. To put it mildly, the prototype was a disaster. Some phones in the market worked faster than this contraption. The battery couldn’t last two hours if a user tried to play video files on it. The touch screen, well, wasn’t “touchy” enough.

The most remarkable Aakash data point was its price: $35. Meant for the millions of students who can’t even afford textbooks, the Aakash was supposed to be India’s iPad knockoff. So what happened?

The then Education minister announced that the government would distribute 10 million at the subsidized $35 price, while millions more would be available for $60 apiece. The device would have videoconferencing capability, a touch screen, and three hours of battery life–not to mention the ability to turn around India’s global reputation. In India, that’s a quarter the cost of competing tablets with identical specifications. Similar tablets in China, the world champion in low-cost components and manufacturing, go for $45 and up, wholesale. Which means the Aakash 2 isn’t just the cheapest fully functional tablet PC on the planet because the Indian government has decided it should be — it’s the cheapest, period.

For all its success at churning out engineers by the hundreds of thousands and sending Silicon Valley countless gifted computer scientists, India has never been much good at producing hardware. This is why people seemed so eager and triumphant, and why the Aakash is so significant. In this one device, you can find the extraordinarily high hopes not just of one ambitious politician but of an entire nation. Or, rather, you would be able to find such things–if you could find an Aakash at all.

The Aakash, you see, never made it to market.

In the developing world, and especially in India, a country where one billion people have a monthly income less than $200, every rupee matters. Aakash means “blue sky” in Hindi, and that’s a fair description of Datawind’s goals for the tablet. Ultimately, says Tuli, the government would like to distribute one to each of India’s 220 million students. India has 900 million cell phone subscriptions, but in a country where smartphones are rare, 95% of Indians have no computing device. Which means the Aakash, or something like it, could become the sole computer for hundreds of millions of people in India, not to mention elsewhere in the developing world.

Even before Aakash was unveiled in 2011, there were attempts to produce a cheap, disruptive device that would break down the divide between the digital haves and the have-nots and transform education, among other areas of public and private life in the country.

Job to be done: need for a blended solution

Before the Mobilis, there was the Simputer, which stood for a simple, inexpensive and multilingual people’s computer. The hand-held low-cost computing device was introduced in 2002 by the Simputer Trust, a non-profit organization formed by seven Indian scientists and engineers. It was touted as a device that would change the low-cost computing ecosystem in the country. By 2011, both the Simputer and the Mobilis had faded from public memory,

The government’s model for technological acceptance is the way in which the mobile phone has been embraced across the nation of 1.2 billion people. India, where the mobile phone made its entry in 1995, has become the second fastest growing wireless market in the world, after China. Thanks to the runaway success of Apple’s iPad, tablets have gained currency and credibility — that was the backdrop to the introduction of Aakash. But tablets haven’t become as widely adopted in India as the high level of interest in them might imply.

The Aakash type of low-cost initiatives, tied as they are to education, have thus far run out of steam for lack of adequate supporting infrastructure, which includes content, unlimited data plans and applications, besides a robust distribution network, according to analysts, manufacturers and industry experts. In addition to this, the user experience with the first iteration of Aakash was disappointing, given its resistive touch screen, and the aforementioned lack of apps, content and a good data plan.

Aakash was a really noble idea, which wanted to equip every college-going student with a computing device, but the specifications were not really supportive and the required content also was not adequate in the version launched. For a device like Aakash to be successful, it has to deliver a great consumer experience. Apple has raised the benchmark for tablets and Indians will not settle for anything less. They will want 3G-like speeds and unlimited bandwidth. Other than these factors, manufacturers who get into this game will also have to focus on logistics, distribution — in short, the entire computing ecosystem.

In a country where education is serious business, the archaic classroom set-up is a challenge to effectively transfer knowledge. Taking that experience beyond the classroom and aiding the student to equip better through interactive learning experiences is what Aeriem aims to achieve.

