TLDR

India’s New Education Policy 2020 has been revised after 28 years. This article provides an analysis of the policy from the lens of Artificial Intelligence. The following are the salient points.

• Embrace Artificial Intelligence instead of fearing it
• Increase the role of Alumni in the education system 
• Embrace massive open online classes (MOOCs)
• Break the language barrier via AI / Natural Language Processing
• Setup a separate AI module in the math curriculum 
• Differentiate between Core AI research and Applied AI subjects
• Provide adequate compute hardware like GPUs for Researchers
• Data is the new oil. Build a comprehensive data policy. 
• Encourage Industry funding in education via corporate social responsibility.
• Inculcate a culture of open collaboration via conferences & published papers

Introduction

The Government of India has come up with the New Education Policy under the able leadership of Dr. Kasturirangan. Social media in India has been buzzing about NEP 2020, and I was pleasantly surprised to see the emphasis given to Artificial Intelligence in the policy document. This article highlights the salient points in the policy concerning Artificial Intelligence and proposes a few areas of improvement. 

1. Embrace AI

The policy document acknowledges the rise of artificial intelligence and its power to shape the future. The document captures the need for AI education and puts forward a balanced policy to harness AI. It acknowledges the fact that there will be a shift in the job landscape with the advent of AI. Rather than fearing job losses, the policy advocates for embracing the change. 

The document proposes the introduction of AI in the curriculum for school and college students. It also talks about upgrading skills of professionals to apply AI to various sectors like Agriculture, Industries, etc. Kudos to the policymakers in identifying the opportunity and laying out a systematic approach in the policy document.

2. Alumni Network: Key to ongoing improvement

The world continues to see technological advancements at an ever-increasing pace. A while back, it was computers and today its AI, blockchain, etc. To keep up with the pace of innovations, our education system should build stronger ties with the alumni network. Our academia should not wait for new education policy to adopt new technologies like Artificial Intelligence.

Academia and students should be encouraged to engage with their alumni via professional social networks like Linkedin. On the other front, professionals should be encouraged to contribute financially to their alma mater via income tax breaks and secure donation process. 

3. Rise of Massive Open Online Course (MOOCs)

The world is moving towards Massive open online education platforms like Khan Academy, Edx, Byjus, Coursera, and Udacity. Apart from these platforms, universities and companies are also offering an online version of relevant courses on the internet. Some of the best minds in the world have prepared these courses in the respective field. These MOOCs also encourage global collaboration among students and help build a fraternity. These top courses and interaction with the best minds will be crucial to be on the cutting edge of Artificial Intelligence.

The New education policy touches upon the topic of online teaching platforms like SWAYAM but falls short due to concerns with the Digital Divide. The external MOOCs can be leveraged by sourcing top content and making it affordable and accessible to the Indian student community.

4. Preserve language and break barriers

India has a diverse culture with thousands of spoken languages and hundreds of scripts. AI-based Natural Language Processing applications can digitize the languages and preserve them for future generations. Additionally, Machine Translations technologies can help break language barriers and encourage cross-collaboration. The day is not far when a Gujarati teacher can teach the symphonic poems by Rabindranath Tagor to a Kashmiri student residing in Kanyakumari. 

India should either invest in developing these apps or collaborate with tech giants like Google and Facebook to open source the same. Following AI capabilities should be built for Indian languages and open-sourced for easy adoption.

• Speech Recognition: The field of Speech recognition has seen a boost with the advent of voice assistants like Siri, Alexa, and Google Assistant. It can convert spoken language into text. (Speech-To-Text). Apart from that Text-to-Speech models should also be developed.
• Optical Character Recognition: OCR is critical to preserve our scripts. Out of the box, OCR libraries are mostly available for English Alphabets. OCR for all Indian languages should be developed. It will help revive ancient Indian scriptures and their teachings.
• Machine Translation: Machine Translation will allow interoperability between various languages in India. It will help with cultural mixture and diffusion of ideas and values across India. 
• Indian Language Benchmarks: Benchmarks are used to evaluate the efficiency of various models. Having robust benchmarks and open-source datasets will boost the development of NLP models. 

