Source: The Times of India
Author: Vidhyashankar Sriram
Buzzwords like Big Data and AI have been in vogue for a few years now and 2023 looks no different. As we get richer in hindsight with larger deployments across industries, it is now evident more than ever that it is “Data before AI”. Firms that went ahead with glossy AI projects without making sure if their data foundations were firm are now stuck with failed efforts.
Data: Hybridizing the cloud approach
Data will continue to grow. The term ‘Big’ is not big anymore as the growth of data has overtaken even the most ambitious projections. Migration to cloud has been a secular trend over the past few years across firms. This was necessitated by nimbler costs and the ability to work with third-party products.
However, the expectations that hyperscale cloud providers would be cost effective and get progressively cheaper with scale have not played out. Senior industry leaders will soon begin to realize that cloud migration isn’t a one-way street. Well established data intensive processes gathering larger volumes by the day are smarter when processed in-house. This is where a hybrid model of making a transition to cloud yet retaining some of the in-house infrastructure will be a smarter strategy.
This is reflected in the way Big Data transformed earlier. When we started with data lakes and unstructured data, everybody went gaga! And then slowly realization hit that some of the classical data warehouse operations initially dismissed in a rush continued to play an important role. Essentially, what we are witnessing is a trend that has already played out. Yes, it’s great to have your data in the cloud with a library of pay-per-use functionalities and resources. But undifferentiated heavy lifting processed on the cloud results in inflated bills especially for larger firms. For start-ups and SMBs, being cloud native will continue to be a winner given the cost of data management, security & disaster management to go along with a vast library of managed services.
AI: Bionic Customer engagement
Many of the mainstream commonplace apps globally are based on content. Yet firms counter-intuitively continue to chase the functionality race. They are keen to answer the question, ‘Can the customer do all the possible transactions on our app?’ It seems straightforward from their perspective. They are looking at digitalization as channel migration of their physical distribution.
They are yet to invest time on the critical question, ‘Have we enabled the customer with the right information to make the transaction?’ In the world of AI, this is about bionic customer engagement. A set of complex rules & models that gives a near-human experience digitally. It’s not about a chatbot throwing up features or T&Cs of a product for customer enquiries, it’s about placing the request in context. A loan query can have very different responses depending on if the customer is a first timer or a repeat buyer who had defaulted in the past. Is it an emergency request for healthcare expenses or is it guilt of overspending on a vacation.
Content must initiate demand, nurture the need over a period and then qualify it as a ready to transact customer. Every app will have an icon for upgrading the subscription plan and no doubt it is important to ease out that journey. But the future is tilted towards firms that invest in content personalization as the journeys get standardized. What makes designing relevant content a forbidden fruit today is the sheer volume available online confusing both the newbies and experts. To be honest it is way too easy to purchase and dump the best of content to clients. Understanding a customer’s past interactions and to digitally engage them with the right content in quality and quantity that enables them to deepen as well as buy more products is the pinnacle of the AI sport. If you really think about it, this is exactly what sales teams take pride in accomplishing!
Data &AI: The invariant & the variant
It’s been a human endeavour for ages to relate correlation with causality and predict the next causal event from similar correlation. This is easier said than done and often the analysis results in spurious correlations arising from either overfilling or plain inaccuracy of data. While computing power has made algorithms on loop ridiculously easy, the need of the hour is to build an ecosystem.
An ecosystem that can help disambiguate ‘explainability of AI’, create a three-path empathy between Data, AI and Humans, the end users of the ecosystem. An ecosystem that can be interpreted not just by scientists and mathematicians in their arcane language of higher order mathematics but by practitioners in their daily commerce. We are closer than ever from AI white papers turning into ubiquitous daily use tools that makes us smarter and productive. Hoping 2023 turns out to be that year of transformation!