Like steam engines, generators, computers, and the internet drove the previous industrial ages of steam, electricity, and information, artificial intelligence (A.I.) is the engine driving the intelligence age. A.I. already powers many real-world applications—from facial recognition to language translators and assistants like Siri and Alexa but, enterprises have barely begun to scratch its surface.
Evidently, the advent of deep learning in the consumer space powered A.I. advances initially. Online translation and photo-tagging services, or digital voice assistants on mobile phones, for example. These A.I.-powered product enhancements, while definitely appealing to consumers, didn’t necessarily lead to a corresponding uptick in sales. While the next wave of consumer A.I. will see even more service and product innovation, the true potential that is yet being unlocked, may lie in the enterprise space.
The real power of A.I. is in its ability to holistically transform the enterprise and redefine business in ways beyond our present frames of reference. A 2017 study by PwC calculated that global GDP will increase by 14% by 2030 because of AI adoption, contributing an additional $15.7 trillion to the global economy. Of this, $6.6 trillion is likely to come from increased productivity and $9.1 trillion is likely to come from consumption side effects.
Digital foundation
A.I. owes some of its promise to other technologies. For A.I. to flourish, it needs an existing base of core and advanced digital technologies like cloud, big data and advanced analytics. A McKinsey study found that 75% of companies that adopted AI, depended on knowledge gained from applying and mastering existing digital capabilities to do so. Data is the fuel that runs A.I. Any organisation in any industry can create business value from A.I., especially those with large amounts of data. In fact, having a digital foundation is necessary for enterprises to gain access to large amounts unique data, which can then be used to train A.I. algorithms.
Part of the reason why A.I., despite being known for the better part of seven decades, wasn’t able to live up to its hype, was the lack of data. That has changed decisively over the past decade. Quantum leaps in processing power and storage, translated to a corresponding jump in data generation. IDC calculated that in 2010 the world created about two zettabytes (ZB) of digital information. Cut to 2020, and the entire digital universe is expected to produce 44 zettabytes. That is 40 times more bytes than there are stars in the observable universe!
In a world awash with data, the enterprise is becoming its data steward. Consumers’ reliance on cloud services especially from connectivity, performance, and convenience perspectives, continues to rise, while the need to store and manage data locally decreases. At the same time, spurred by tightening data regimes, businesses are looking to centralize data management and delivery to control their businesses and user experiences. IDC estimates that from 2019, more data will be stored in the enterprise core than in all the world's existing endpoints.
Opportunity amidst challenges
While enterprises were already along the path to embracing the transformative potential of A.I. and its underlying digital technologies, came the great accelerator—Covid-19. The pandemic emphasis on low or no contact, translated to the volume of online delivery increasing by the same amount in eight weeks, as it had over the previous decade. Or, telemedicine experiencing a tenfold-growth in subscribers in just 15 days. Similar acceleration patterns were seen across industries, as the black swan event changed all aspects of life and customer behaviour, prioritising business continuity. This prompted enterprises to turn to digital technologies such as A.I., to enhance customer experiences on digital channels in the new low/no-contact economy.
Another impact of the pandemic, is the debilitating negative impact it has had on businesses—from size trimming to even closure. Even prior to 2020, with the slowdown in the global economy, there was an increasing emphasis on AI-powered automation for efficiency gains. Enterprises are now casting their nets wider and deeper to identify process and use cases where the magic of A.I. can help overcome the economic shock of the pandemic. From remote patient care, to slashing administrative workloads for doctors in healthcare, to helping governments track and manage the contagion’s spread, to aiding HR departments in keeping their fingers on the pulse of employees’ mental health, use cases abound. And not just in healthcare.
Part of the reason why A.I., despite being known for the better part of seven decades, wasn’t able to live up to its hype, was the lack of data. That has changed decisively over the past decade. Quantum leaps in processing power and storage, translated to a corresponding jump in data generation.
An idea whose time has come
As enterprises grapple with the various economic, social, and technological forces that shape the market amidst the pandemic, expect them to throw the might of their resources to tackle emerging problems. For example, virtual experiences have gained much prominence over the past few months. Consequently, this will catalyse the maturing of A.I. technologies that interpret and respond to non-verbal emotional cues such as facial expressions, gestures, body language, and tone of voice. In parallel, newer approaches like transfer learning, few-shot learning, and multi-task learning will see broader adoption among application developers, promoting training A.I. from smaller data sets.
One of the risks that A.I. carried was the human inability to understand decision making from large and complex models. This challenge of explainability, or showing what factors led to a decision or prediction, and how, takes on particular weight, where trust matters and has implications. For example, in healthcare, criminal justice or financial lending. Even in industries that aren’t as heavily regulated, recognition of risks from lack of explainability, is rising. Some nascent approaches, including local interpretable model-agnostic explanations (LIME), aim to increase model transparency and should help drive further adoption of A.I. technologies in hitherto out-of-bounds areas. There is no doubt that this will develop the broader recognition of the need and consensus for the oversight of bias and ethics in A.I.
While it is still early days in enterprise AI adoption, the confluence of technological, economic and social factors, along with the emergence of the pandemic, it is now at an inflection point. Over the past few years, AI has emerged as the top technology innovation-driver, finally starting to deliver on its promise and securing its future.
Views are personal. The author is Senior VP, head, A.I. & Automation Services, Infosys.
