At its first AI Day conference in India on Tuesday, Intel outlined its vision on artificial intelligence (AI) in India, and announced a developer education program targeted at educating 15,000 scientists, developers, analysts, and engineers on key AI technologies.
“As India undergoes rapid digital transformation, the data centre and the intelligence behind the data collected will enable the government and industry to make effective decisions based on algorithms," said Prakash Mallya, Managing Director, Intel South Asia.
"This means increasing opportunities for using AI in the country," he added. "Our collaboration with the industry and the academia will help democratise AI, by reducing entry barriers for developers, data scientists and students. In India, we are targeting the BFSI, telecom, and e-commerce sectors, across High Performance Computing (HPC), big data, and Internet of Things, all of which are complementary to AI."
Speaking at the AI Day event, Professor Pushpak Bhattacharya, Director, IIT Patna talked about the work that the IITs have been doing in this space. "Research and academia are great platforms to initiate AI into the society," he said. "IIT Patna has been doing cutting edge research and development in Artificial Intelligence, distributed computing, network security, social networks, and beyond, using data driven machine learning, as well as knowledge and deep learning based methods."
India, said Pradeep Dubey - Intel Fellow and Director Of Parallel Computing Lab, Intel Labs - is very well suited to working on problems around AI because it now has the connectivity and also the scale required to generate a large amount of data, which is essential for technologies like machine learning. "You can't do this kind of work in Sri Lanka, for example, because the amount of data being generated is not enough," he said. "India and China are very well suited to take advantage of the new developments in AI."
General intelligence, AI agents, and AI applications
Part of the problem today is that there is a fair amount of confusion about what the word AI itself means. Kicking off the day, Mallya explained that what we are mostly talking about in terms of AI these days is the idea of Machine Learning - self learning algorithms that apply rule based logic to be able to make better and more accurate decisions without human intervention.
This is an incredible development, and lies at the heart of technologies like machine vision (image recognition), speech recognition, and natural language processing. These are powerful technologies and we can see their effect in almost all the technology we use today - whether it's talking to an assistant like Samsung's Bixby, the Uber app on your phone, or even the predictions from your phone's keyboard.
But it's a far cry from a general intelligence, which people possess. Yet that's what we think of when we talk about artificial intelligence, and what science fiction always presents us with. Perhaps that's why AI makes us think of the likes of Siri and Cortana instead of the recommendation engine on Myntra.
Dubey says that's not happening just yet, but adds that the ways in which AI can be used today are potentially already amazing. "AI is becoming pervasive," said Dubey, adding, "and it's taking computing to the realm of social problems, faster than anyone expected."
"By 2020, just three years from now, it is estimated that the AI compute cycles will grow by 12x at datacenters," Dubey aded. "And with connectivity getting better, this will allow real time decisions, which is very important when you're making something like a connected car."
To meet this need, Intel has a roadmap of specialised chips that are designed for this purpose, explained Nidhi Chappell, Director, Machine Learning, Data Center Group - Intel.
"No other workload will be as big in the data centre by 2020," said Chappell. "We have a full lineup of processors, and we have the Xeon Phi which is developed specifically for machine learning."
"What it does is brings the data closer to the CPU, and allows the data to be analysed much more quickly, so that the researchers can develop and test their models without as much delay," she added. "We are going to see a 100x reduction in time required for training models by 2020."
AI powered governance
Mallya also pointed out that AI has a role not just in new products, but in every aspect of life. There are already projects in India that work to bring Machine Learning to hospitals, and IoT based systems to help fishermen avoid poor weather conditions.
Describing AI as the next evolution of computing, Mallya said that it will have an impact on the government, health, IT, and all sectors. "In India, there is one doctor for every 1,680 patients. Imagine if we could reliably build diagnostics capacity in machines," said Mallya.
"Data analytics and machine learning can automate processes across e-Governance, research and the private sector," added D N Narasimha Raju, Chief Executive Officer, National Institute for Smart Government. "It is recognised that the potential of Artificial Intelligence is high in sectors such as healthcare, disaster mitigation, and financial services. Or you could use AI for finding tax evaders by finding patterns that would otherwise go unnoticed."