With a large student population, India uniquely requires varied type(s) of educational tools across multiple streams of knowledge. The Tablet PC — as an Internet enabled device — provides a powerful platform and an ecosystem to deliver quality education to the student fraternity.

Tablet PCs have the potential to address many of the challenges that are faced by Test Preparation Institutes — changing exam patterns, evolving assessment paradigms and the sheer growth in number of students appearing for these competitive exams

The tablets have succeeded so quickly across the rest of the world in part because they are portable, intuitive to use and bring a new dimension to digital learning.

Aakash died a slow death, but the idea remained before, and stood after

The race to build the world’s cheapest computing device started when the One Laptop Per Child (OLPC) project was announced in 2005. Headed by Nicholas Negroponte, best known as the founder of MIT’s Media Labs, it was a non-profit entity and funded by global majors like AMD, Google and Nortel among others. The central theme to this idea was to build a laptop that would cost no more than $100.

They provide a different modality of learning, one that has an element of fun.

The most successful teaching for children with special education needs involves visual, auditory and kinesthetic (tactile) learning. The software is self-paced and allows students to skip quickly through parts they are strong in and spend more time on the parts the struggle with.

The Opportunity: Full Stack Learning at Indian Education Centres

When I was a student, I spent much of my time watching this unfold from the sidelines. The opportunity to meaningfully impact digital education in the country was large. My Co-Founder and I were barely 18 and what we lacked in experience, we compensated in excitement. We chose to put together the building blocks for a learning solution we’d have liked to use

At Aeriem have a simple goal that we set out to achieve; herald the Post PC learning ecosystem into the country and uplift the quality of education nationally. We realized the solution is not a race to the bottom on cheap hardware, but rather a job to be done from a student’s shoes. Broadly, the important pillars behind a good digital learning solution were:

  • The hardware
  • The operating system
  • The application ecosystem
  • The content and learning materials
  • The cloud as a delivery medium for pedagogy

To do just one of these aspects, at scale, to any degree of success is a daunting task but we thought it’s best to tackle all of them, together, at once. Without saying, it meant some of the most consistent periods of all-nighters, missing classes and even exams to put time, attention and energy into building Aeriem.

  1. We designed the shell and sourced our own tablets from China
  2. We built a learning management system that was actually useable on a tablet
  3. We partnered with a popular ed-tech app vendor for a whole suite of additional apps
  4. We built the LMS to be cloud-native and not rely on an SD card, which was still the dominant method of content transfer back then
  5. We pitched to the largest test prep institutes in the country to make their content power the device

A design house for the future of tablet learning: Aeriem

Aeriem is an innovation driven organization that aims to take the classroom experience to the next level in a new age of connected devices. We believe in transforming the way in which learning takes place and redefine the teacher student interaction. Aeriem endeavors to create new benchmarks and truly energize the learning ecosystem and usher in a new era of learning, one that leverages the power of the information age to complement traditional methods. Focusing on trust, quality and reliability, we aim to deliver a truly unique, engaging & inspiring experience for all your stakeholders. We enable you to make this transition to the future in a seamless and intuitive manner. “A great learning solution is not merely software excellence or hardware expertise; it is an amalgamation of the stellar design with a deep understanding of the human thirst for knowledge. Aeriem helps you redefine the interactions in your classrooms and helps you surpass traditional barriers to enable your stakeholders reach their fullest potential” Founded by a team passionate about redefining the way interact with technology, we are a Bangalore based Design House with qualified manufacturing units in Shenzhen, China. Aeriem boasts of a young product team drawn from premier institutions of the country. This enables us to deliver certainty for our institutional partners

Aeriem is the single most comprehensive solution to the problem of a fragmented user experience for its tight integration of services and products helps achieve seamless interaction with the device and sustained usage, a joy

Over a period of time it became clear that almost every part of this stack was eventually replaceable if so desired. You could have the same solution with better hardware, better applications and even from a different institute or content provider. The one thing that ties all of this together as a technology moat? The core learning management system. We called it the Aeriem LMS.

