-Authored by Gautam Bandyopadhyay, Co-Founder, and CEO, Trustt
As published in Times Now
For a highly regulated industry such as Banking and Financial services, data privacy and data residency as per country regulations are paramount & non-negotiable. In early May 2023, Samsung banned the usage of ChatGPT by its employees after a leak of proprietary data. On the 19th of May 2023, Apple joined other corporations in restricting the usage of ChatGPT by its employees. So, do you think that a rather regulated industry will adopt Generative AI? Well, my personal experience of showcasing our own Generative AI-based product viz. Trustt GPT has met with great interest. Several pilots and paid POCs are underway for our Customers. The decision-makers are aware of the disruptive powers of Generative AI and do not want to be left behind. It will not be far before the adoption of Generative AI from education/experimentation moves to be at the center stage of Business as usual.
With over a few decades of experience in the realm of technological advancements, I have closely observed the ever-growing challenges faced by the banking and financial services industry. The need for efficient customer service, risk management, fraud detection, and regulatory compliance has always been at the forefront of this industry. These problems required very sophisticated solutions based on Artificial AI - which only the deep-pocketed budgets could experiment with. The shortage of talent and the fast pace of changing technology made it very difficult to keep up. The advent of LLMs has brought about a paradigm shift, which I call as “Democratization of Generative AI”. There is a high-stake full-fledged battle to make the LLMs bigger, better, and more versatile. LLMs are metamorphing themselves as “operating systems”, around which Business applications can be built by a community of collaborators.
This is the time of the year when we set goals, place bets, and predict technology trends (Lol). Let me place my bets as to where Generative AI will make the most impact and will probably be the first set of use cases to be adopted at scale.
Customer Service and contact center experiences - be in BFSI or otherwise - is a broken experience and it is begging for solutions that can improve productivity, be contextual to the customer, and bridge knowledge gaps & language barriers. First-generation NLP-based chatbots have been sub-optimal. They take an immense amount of training time, are menu-driven, and are incapable of handling natural human conversations. This makes Customer Service an ideal candidate to look for help from Generative-AI-backed technologies. However, there is a major challenge with Generative AI, its responses are probabilistic and we can never be certain that it will not hallucinate. Hallucination will lead to wrong information being provided to Customers and it will be a strict No-No for BFSI. As a result, what will work is a “Co-Pilot” for human support executives. Generative AI will listen to the customer conversation and provide a response suggestion. The human in the loop will decide whether or not a response will be promoted to the end customer. This will be a “banking” (read conservative) way of using the power of Generative AI until we know that certain processes have matured enough to be promoted to a self-service mode.
Generative AI-based Co-Pilots will reach stardom, not only in customer service but in other areas as well. AI will truly stand for “Augmented Intelligence” rather than Artificial Intelligence. Field sales staff will be assisted by the chat support to search product details more efficiently, make the best suggestions (assisted by AI), and most likely win over a customer or let go (with the least wastage of time) by assessing the customer sentiment.
The same is true for internal training of staff, be it product training or process training.
Take, for example, the area of Regulatory Compliance. The BFSI sector operates in a highly regulated environment, and compliance with complex regulatory frameworks is a perpetual challenge. Gen AI-based Co-pilots will prove to be a valuable ally in ensuring compliance by automating the analysis and interpretation of regulatory guidelines. These models can be tuned to comprehend intricate legal jargon and extract actionable insights, read & interpret RBI circulars, make connections between circulars issued by different authorities, and interpret them with respect to the current policies of a Bank or NBFC. The Compliance team and CRO will be relieved by high-quality suggestions to streamline the compliance process. As an example, Gen AI models will be trained to scan contracts, identify potential risks, and suggest necessary amendments to align with regulatory requirements.
In the era of big data, banks have access to massive datasets, making it challenging to extract meaningful insights. On top of that, large unstructured data such as social media interactions, and voice files make the processing & insight generation an expensive affair. The BFSI sector is hugely invested in analytics technologies. However, Gen AI will make its adoption even more broad-based, democratizing access to everyone based on their span of control. It will simplify the generation of insights and predictions based on operational data.
Given the risk-averse nature of BFSI, this use case is also likely to be exposed to end customers as one of the early use cases for Gen AI. Imagine the customer delight, if a Bank can categorize and give graphical insights on her spending patterns across the last 12 months on 3 different cards, and the Customer, at her will, asks questions to slice and dice the analytics in many different ways! These will be possible, without having to build specific analytics dashboards.
Everything I described above is around the improvement of productivity and customer experience. It does not consider anything around how Customers can directly deal with BFSI providers, leading to huge revenue at a much lower cost of acquisition! Well, all is not lost. I think Gen AI will democratize access to Finance. The ability to understand and converse in natural languages in diverse languages will make BFSI providers more accessible to a larger section of society. The initial use cases will be around product discovery in voice-first interfaces and simple lead captures. Such leads may be completed in assisted mode using traditional approaches, at least for now. This will help drive Financial Inclusion, address new target segments, and add new revenues.
2024 will be an exciting new year and one can expect big innovations & large investments in AI to continue to grab headlines. At the foundation level, it will be an interesting play between OpenAI, Google, X, and Mistral AI. India will have its own country-specific AI models coming up, such as Sarvam which raised $41 million in Dec’23 to focus on voice-first interfaces supporting diverse Indian languages. I believe LLMs will unfold as “operating systems”, leading to huge collaborative work and diverse applications, bringing massive impacts much like what happened to IOS & Android ecosystems.
Enterprises & especially BFSI - will adopt Gen AI - albeit on a conservative footing. The majority of the use cases will continue to have humans in the loop, where Gen AI will be augmenting human intelligence. Only after the enterprises are reasonably sure of the groundedness and have tight control of Hallucination, will we see direct Customer-AI interfaces gaining dominance. Maybe that will be the hero of 2025!