Home Stories

Artificial Intelligence: Is India ready for an AI revolution?

In 2018, India produced 10 times more patents than it did in 2012. However, it still ranks 10th globally among AI patent producing countries. India’s AI maturity stands at 2.45 out of 4, suggesting that we have a long way to go to become a global AI leader.

By Dr Srini Janarthanam
New Update


Have you been hearing about Artificial Intelligence a lot lately? In movies, TV series and even children’s cartoons? I am sure you have. AI has become a buzzword and not for some superficial reasons. Artificial Intelligence technologies have come out of research labs and are now out in the world, permeating every sphere of human activity. Not only are we becoming consumers of AI services but also its creators.

AI is creating numerous job opportunities for us in several domains and diversified business functions. It is important to see AI beyond its hype. In this article, let’s demystify Artificial Intelligence, look at what AI is, its impact on us, the types of jobs it has created and finally, the Government of India’s initiatives to get us on the AI bandwagon.

Imagine you are out for a jog. You want to hear some music. You open Spotify, and it starts suggesting a list of songs that you will like. You find it interesting that it knows you well and gets the songs you enjoy. How does Spotify do that? This is an example of AI at work. Surely, you have heard of smart assistants like Siri, Alexa and Google Assistant. How do they make conversation with you in natural human languages? This again is AI at work. Finally, you have probably heard of cars driving themselves with no human drivers. How do they do it? Again, this is also the work of AI.

So, what is AI?

Artificial Intelligence (AI) is the technology that enables us to build computing systems that are capable of human-level intelligent behaviour. For ages now, we have been developing software applications to do our bidding. But some processes cannot be fully automated using traditional software.

What if, instead of the traditional visual interface with buttons, text boxes, and drop-down menus, you want to interact with the system using your voice? What if you want a system to visually take in information through onboard cameras and make decisions with it? What if you have a number of unstructured tasks that need a multifunctional humanoid robot that quickly adapts to tasks, perceives the environment and acts accordingly? These tasks are beyond traditional software development and require AI solutions.

“We cannot solve our problems with the same thinking we used when we created them.”
Albert Einstein

Types of AI problems

AI technology has evolved over time to solve challenging problems. These problems can be largely categorised into four types based on the type of input data and decisions made using them.

1. Language - These problems are about understanding and generating human languages in all forms - text, speech and images and making decisions with them. These are collectively called Natural Language Processing (NLP) problems.

2. Visual - These are problems relating to understanding, generating and decisioning based on visual data in all forms - static images and videos. These are collectively called Computer Vision (CV) problems.

3. Numeric - These are problems where the system needs to crunch lots of numeric data, find patterns, make predictions and recommendations and generate explanations. These data include numbers like product sales, departmental expenses, sensor data, website clicks, and social media likes, among others.

4. Agents - Agents typically combine data domains together in physical and virtual settings where AI agents perceive, understand, infer and act to achieve goals they are set to achieve. These include problems that are solved by warehouse robots, home assistants, self-driving cars, AI that plays chess, Go and Atari games, non-player characters in games/metaverse and others.

AI toolbox

So, how do we solve problems in AI? Machine learning (ML) is a popular approach to solving AI problems but not the only one. There are three kinds of ML algorithms (an algorithm is a set of instructions that you give a computer to do a specific job).

  • Supervised learning - We can learn from data that is annotated by human supervisors. AI can classify images or text based on examples and labels provided. For instance, by training itself on hundreds of images labelled with their names (e.g. cat, dog, monkey), it can predict the name of new unseen yet similar images.
  • Unsupervised learning - We can explore data without any annotations from humans. Data can be clustered based on similarity to each other. Clusters so identified can help analyse new data and derive insights.
  • Reinforcement learning - RL techniques are used to learn sequences of steps or actions to take to go from the current state to the desired goal state. These techniques are typically used in gameplay where the final desired state is that of the player winning the game.

Deep learning, which is currently a very popular approach to AI, is an advanced version of machine learning. In addition to ML, there are other tools and techniques to solve AI problems. These include heuristics, search algorithms, logical reasoning, statistical reasoning and others. Using these tools and techniques, you can today create the magic of human-level intelligence in very specific domain tasks.