Why an LMS?

A robust, high-quality LMS is a vital tool to the success of any online course and can make or break an institution’s reputation in the highly competitive education market. LMSs not only need to provide content to learners, but they must also facilitate timely and accurate communication between learners, course facilitators, and other institutional stakeholders.

While there are several definitions of a learning management system (LMS), the basic description is a software application that automates the administration, tracking, and reporting of training events. However, it’s not that simple. A robust LMS should be able to do the following:

  • centralize and automate administration
    • use self-service and self-guided services
    • assemble and deliver learning content rapidly
    • consolidate training initiatives on a scalable web-based platform
    • support portability and standards
    • personalize content and enable knowledge reuse

Back in 2012, these learning management systems were far from being usable on a tablet PC. Software vendors shipped poor products, existing products were built for the web, and poorly translated to an app and open-source alternatives were simply not focused enough on the specific Indian learner’s journey or use case.

We started with the basics and built from there on, for a full featured LMS that was purpose built for how India learns. The broad features included:

Mobile Learning Environment
The mobile learning environment provides a rich set of productivity tools that helps the student to generate his/her study plan and keep track of the progress with usage analysis.

Multi Publisher Textbook Catalog
All your prescribed textbooks from the leading publishers available in rich interactive format.

Author Content
Easy tools for teachers to create rich “mobile ready” content that can easily be shared with students.

Interactive Textbook Reader
Textbook functionality which supports pdf, epub2, epub3, and LivePage textbooks with in-built quizzes and full annotation like free pen highlighting and note taking.

We made some course corrections to include multiple delivery models too. This included:
Deliver On Micro SD Card
Deliver On USB Device
Deliver From Cloud
Deliver From On Premise Server

As any parent can tell you, kids love touchscreen devices. And Nielsen’s now dated quarterly Connected Devices Report found that during Q1 2013 more than three-quarters of tablet-owning parents (78%) let their kids under age 11 play with tablets in their homes. The majority of these parents (54%) said their children use tablets for educational purposes. What’s more, among those who don’t let their kids play with tablets, 20 percent said they would if there was more educational content on them. This was compelling data for us to build Aeriem atop of. The thesis was that tablets are bringing a level of ubiquitous computing unlike any other device

A ubiquitous learning environment offered by tablet technology, keeps the students engrossed in the learning process and more importantly teachers are able to adopt the tablets due to its ease of use, small form factor and most importantly affordability.

Project Acetab

To build Aeirem’s vision out, we decided to partner with an educational institution on the launch of a custom branded hardware solution better known as Project Acetab.

The AceTab project witnessed a pilot rollout in 2013, with devices provided to students of select campuses and course material loaded in them. Post device deployment surveys were conducted and the results indicated that the hardware fell short of satisfaction. Evident problems in core features such as battery and touchscreen were proving to be a hindrance between the student and the accessible content.

Some of the data we received through the pilot were promising:

This concept was competing squarely with legacy approaches such as:

1. Sale of SD Card

2. Sale of USB/DVD/PC based media

3. Sale of Cloud based web portal to access content

4. App based business model

5. BYOD (Bring Your Own device) deployment

6. Restricted device option deployment

7. Branded device deployment

It was during this time that we relized working with educational institutions is a notoriously difficult process. We had to justify the smallest of things, through multiple iterations. We were still bullish on the LMS for project Acetab, as the pros outweighed the cons:

Pros of an App based LMS:

  • Any tablet/smartphone accessible
  • App makes sure that the content is pre encrypted from the cloud and can be accessed only securely
  • The app enables you to store offline content in a restricted and a safe manner
  • Highly customizable to the needs of the educational institution, and also can be provided with some robust featuresets
  • Easily deployable and efficient student demographic reach
  • Analytics and test taking options with detailed answers can be provided
  • This scenario in essence is the software offering that is allowed on any hardware platform

Cons of an App based LMS:

