India has recorded the highest increase in AI adoption amidst the pandemic as compared to major economies including the US, UK, and Japan.
Artificial Intelligence (AI), while repeatedly touted as the ultimate business disruptor, has always grappled with low adoption levels in the real world. Entrepreneurs and CXOs in the Indian business landscape constantly quote the low level of enterprise digitisation and the resulting lack of quality datasets as a major reason for the adoption gap. However, the ongoing pandemic has accelerated digital uptake, which will help bridge this potential-realisation gap.
Over the next decade, as more businesses embrace digital transformation and generate considerable amounts of data from various touchpoints, AI strategies will begin to see rapid execution. In fact, India has recorded the highest increase in AI adoption amidst the pandemic as compared to major economies including the US, UK, and Japan.
With robust AI frameworks, businesses can identify game-changing patterns, optimise processes, provide contextual insights, and maintain a competitive advantage.
State of enterprise AI in India today
Currently in India, the most common real-world enterprise AI applications with proven benefits are limited to the auxiliary functions (sales, marketing, and customer support) of a business. A number of digitally mature Indian companies have completely automated their L1 support with well-trained AI platforms, such as HDFC Bank’s chatbot EVA, which has handled millions of customer queries ever since its launch in 2017.
At the same time, it’s highly unlikely that a bank will let an AI system single-handedly manage its core operations such as loan processing which involves creditworthiness assessment, risk rating, etc. Such business-critical use cases are still in theory stages or pilot projects, yet hold transformative potential if successfully implemented.
Implementing AI for your business
If you are a business owner or CXO looking to build an AI strategy with optimum value addition, here are three important things to keep in mind when you begin evaluating ideas:
1) Define your business’ unique set of AI use cases: Because AI is a general-purpose technology, its applications can be practically limitless in any environment and also tempting to experiment. To avoid getting lost in an ocean of possibilities, it’s critical to be aware of the digital areas where AI can truly bring value to your industry/business and then work your way forward with a set of specific use cases to maximise results.
A common mistake that most organisations make at this stage is using complex deep learning algorithms for every identified problem without checking if their information systems are capable of generating the required amounts of data. Instead, try starting out with simple business problems such as automating redundant tasks using RPA (robotic process automation) bots in areas where data collection is adequate, simple, and doesn’t involve personal information.
2) Invest time in training and validating your AI model with constructive feedback loops: AI algorithms learn faster and produce better, relevant outcomes when put through continuous feedback mechanisms such as human-in-the-loop (HITL) machine learning concepts where a human expert steps in and rates the AI’s decision accuracy or even guides the machine in choosing the best outcome prediction. HITL AI models are highly recommended best practices during testing/model training stages.
Another way to effectively build and train your model is through real-time shadowing techniques where both humans and AI manage a certain set of business processes in parallel until the model learns to mimic human decisions.
3) Keep privacy and ethical responsibility in mind: While AI is a potential game-changer, it could hurt user privacy at the same time. AI erodes all data boundaries, which allows it to make powerful predictions while simultaneously being very intrusive to user privacy. It’s crucial to implement data anonymization tools to strip your datasets of PII before feeding them to your AI model. Also, remember to update both your internal and external privacy policies on a regular basis. On the flip side, the fight against privacy is gaining traction, and government regulations are becoming more strict on enhancing data privacy. Currently, the AI research community is investing in techniques such as differential privacy and confidential computing to enable privacy-first AI techniques.
Going forward, provided that Indian businesses address the gaping talent shortage through upskilling initiatives along with government support and invest smartly in AI projects that deliver maximum impact, we can expect the effects to be revolutionary for most organisations.