“Any sufficiently advanced technology is indistinguishable from magic.”
Issac Asimov

Why is AI important?

The use of AI technology has risen exponentially over the last few decades. Recent development in AI techniques like Deep Learning has propelled the use of AI in several sectors. Transformer architecture has created a number of notable AI models - BERT, T5, GPT-3, CLIP, DALL-E, and GATO, amongst others. This technology can be used to understand language, generate articles, compose music, generate art, translate between languages, have open-domain conversations, beat humans in traditional games like Go, predict protein structures and even drive cars. These state-of-the-art techniques have moved the bar on AI research, paving the way for more interesting solutions to come.

Investments in AI are soaring at unprecedented rates. AI solutions are being used across all sectors – travel, healthcare, e-commerce, education, banking, and governance, to name a few. This has resulted in a number of AI startups being born to take advantage of the global demand. AI provides startups with the tools and techniques required to scale up massively in response to global demands on the innovative products and services they are building. Product manufacturing and service delivery which were limited by the availability of qualified human resources, are now delivered by trained AI solutions instead. This enables startups to deliver services using a small team of people augmented with AI capabilities. Enterprises are not behind. They use AI to be available 24x7 to their customers across the globe and cut costs.

International Data Corporation (IDC) reports that the global AI market is valued at $87 billion currently and is estimated to grow to $500 billion by 2024. It is then estimated to grow to $1.6K billion by 2030 as per Precedence Research. In India, the market is valued at $7.8 billion as of Aug 2021, according to research by Analytics India Magazine. In terms of AI related patents, India’s stance has boomed since 2012 and in 2018, India has produced 10 times more patents than it did in 2012. However, it still ranks 10th globally among AI patent-producing countries, according to a recent report by the Centre for Security and Emerging Technology. In a recently released report, NASSCOM’s AI adoption index 2022 pegs India’s AI maturity at 2.45 out of 4. All these suggest that India still has a long way to go to become a global leader in AI.

Jobs in AI

Adoption of AI in diversified sectors has created new job opportunities. The median salary for AI roles in India is INR 14.3 Lakhs with close to 1 Lakh professionals working in the sector as per Analytics India Magazine. With soaring investments, these numbers are only going to grow in the years to come.

AI is adding a number of jobs to the traditional roles we find in software development. Here are the most common roles in AI.

Research Scientist

Research Scientist roles exist at universities and AI research labs across the globe. Although AI is being industrialised, there is still a long way to go and many research problems to solve. Big tech companies like Google, Apple, Microsoft, Facebook, etc do have well-funded labs that focus on a number of AI problems.

Data Scientist

Data Scientist is one of the most popular jobs when it comes to AI in industry. A Data Scientist analyses data, explores algorithms, builds machine learning models, experiments with them to identify the best fitting model for a given problem.

Data Analyst

Data Analysts analyse data and get answers. They find trends and patterns. They make reports and build dashboards.

Data Engineer

Data Engineers collect, store and make data available to use by analysts and data scientists. They are responsible for designing, building and running data pipeline infrastructure.

MLOps Engineer

MLOps Engineers deploy the AI systems and keep them running. They set up and monitor the systems to ensure that the models don’t drift in performance and that the system as a whole is resilient and robust.

Besides these generic AI and data roles, there are roles specific to the problem domains, such as Conversational AI, NLP, Computer Vision and many others.

Conversational AI / Vision / Speech / NLP Engineer

These AI Engineers create the infrastructure using tools and platforms specific to the problem. For instance, a Conversational AI engineer will create chatbots using platforms like IBM Watson, Google Dialogflow and others and integrate them into frontend chat user interface (UI), ML models built by data scientists and backend services to consume data or content required by the user.

AI Game Programmer

AI is used a lot in computer games. AI Game programmers are responsible for designing and building the AI elements of a game. They design and develop non-playable characters (NPCs) that are humans and other entities in the game that behave intelligently but are not played by actual humans. Besides NPCs, they build the elements that enhance the game-playing experience.

AI is for everyone and not just techies. Good AI solutions emerge from collaborative efforts between data, tech, business and creative folks.