  • Creates the problem of software fragmentation which significantly increases the upkeep cost of debugging and optimizing for every device
  • Internet access control cannot be restricted since the devices can range from smartphones to tablets of varying sizes and specifications
  • 3rd party apps can be accessed but there is no method to control the access, anything can be downloaded from the App Market
  • No form of content restriction in terms of an SD card deployment is possible. Any locally stored content must be on the device itself.
  • The app will be a standalone addition to the device and not a complete part of the device core functionality
  • The App cannot have system level privileges. Hence, it cannot function core system level tasks such as access to nested applications
  • The App is more service oriented than device oriented and represents a fundamental change in the AceTAB project

We ended up rolling out these solutions to thousands of learners over the period of time it was active. The LMS eventually was too brittle, and needed a more intelligent system to help surface the right content for the student to learn from. Something that was deeply personal. Almost like the Netflix recommendation engine, or the Spotify algorithm. In the education space, this was better known as adaptive learning. We then set out to build these adaptive learning features into the Aeriem LMS.

Adaptive Learning Management Systems:

College students and high school seniors believe that tablets are just as valuable for educational purposes as they are for personal entertainment.

Adaptive learning is an educational method which uses computers as interactive teaching devices. Computers adapt the presentation of educational material according to students’ learning needs, as indicated by their responses to questions and tasks. The technology encompasses aspects derived from various fields of study including computer science, education, and psychology.

The focused program on Adaptive Learning Systems supports high-risk research to accelerate development of this infrastructure. Adaptive learning systems emphasize:

● systems that readily adapt the learning experience to the skills and needs of the learner

● systems that are flexible and scalable, that allow the content and courseware to be widely reusable and easily organized in different ways at different instructional levels

● systems that provide content development tools that function across a broad range of platforms, easily used by domain experts who are not programmers — systems that are adaptable to the educators as well as the learners.


● The costs of producing and disseminating educational content would drop.

● The user community for instructional systems would expand and become more diverse. Distributed instructional systems would be manageable and provide a high quality of service.

● Learning performance and workplace productivity would increase.

● Training and learning would become more accessible than ever before, to the benefit of workers in small and medium-scale enterprises and others in need

Adaptive learning has been partially driven by a realization that tailored learning cannot be achieved on a large-scale using traditional, non-adaptive approaches. Adaptive learning systems endeavor to transform the learner from passive receptor of information to collaborator in the educational process. Adaptive learning systems’ primary application is in education, but another popular application is business training. They have been designed as both desktop computer applications and web applications.

Adaptive learning has also been known as adaptive educational hypermedia, computer-based learning, adaptive instruction, intelligent tutoring systems, and computer-based pedagogical agents

Some basic features that were built out as a part of the Adaptive Learning Management System were:

Reading Software and Ecosystem

Reading software for e-textbooks must combine the simplicity and convenience of the printed book with the unique and complex ways that students interact with educational content. The software should mimic the physical book where appropriate (e.g., page numbering) and enhance studying through features unique to the digital format (e.g., full-text search). To accommodate students’ tendency to collect and refer back to other content sources, the software should also accept external content such as personal notes and/or a professor’s PowerPoint lectures.

During our research, we found that students wanted features that replicated the physical book experience, such as highlighting and note-taking. However, because this was a “digital” format, students also expected features that were unique to the digital environment, such as quick Internet searches, tagging content, and automated study-guide creation. Commonly requested features included the following:

Highlighting and annotation (i.e., book markup)

Content tagging

Full-text search within and across content

Faculty sharing of annotation and highlights

Integration with other content sources and learning management systems

Integrated web resource lookup

Study tools such as study-guide creators or flashcards

Adaptive learning or Intelligent tutoring has its origins in the artificial-intelligence movement and began gaining popularity in the 1970s. It has been a technique that’s been tried a few times to be implemented however has failed to attain a global audience in terms of implementation and active use.

