AI Adoption: Some Key Learnings

Keshav R. Murugesh,

Group CEO, WNS Global Services; Vice Chairman, NASSCOM

In recent months, there have been some significant shifts in conversations around artificial intelligence (AI) and how it can impact business. We are slowly going past the exploratory stage and getting down to the brass tacks of deploying AI to alleviate real-life business problems. Business leaders are keen to understand how they can make strategic investments in AI, rather than selectively pick processes for AI ingestion without a long-term vision.

AI guru Andrew Ng believes AI has the potential to impact human lives in a way that electricity had done in the late 19th and early 20th century. And quite like its predecessor, this current market disruptor has led to both fear and enthusiasm in equal measure. But now that the initial hype has cooled off, we have begun to unravel the true potential of AI to drive business value, along with its limitations and adoption-related challenges.

A McKinsey Global Institute report published earlier this year expects around 70 percent of companies around the world to adopt some form of AI by 2030.[i] These companies will do well by learning from the experience of those who have already embarked on AI.

  1. Fix the Talent Shortfall

Chinese technology giant Tencent estimates that there are only 300,000 AI engineers around the world, whereas the demand is for millions.[ii] India is at the 10th place in terms of countries with the most number of AI researchers, and unless we fix the shortfall, this will be the biggest barrier to AI adoption in the country. Besides government push through the AI Task Force and industry-wide efforts such as the NASSCOM Centre of Excellence in Data Science and Artificial Intelligence, companies must draw up their own AI talent development programmes and build AI labs to meet the rising demand.

  1. Understand the Limitations of Data

One of the biggest questions that faced us in 2018 is whether AI can be deployed to effectively tackle fake news. It has also led us to test how far we can rely on data science, without which AI cannot produce accurate results. To detect fake news, algorithms need to be trained to detect dubious content but that is tricky since falsehood may come in various shades of biased, subjective views. Similarly, when we deploy an AI tool in a business that is constantly evolving, the effectiveness may be limited as the new business data may be unstructured and not fit into the defined data sets.

  1. Technology Capability vs Industry Readiness

Within AI and machine learning, maturity is growing across fields such as named entity recognition, natural language processing, computer vision, and image recognition and tagging. This year we saw both high-profile deployments such as the police in China and Singapore using facial recognition for better law enforcement and enterprise use cases from sectors such as finance, healthcare and transportation. So, we have the technology capability, but is my business or industry ready for it? Organisations and industries at the lower end of the digital spectrum need to improve their data-readiness before they can expect success in AI deployment.

  1. Re-imagine Business Value

Measuring the business value of AI may not be as straightforward as measuring the impact of automation such as direct productivity or revenue gain. The metrics to use may sometimes be indirect – improving decision-making, forecasting demand more accurately or predicting failure and hence saving on downtime costs. Executives need to think innovatively to capture the true ROI of AI adoption. This is also a good time to share knowledge about how a business unit or function has successfully measured the value of AI so that others can learn from it.

In conclusion, I’d urge business leaders to develop a strong understanding of the capabilities and current limitations of AI. AI is not an instant revenue multiplier and does not add to a company’s core competence. AI is a tool that can provide you an edge in areas such as decision-making, customer experience, operations and inventory management. For optimal results, adopt a strategic approach where you assess your organisation’s readiness before you set your AI goals.




Share This Post

Leave a Reply