5. Math for AI

AI can be split into two distinct eras, the traditional machine learning era and the current deep learning era. Traditional machine learning algorithms like Decision Trees, Support Vector Machines, etc. require algorithm-specific math concepts that are not generalizable. In comparison, Deep learning uses basic math building blocks called neural nets, which can be stacked in layers to build complex algorithms. Deep learning techniques became popular in the 2010s and have taken the world of artificial intelligence by storm beating many benchmarks, including human performance.

In my humble opinion, the current CBSE math curriculum covers most of the math concepts required for deep learning. Topics such as probability, statistics, matrix, and calculus taught today are foundational to DL, but they are part of independent modules. Math concepts required for AI should be collated under a common module, just like calculus, algebra, and trigonometry. AI-related math concepts can be incrementally introduced to the students as part of this module, depending on the grade level.

6. AI research vs. AI applications 

The policy document suggests the introduction of AI-related subjects in a Ph.D. But we should distinguish between Core AI research and Applied AI. 

Core AI Research involves coming up with newer approaches for machine learning, new neural architectures, etc. whereas Applied AI consists of using these algorithms in solving domain-specific problems. For example, finding an efficient AI algorithm to identify objects in an image will involve core research. But the object identification algorithms can be applied to a wide variety of tasks from self-driving cars to agricultural produce sorting. 

Topics related to Applied AI should be introduced much earlier in the curriculum. The AI community has done a wonderful job of packaging common AI Algorithms as pre-built open source packages. The usage of these packages should be part of coding practice in pre-college and undergraduate itself. 

7. Access to Hardware like GPUs for AI Research

Recent deep learning models like BERT, ResNet, etc. require a significant hardware setup during training. The availability of this hardware, like TPUs & GPUs, determines the speed of innovation. 

Colleges and universities can partner with cloud providers like Google GCP, Microsoft Azure, and AWS to get access to high-end hardware for their research. 

8. Data is the new oil

Just like hardware, modern AI models require large amounts of data. Data can be coming from various sources. It can either be collected or built via crowdsourcing. It also has privacy and bias concerns. If unchecked, these concerns can bleed into the AI algorithms and applications.

The government should come up with a cohesive data policy. Data policy is a big topic in itself that needs the attention of Govt and intellectuals alike. Academia, Industry and the Government should collaborate in building and sharing robust datasets for their AI training needs.

9. Industry lead innovation

The academia-Industry collaboration will be crucial in having a constant influx of cutting edge technologies into the curriculum. Often private sector industries are stuck in a race to capture short term opportunities and lose sight of the big picture. In the west, industries turn to academia to keep an eye on the long term and explore the opportunity with minimal investment.

Alumni networks will play an important role in furthering the engagement between academia and industry. Apart from that, the Government should streamline the process of research funding and clarifying Intellectual property rules for such collaboration. Adding aspects of education to the Corporate Social Responsibility of the private sector will fuel the growth of AI and other technologies in academia.

10. Collaboration via Conferences and research papers

Exchange of ideas is crucial for the understanding and development of cutting edge technologies like AI. Top universities and scholars across the world share their findings via research papers and presentations at conferences. Indian academia should get plugged into these highways of information flow.

Teachers and students alike should be encouraged to participate and present at leading AI conferences like NeurIPS, ICML, ICLR & AAAI. These conferences should be invited to host in India, and similar local conferences should be created. Travel cost to such conferences is usually the biggest deterrent and should be addressed via grant or alumni collaboration. 

Conclusion

Just like Iron, steam engines & computers ushered in an era of growth during various stages in history, Artificial Intelligence will shape the future of the world. Equipping students, teachers, and professionals with the right knowledge and resources about AI will be paramount for the future of India.

The foundations of the Indian Education system are robust. The New Education Policy 2020 and future policies will provide guidelines for growth conscious of the era. These policy documents might not get everything right, but as policymakers and good citizens, we should put our best foot forward. 

The only constant which does not change is CHANGE itself.