Bottom Line: Historically, a very very heralded concept. Its something people look up to, but have not been able to find an answer. The closest parallel would be Google Self Driving cars, its something that is a lot on hype but not a real product yet. Similarly, adaptive learning is something every western educator propagates, but theres no real product to do it yet.

Overall, the features that were most popular with students were those that allowed them to create a comfortable and organized study environment. Our research clearly indicates that a student-focused study platform should be built around tools for content organization because studying, at its core, is an exercise in organization and access.

Over time, digital content will explode out of the context of the printed book. Content will be offered at the micro level, such as a chapter at a time. We could also see never-ending links to additional content — whereby accounting content bleeds into finance content which bleeds into economics content, giving faculty the ability to cut out the exact patches of content they want. Digital platforms have also unlocked opportunities for the creation of open-source and free content, apart from textbook publishers.

At the end of the day, it is about connecting the dots in the student ecosystem. Making the standard classroom adapt better to a student and adding on the experience of the classroom to the students’ fingertips are two broad goals that these devices would ultimately achieve.

Mobile Learning is about self-actuated personalization.

When building the Adaptive learning environment, we had the following points in mind:

1. Access

A mobile learning environment is about access to content, peers, experts, portfolio artifacts, credible sources, and previous thinking on relevant topics. It can be actuated via a smartphone or iPad, laptop or in-person, but access is constant–which in turn shifts a unique burden to learn on the shoulders of the student.

2. Metrics

As mobile learning is a blend of the digital and physical, diverse metrics (i.e., measures) of understanding and “performance of knowledge” will be available.

3. Cloud

The cloud is the enabler of “smart” mobility. With access to the cloud, all data sources and project materials are constantly available, allowing for previously inaccessible levels and styles of revision and collaboration.

4. Transparent

Transparency is the natural byproduct of connectivity, mobility, and collaboration. As planning, thinking, performance, and reflection are both mobile and digital, they gain an immediate audience with both local and global communities through social media platforms from twitter to facebook, edmodo to instagram.

5. Play

Play is one of the primary characteristics of authentic, progressive learning, both a cause and effect of an engaged mind. In a mobile learning environment learners are encountering a dynamic and often unplanned set of data, domains, and collaborators, changing the tone of learning from academic and compliant to personal and playful.

6. Asynchronous

Among the most powerful principles of mobile learning is asynchronous access. This unbolts an educational environment from a school floor and allows it to move anywhere, anytime in pursuit of truly entrepreneurial learning. It also enables a learning experience that is increasingly personalized: just in time, just enough, just for me.

7. Self-Actuated

With asynchronous access to content, peers, and experts comes the potential for self-actuation. Here, learners plan topic, sequence, audience, and application via facilitation of teachers who now act as experts of resource and assessment.

8. Diverse

With mobility comes diversity. As learning environments change constantly, that fluidity becomes a norm that provides a stream of new ideas, unexpected challenges, and constant opportunities for revision and application of thinking. Audiences are diverse, as are the environments data is being gleaned from and delivered to.

9. Curation

Apps and mobile devices can not only support curation, but can do so better than even the most caffeine-laced teacher might hope to. By design, these technologies adapt to learners, store files, publish thinking, and connect learners, making curation a matter of process rather than ability.

10. Blending

A mobile learning environment will always represent a blending of sorts–physical movement, personal communication, and digital interaction.

11. Always-On

Always-on learning is self-actuated, spontaneous, iterative, and recursive. There is a persistent need for information access, cognitive reflection, and interdependent function through mobile devices. It is also embedded in communities capable of intimate and natural interaction with students.

12. Authentic

All of the previous 11 principles yield an authenticity to learning that is impossible to reproduce in a classroom. They also ultimately converge to enable experiences that are truly personalized

Adaptive learning systems have traditionally been divided into separate components or ‘models’

These models are nothing but content delivery mechanisms in this context.

