Artificial Intelligence may be around everything and in everything conceivable today. While its application is certainly scaling industries, the technology is simultaneously paring back layers and delving deeper every day.

At Xperience AI Virtual Summit, Srikanth Velamakanni, Co-Founder, Group Chief Executive & Executive Vice-Chairman, Fractal Analytics spoke about the evolution of applied AI, the various schools of thought thriving within the AI landscape and the critical aspects needed to enhance adoption and nurture this technology in the right direction.

In 2006, the Dartmouth AI Conference: The Next Fifty Years or AI@50 was held at Dartmouth College and organized by Professor James H Moor. The objectives of this conference were three-fold - to celebrate the Dartmouth Summer Research Project, which occurred in 1956; to assess how far AI has progressed; and to project where AI is going or should be going. Leading researchers and scientists made presentations on the progress AI had made in 50 years, in areas like learning, search, networks, robotics, vision, reasoning, language, cognition and playing games. Over time, we have seen incredibly successful developments in neural networks, General Adversarial Networks (GANs) and more. But there were some dramatic disagreements between the pedagogy assembled at the conference:

• Should AI be probabilistic or logic-oriented?
• Are neural network techniques good enough?
• Should AI attempt to understand human psychology or focus on pragmatic problem solving?

Given the rate at which AI is progressing and proliferating, Velamakanni makes a case for another landmark event now – it isn’t 50 years away but 15 years since AI@50 – a reflection of how far this technology has come in less than two decades. Aptly naming this juncture as AI@65, Velamakanni asks:

• How do we drive responsible AI to ensure no harm?
• Where will the next AI breakthrough come from?
• How can organisations drive AI results at scale? (even though Artificial General Intelligence is still decades away)

Google’s Peter Norvig makes a case for AI to be developed in a way that it services man, not render him irrelevant. The technology is purely to augment human tasks, especially those that require high speeds, high accuracy and need to be completed in limited time.

Driving results at scale with AI need algorithms that can match or exceed human performance, seamless engineering and interconnected data pipelines in addition to human-centric design. Once these factors are addressed, there is scope for limited errors in the technology and this will consequently ramp up the pace of adoption. Consider the initial apprehension towards Siri, the Apple smart assistant. In 2012, the percentage of errors made by the technology was 17% but by 2016, this number reduced to 3%. Consequently, Siri’s adoption has been 3X since January 2015. By building a richer sensor network, enhancing compute abilities and infrastructure, building larger learning models, errors in AI will reduce.

While one can vouch for data integrity, data mining, sound engineering & design and faster compute abilities, none of these come together without effective leadership. It is imperative for organisations to invest in talent, even if the short-term cost may be higher because in the long-term, this investment will reap rich dividends. For instance, hiring Professor Geoffrey Hinton proved to be a more than daunting task for Google. Hinton, along with other industry legends like Yann LeCun of New York University and Yoshua Bengio from the University of Montreal, are among the most sought-after names in the field of AI for their pioneering work in deep learning. Google somehow managed to hire Hinton, and he’s changing the AI landscape for the tech giant with his work on deep learning to improve voice recognition and image tagging. LeCun is a prized asset for Facebook, while Bengio continues to teach at University of Montreal, while assisting companies like Botler AI and Recursion Pharmaceuticals.

“The next 20 years are going to be fun for Artificial Intelligence,” said Velamkanni just as his session concluded. And he’s absolutely right – the AI ecosystem is at an exciting precipice with tremendous tech talent, infrastructure to run complex data models and myriad opportunities across industries to experiment on. AI applications will permeate every imaginable aspect of business and the AI applied landscape is only going to get richer.