AI Product Manager

AI Product Managers are PMs that understand how AI solutions are different from traditional software solutions. They understand the need for data, market testing MVPs and skillfully combine AI tech with the UX to be delivered.

AI Product Designer

AI Product Designers are product designers specialising in AI solutions. They understand user needs and constraints and design AI-powered experiences that solve user pain points. They ideate solutions, create demos and blueprints and build prototypes. They work with engineers and data scientists and help them understand the requirements of the solutions being built.

Conversation Designer

Conversation Designer designs conversations that chatbots have with their users. This is currently low on AI tech and high on the UX side. This is because conversational AI solutions use preplanned conversational pathways while engaging a user, and they do not make conversational decisions based on what they learn from data. However, I believe that in the next few years, this role will become a blend of UX and AI, wherein the conversational flow will be a hybrid of preplanned scripts and dynamic planning and learning.

In addition to these specific roles, it is now also becoming essential for everyone to become AI literate. This means that AI is going to impact each one of our jobs. It also means that we need to learn what AI means, how it can help us in our jobs, and learn how to work with it.

How do you get started?

Wondering how to get started in AI? There are many ways to go about this question, and you can get started from three vantage points.

What type of AI problem (linguistic, vision, or numeric) excites you the most?
What roles in AI are you cut out to do?

  1. How can AI solve the hard problems in your domain (e.g. travel, legal, healthcare, etc.)?
  2. What type of AI problem (linguistic, vision, or numeric) excites you the most?
  3. What roles in AI are you cut out to do?

Answering these questions may need some exploration, but once done, acquire the skills needed to understand the problem domain and the data available and perform the tasks for which the role is responsible.

Where do you get the skills? There are many options available. The traditional approach is to get a degree in AI at a reputed university. But if you are upskilling yourself to get into AI, you may want to look into online learning platforms like Coursera, Pluralsight, Udacity, etc., that allow you to get certified in specific AI skills.

The Govt. of India Ministry of Electronics & Information Technology (MeitY) and NASSCOM have created a platform called FutureSkills Prime in 2018 to achieve this very objective – to upskill you in areas of new and emerging technologies. The platform had trained over 3.8 lakh people until 2021 as per Hindu Business Line.

The aim of FutureSkills Prime is to upskill 1.4 million people in future technologies over the period of 5 years. In order to make this viable, the government is supporting learners by providing them incentives up to INR 14500 if they are taking courses in AI, Big Data, Cloud Computing, Cybersecurity, Robotic Process Automation and others.

FutureSkills platform brings together a number of course providers, including well-known industry names like Accenture, Microsoft, CDAC, Cisco Networking Academy, Digital Vidya, SkillUp Online, Jigsaw Academy, and many more. Learners have a number of options in terms of levels of courses, starting from foundation to bridge courses all the way to deep skilling courses that include capstone projects.

AI is a fast-growing field, touching on many facets of our lives. AI is used everywhere, from choosing music tracks and things we buy to whether we get through in our job and loan applications. While Covid-19 researchers, scientists and virologists worked for several months to build testing capacities to detect the virus and manufacture vaccines, a recent report suggested that researchers have manufactured AI powered wearable health activity trackers that might be able to detect Covid-19 even before the onset of the symptoms. From the ability to spot burnout in employees, to detecting cancer, from mitigating climate change to revealing the secrets of the universe, AI has become an integral part of our world.

According to The Brookings Institution, India is one among the top ten countries in the world in terms of technological advancements and funding in AI. A latest report by Nasscom states that the adoption of AI and data utilisation strategies can help India add $500 billion to its GDP by 2025. Though India needs to go a long way when it comes to AI, the government of India has launched ambitious programs and campaigns like “AI for India” with an aim to make India the “AI Capital of the World”. While that may be a really long road, India is slowly but steadily making strides in harnessing its AI potential.


Srini Janarthanam is a PhD in Artificial Intelligence & Natural Language Processing from the University of Edinburgh. He served as a Postdoctoral researcher at Heriot-Watt University, UK, working on several EU-funded projects that focus on Conversational AI research. He authored a book titled ‘Hands On Chatbots and Conversational UI development’ published in 2017 when chatbots and conversational assistants became a rage.