● Expert model — The model with the information which is to be taught

● Student model — The model which tracks and learns about the student

● Instructional model — The model which actually conveys the information

● Instructional environment — The user interface for interacting with the system

Expert model

The expert model stores information about the material which is being taught. This can be as simple as the solutions for the question set but it can also include lessons and tutorials and, in more sophisticated systems, even expert methodologies to illustrate approaches to the questions.

Bottomline: Expert model is the teacher’s guide to delivering content to the student. Nothing adaptive in terms of the teacher delivery system but this would help set course a classroom-based environment in the blended learning ecosystem.

Student model

Student model algorithms have been a rich research area over the past twenty years.

Biggest Success story: CAT & AIEEE exams

In CAT & AIEEE, the subject is presented with questions that are selected based on their level of difficulty in relation to the presumed skill level of the subject. As the test proceeds, the computer adjusts the subject’s score based on their answers, continuously fine-tuning the score by selecting questions from a narrower range of difficulty.

An algorithm for a CAT-style assessment is simple to implement.

1. A large pool of questions is amassed and rated according to difficulty, either through expert analysis, experimentation, or a combination of the two.

2. The computer then performs what is essentially a binary search, always giving the subject a question which is half way between what the computer has already determined to be the subject’s maximum and minimum possible skill levels.

3. These levels are then adjusted to the level of the difficulty of the question, reassigning the minimum if the subject answered correctly, and the maximum if the subject answered incorrectly. Obviously, a certain margin for error has to be built in to allow for scenarios where the subject’s answer is not indicative of their true skill level but simply coincidental.

4. Asking multiple questions from one level of difficulty greatly reduces the probability of a misleading answer, and allowing the range to grow beyond the assumed skill level can compensate for possible misevaluations.

Richer student model algorithms look to determine causality and provide a more extensive diagnosis of student’s weaknesses by linking ‘concepts’ to questions and defining strengths and weaknesses in terms of concepts rather than simple ‘levels’ of ability.

Because multiple concepts can influence a single question, questions have to be linked to all relevant concepts. For example, a matrix can list binary values (or even scores) for the intersection of every concept and every question.

Then, conditional probability values have to be calculated to reflect the likelihood that a student who is weak in a particular concept will fail to correctly answer a particular question. A student takes a test, the probabilities of weakness in all concepts conditional on incorrect answers in all questions can be calculated using Bayes’ Law (these adaptive learning methods are often called bayesian algorithms)

Outtake: There are proven algorithms that scientifically lead to better student testing. Bayes Law identifies weaknesses through the linking of concepts.
Real life scenario will be :

Question | Tag difficulty level | Tag all concepts relevant

Student attempts

Algorithms runs a search based on Correct(1) or wrong answer(0)

Algorithm stores the question result in memory and accordingly selects the next Q

Instructional model

The instructional model generally looks to incorporate the best educational tools that technology has to offer (such as multimedia presentations) with expert teacher advice for presentation methods. In a CAT-style student model, the instructional model will simply rank lessons in correspondence with the ranks for the question pool. When the student’s level has been satisfactorily determined, the instructional model provides the appropriate lesson. The more advanced student models which assess based on concepts need an instructional model which organizes its lessons by concept as well. The instructional model can be designed to analyze the collection of weaknesses and tailor a lesson plan accordingly.

When the incorrect answers are being evaluated by the student model, some systems look to provide feedback to the actual questions in the form of ‘hints’. As the student makes mistakes, useful suggestions pop up such as “look carefully at the sign of the number”. generic concept-based hints being offered based on concept weaknesses, or the hints can be question-specific in which case the student, instructional, and expert models all overlap.

Classroom Implementation

Adaptive learning is also called intelligent tutoring system.


● Systems need to be able to dynamically adapt to the skills and abilities of a student.

● Environments utilize cognitive modeling to provide feedback to the student while assessing student abilities and adapting the curriculum based upon past student performance.

● Inductive logic programming (ILP) is a way to bring together inductive learning and logic programming to an Adaptive Learning System. Systems using ILP are able to create hypothesis from examples demonstrated to it by the programmer or educator and then use those experiences to develop new knowledge to guide the student down paths to correct answers.

