While Artificial Intelligence (AI) has caught our fancy today, humans have always been intrigued by the possibility of creating machines that mimic the human brain. The term “artificial intelligence” was coined in 1955 by John McCarthy. This beginning led to the creation of machine learning, deep learning, predictive analytics and now to prescriptive analytics. The capability to process large amounts of data sets at a high rate has today enabled AI to offer actual services. It has also given rise to a whole new field of study, data science.
Building AI through the creation and application of algorithms built into a dynamic computing environment requires computational systems, data and data management and advanced AI algorithms.
In India, a government-appointed task force has come up with a plan with recommendations to boost the AI sector in India, from developing AI technologies and infrastructure, to data usage and research. Emphasising its role, the report clearly states: Data is the bedrock of AI systems and reliability of AI systems depend primarily on the quality and quantity of the data.
In a populous country like India, mass impact sectors like healthcare, national defence, agriculture, environment and water can benefit from AI applications. There are India based startups that utilise AI today to offer diagnoses based on pictures of patient blood sample. The question is how we employ, and even before that how we identify data that will get us the desired outcomes on applications and analytics.
AI (its logical evolution of machine learning) and deep learning are the future of business decision making. But all this is hinged on one prerequisite, which is data. This data allows machines to observe repetitive patterns and learn to build predictability. The key aspect to this learning is to have humungous data points that allow the learning over time.
Organisations and governments today are eager to take advantage of AI technologies as a means to introduce new services and enhance insights from data. However, as data science teams begin to operationalise deep learning, many are experiencing issues with data management. This brings to focus the creation of an appropriate environment and infrastructure required for data to deliver on AI outcomes.
Employable data is a clear pre-requisite for India to catch up with countries like China. India’s GDP could reach $6 trillion in 2027 because of its digitisation drive, according to a forecast by Morgan Stanley. That would make India the third-largest economy in the world. Yet it remains far behind China which aims to become the world leader in AI by 2030. How quickly we can re-skill and reshape our data infrastructure will decide how much of catching up we can do.
While the digitisation process from a citizen perspective is still on, the ability to take decision making to remote areas in the form of sectors such as education and agriculture will really drive the inclusive growth of our country.
To create such an environment, where AI in India moves beyond the proof of concept phase, we can look at a four-way outlook on data.
Know where your data is located: This is really the ability to have visibility and insight into your data. The data could be in a data centre, a cloud or a hybrid model, but this is the first step in figuring out what we want to do with it.
Know what you are doing with data: Get the data in the quality that you need to derive value and potential decisions. Get your data to a form where you can analyse it. This means the right format, frequency and time period the data is sampled in. To understand this, consider the potential of life-saving information from such data. Say, a percentage of people show the symptoms of a disease. Extrapolated over large enough data sets, the symptoms can be generalised enough to diagnose diseases early and save lives.
What algorithms or techniques you want to apply on data: Now it is about analysis and decision making. Data-based services could be about new businesses, customers, new revenue streams, or new solutions. What kind of algorithms you really want to use depends on what ends you want to achieve. Peer benchmarking or community wisdom-based anomaly detection can help find the blips in data that point to issues or potential revenue streams. For example, environmental impact of a certain pollutant can be derived in such a manner.
Continuing to innovate based on the predictions derived: Change based on recommendations from data is the final step, but also one that needs continuous learning. For example, malnutrition data of all genders and age groups versus specialised subsets of this data may provide different insights to make decisions in healthcare.
While AI in India is now visible in how we book taxis, order food, pay bills etc., this is more of the consumer and aggregator services use case. To make AI a reality in the big impact areas, the b2b ecosystem will need to innovate beyond just revenue streams. That’s what will turn our villages into smart cities. We will need to roll up our sleeves and invest in building data-led AI infrastructure and get on the global innovation expressway.
What is AI?
Does AI aim to put the human mind into the computer?
How far is AI from reaching human-level intelligence?
What should I study before or while learning AI?
(Source: Dr McCarthy on AI @ http://jmc.stanford.edu)
Evolution of AI
Reactive
Limited memory
THE understanding
Fully Self aware
Status in India
The task force on AI for India’s economic transformation has recently submitted its recommendations to the government. It has identified 10 important domains relevant to India that includes manufacturing, fintech, healthcare, agriculture, education, national security and public utilities
The potential
Major challenges
—The writer is Managing Director, NetApp India
4
7