By Soumik Dey

The first algorithm for Artificial Intelligence (AI) was written in the 1960s. But the phenomenon had not scaled in the absence of enough computing power. But now, with advances in computing storage and processing, in less than a decade artificial intelligence has leapfrogged. The machines are taking over fast. They seem to be running our lives and even getting better than us at it.

For businesses though, this has created a do-or-die situation. Gone are the days when firms could choose to ignore the power of AI or stay on the sidelines with a wait and watch mindset. Nine years after SIRI was first introduced, AI is taking over business functions and leaving behind telling impact for the business value chain and society as a whole.

To understand this shift, the Srini Raju Centre for Technology and Network Economy (SRITNE) at ISB, Hyderabad, laid special emphasis on AI during its recently concluded Digital Transformation Workshop. The workshop discussed and dissected the impact of AI on the future of work, products and services, and a host of other issues related to this emerging technology.

A peek into the future of this AI-led transformation of the economy came from ISB Professor Anand Nandkumar’s presentation, based on his ongoing research on AI, and its impact on jobs and industry dynamics. He defined AI as a General Purpose Technology (GPT), akin to that of innovations like electricity and computer.

“This has a pervasive effect on several industries. There is a paper which shows that deep learning is an invention for a method of invention,” he said.

There is a general believe that automation would only impact manual tasks and not the cognitive ones. This, however, is going to be different for AI. “We study something cognitive so that we cannot be replaced. But that holy grail is going to be broken. It is the promise of AI,” he said. According to him, shortly this emerging technology can take over not just manual and routine jobs but even cognitive and non-routine ones as well.

Notching up the discussion, Professor Nandkumar pointed out, that the advent of Information Technology in the US created a paradox. Jobs were replaced, productivity increased and wages dropped. However, he contested the doomsday predictions, that this is what is going to happen with AI too.

In a study being conducted in India by Nandkumar and his researchers, similar to a 2018 US study titled ‘What Can Machines Learn and What Does It Mean for Occupations and the Economy?’ by Erik Brynjolfsson, Tom Mitchell, and Daniel Rock, the impact of AI was found to be pervasive across all professions. In a measure of how susceptible a profession could be to the advent of AI, both the research studies concluded that AI was just as pervasisve as it was thought to be and that its pervasiveness would impact cognitive, manual, routine and non-routine jobs.

“So pervasiveness of AI is the message. We find that evidence,” he said. However, interestingly, his study also found that this pervasiveness of AI would have low correlation with wages. Also, that AI would probably not spell the end of human labour.

He admitted that in the short run, post introduction of AI, there are going to be some job losses (0.18%-0.34% per 1,000 workers for every robot introduced) and drop in wages (0.25%-0.50% wage points) as indicated by some earlier US studies, but in the long run there is good news. This loss of jobs in the short run is going to be compensated by newer kind of jobs, like AI Trainers, automation ethicist, automation economist etc, being created in the long run, he said.

“If this is what is going to happen, AI Transformation may be the key business strategy to follow,” he concluded. He said the need of the hour for businesses was to engage in small experiments to adapt, learn and deploy them at a larger scale. While governments would need to set up the infrastructure and invest in AI readiness, individuals needed to develop a growth mindset and believe that intelligence is not just innate but also acquirable like AI skills.

“The message then would be to either prepare or perish,” Nandakumar said.

During further discussions through the workshop, newer dynamics about AI came to the fore from technology business leaders speaking at the workshop. This change was clear from the perspective of Big 4 management consulting firm Deloitte, industrial automation leaders Honeywell Connected Enterprise and software giant Microsoft.

Technology from just being enablers and remaining a cost point in business value chains have evolved greatly and transforming businesses across domains. “It is not just about cost reduction anymore. We are seeing technology as co-creators of business value rather than just enablers of it. That is a significant shift,” said Ajit Nema, MD, Deloitte (Office of the US), speaking at the workshop.

Interesting dynamics are also coming into play across traditional industries where AI, data-based decision making comes into play. “If an alarm comes on in a refinery, how do I within a milli or microsecond say okay this is critical and I need to act. Versus the 500,000 alarms that came, and which are just noise,” said Karthik Ganapathi, MD (India), Honeywell Connected Enterprise, explaining the profound implication of change in workflow across organisations brought in by AI.

Terming AI as an ‘exponential technology’, Dr Rohini Srivathsa, National Technology Officer & Strategy lead at Microsoft India, said that narrow focus Ai technology in the fields of image recognition, speech recognition and language understanding is doing the job very well. However, she adds, that one may need to also step back slightly and not get enamoured by technology.

Giving an example of a bot created to counsel teenagers, Dr Srivathsa highlighted on the ethical dilemmas of using AI for certain general-purpose application. “The question that we at Microsoft are thinking deeply about is that what can happen with AI. There is intended use and there are also unintended fallouts. We should be cognizant that technology is not perfect, knowing that data is not perfect. And the fact that in certain scenarios you do not want the machine to do something even if it could,” she said.

She said in Microsoft such thought groups, called ‘wallows’, exist and they think about what can go wrong with the use of any particular Ai technology they develop. “There is a group at a global level and gives its recommendations to Satya,” she said. She said in certain alarming situations the company had even said ‘no’ to business. In certain situations, they had even helped their customers think about these issues.

This framework, she says, comprising of issues related to fairness, security and privacy, safety and reliability, inclusion and transparency and accountability, have been integrated over the past three years in the workflow of Microsoft, in terms of standard-setting, engineering, product development and research.

Talking about the bias that creeps into data, Ajit Nema of Deloitte said that within his organisation there is a debate about using AI for recruitment. “The data that it (AI) is using has human biases already built-in. So, how do I eliminate that,” he asked, furthering the point about ethical issues related to the use of AI.

Summing the conversation, Professor Deepa Mani, executive Director, SRITNE said that there seem to be two large issues about the proliferation of AI. First, it is about the behaviour or comfort level of a person with this technology. Giving an example of using auto-pilots on flights, she highlighted that despite the initial inhibition of people about the technology, mindsets soon changed.

The second challenge, she said, was the ethical challenge about AI. “This is the issue about statistical discrimination versus discrimination,” she said giving an example of how Google was showing jail bond ads to users from the African-American community. “So, those are the kind of ethical challenges, which I guess may be different from the behavioural adoption of this technology. I think that policymakers, as well as organisations, will have to grapple with it,” Prof Mani said.

On the problem about lack of right kind of data in India to power growth of AI, Ajit Nema of Deloitte expressed that building models based on this flawed data sets will result in incorrect models. Rohini Srivathsa of Microsoft, however, expressed that data has started to come in the right way in India with Direct Benefit Transfer, data mining to detect tax fraud and others being implemented in India. “This is where there is a need to think about data strategy, and this becomes very timely, as those seeds are getting sown right now,” she commented on the need-of-the-hour for policymakers and businesses to deal with the advent of AI.

During the discussion, AI was seen to match up and even compete with human intelligence over the next 50 to 100 years. But in the interim, the experts suggested that the AI beast should be kept on a tight leash. “If you can automate processes and make them learn within a constrained domain, that’s where it is going to be of real benefit. We can focus on narrow AI solutions for the time being,” said Karthik Ganapathi of Honeywell Connected Enterprise.

Feedbacks to: soumik_dey@isb.edu