● Systems must have the ability to be flexible and allow for easy addition of new content.

● Cost of developing new Adaptive Learning Systems is often prohibitive to educational institutions so re-usability is essential.

● School districts have specific curriculum that the system needs to utilize to be effective for the district. Algorithms and cognitive models should be broad enough to teach mathematics, science, and language.

● Systems need to also adapt to the skill level of the educators.

● Many educators and domain experts are not skilled in programming or simply do not have enough time to demonstrate complex examples to the system so it should adapt to the abilities of educators.

Distance learning implementation

The field of distance learning is now incorporating aspects of adaptive learning. Initial systems without adaptive learning were able to provide automated feedback to students who are presented questions from a preselected question bank. That approach however lacks the guidance which teachers in the classroom can provide. Current trends in distance learning call for the use of adaptive learning to implement intelligent dynamic behavior in the learning environment.

During the time a student spends learning a new concept they are tested on their abilities and databases track their progress using one of the models. The latest generation of distance learning systems take into account the students’ answers and adapt themselves to the student’s cognitive abilities using a concept called ‘cognitive scaffolding’.

Cognitive scaffolding is the ability of an automated learning system to create a cognitive path of assessment from lowest to highest based on the demonstrated cognitive abilities.

● Personalization of links to information sources based on the user’s stated interests or the user’s surfing habits. This is viewed as the biggest asset adaptive learning was and IS bringing to the table. Hence, it looks to be a case of stagnation in terms of an innovative content delivery as well as any kind of a real time adaptive method

Adaptive learning seems a big deal, why isn’t it a hotter field of innovation?

Several factors are at work.

The cost and complexity of producing instructional software and systems is prohibitive (except for certain narrow applications, such as selected children’s software.)

Educational software and systems are not easily usable for many learners and educators, and for that matter present obstacles for educational institutions.

Educational systems are increasingly interactive and difficult to manage at the institutional level. While this problem is not unique to educational markets, it presents an unusually high barrier given the skills and organizational dynamics of today’s education and training organizations.

The necessary business models and key transactions that enable the missions of knowledge-driven institutions are not yet adapted to computers and distributed systems.

Equally problematic: educational networks do not yet offer sufficiently high reliability to become a viable alternative to many educational media, such as desktop systems or traditional classroom techniques.

Pressure points:

● Content — support for efficient, cost-effective production of instructional components that emphasizes interoperability and reusability, as well as the import or transformation of legacy data.

● Modes of Delivery — solutions for synchronous and asynchronous collaboration, both in content development and instructional use, of data that might be distributed widely, and that supports the flexible adaptation of the content to support different learning needs.

● Search and Retrieval — highly interactive infrastructure technologies to support educators, students, and workers with complex information acquisition and management requirements and that deliver outputs precisely tailored to the needs and constraints of the user.

● Efficient Performance Support — middleware applications that are robust and reliable for support of high-performance applications even when scaled-up and extended. Special transactions specific to education, and training, such as translation, accreditation, learner evaluation and collaboration around learning tasks will have to be supported.

Ed-Tech & Learning apps since then

A lot has happened in the space over the years of 2018–2020. When we started Aeriem a number of other learning apps were just beginning to get going as well. A name that used to be circulating often then, and now, is Byjus. The

Since 2013, the world of ed-tech has evolved dramatically and in 2020, is witnessing its strongest moments yet in countries like India. At Aeriem, we were early to the market, missed the timing by a few years and learnt a lot through the process in terms of going in search of product-market fit. I remain bullish on the possibilities of building algorithmic curation services for education and believe there’s value at scale, in the billions. Aeriem was one attempt at possibly unlocking that value, one tablet at a time.



Sai Krishna V. K

I write about possibilities in the Metaverse & productivity in the Meatverse ♦︎ Founder, Scapic (acquired by Flipkart)