Why implementing AI projects can run into rough weather in India

Artificial Intelligence (AI) can be a black box. Users can't really figure out what drives the algorithm - what data has the machine been fed with, how it is nurtured and why machines make the decisions they do. Can this black box image impact AI adoption?  

Advisory firm PwC's study, 'With AI's great power comes great responsibility', comes up with interesting answers. The survey, which polled 1000 CXOs globally and in India, was released on the sidelines of the 33rd India Economic Summit in Delhi on Thursday. The summit is hosted by the World Economic Forum in collaboration with the Confederation of Indian Industry.  

The study wanted to figure out  the scale of deployment of AI solutions, how they are working, and whether there are any ethical or legal challenges. Turns out, India is not far behind global standards when it comes to adoption. About 62 per cent of Indian respondents surveyed stated that they have implemented AI in some form versus 65 per cent globally. "However, the worrying part is that Indian respondents (53 per cent) significantly outnumber their global counterparts (36 per cent) in admitting that they have no formal approach to identify AI risks," the report stated. About 29 per cent had no tools to assess security flaws in their AI systems versus 16 per cent globally. "This suggests that the enthusiasm to implement AI projects is very likely to run into headwinds unless organisations adopt a robust framework for using AI responsibly," the report added.  

What inhibits AI adoption in India? The top five reasons are vastly different from what global counterparts of Indian CXOs would tell you. In India, 'lack of understanding of how AI models and applications make decisions' topped the inhibitor list, followed by 'insufficient/not usable data', the fact that it takes 'too long to demonstrate value', 'skills mismatch' and finally 'budget constraints'. Globally, budget constraints top the list followed by skills mismatch and insufficient or usable data.  

The Indian concerns largely have to do with lack of interpretability in AI-powered systems - the cause and effect of a decision that the algorithm makes. PwC said that better benefits of AI would be realised when an appropriate governance framework is in place, and when humans and machines can collaborate effectively.   

This is not to say that there are no benefits at the moment. Indian respondents to the survey pointed to efficiency gains as the top priority for AI investments, followed by innovation and revenue enhancement. "Surprisingly, when compared to global respondents, a significantly greater number of respondents from India cited market-leading innovation as their primary driver to invest in AI. This could be a sign that organisations in India are investing in AI to create a competitive advantage rather than just using it for revenue enhancement or efficiency improvement," the study stated.

Also Read: Single-screen cinemas are shutting down, but there are still not enough multiplexes

Also Read: IRCTC IPO subscribed nearly 5 times on last day, receives bids for over 9 crore shares

Also Read: Action back in the SME IPO mart, but second line of investors missing