Source: Finance Derivatives
By Kevin Levitt, Global Industry Business Development, Financial Services, NVIDIA
As we enter the new year, many are left in a continued state of uncertainty from the unprecedented financial challenges 2022 posed. With high inflation and rising interest rates, financial institutions are more inclined than ever to turn to innovation to address top opportunities: leveraging AI to deliver more value, create relevant customer experiences and drive operational efficiency.
Artificial intelligence (AI) has long been a hot topic of interest in the financial sector, but now as firms are beginning to see its concrete value, it is moving from experiments in the research labs to production deployments across lines of business. AI’s value to financial enterprises will accelerate in 2023, solidifying its well-deserved spot in the technology toolbox of financial institutions.
Real-Time Risk Management
This year’s sudden collapse of cryptocurrency exchange FTX serves as a great example of the speed at which unknown risks can manifest for businesses. Financial firms will advance AI for risk management in 2023 as the importance of real-time risk calculations is more evident than ever. We live in a volatile financial climate where more accurate and immediate asset and risk values are needed to make gametime decisions. Increasingly, firms will turn to accelerated computing to speed price discovery, risk calculations and back testing.
The simulation techniques used to value risk in derivatives trading are computationally intensive and typically consume large swaths of data center space, power and cooling. In 2023, what used to run all night on traditional compute will run over a lunch break or faster on accelerated compute. A real-time value of sensitivities will enable firms to better manage risk and improve the value they deliver to their investors. Firms need to adopt AI and accelerated compute or run the risk of being left behind.
This technology won’t just be used for the capital markets sides of global banks. It will also apply to hedge funds, who will use accelerated compute to train algorithms to discover signals and produce more accurate predictions. They will deploy increasingly larger and complex models in production to capture the market opportunity and improve investor returns.
Cloud-First for Financial Services
In 2023, banks will have a new imperative: get agile fast. Facing increasing competition from non-traditional financial institutions, changing customer expectations rising from their experiences in other industries, and saddled with legacy infrastructure and economic headwinds, banks and other financial institutions will embrace a cloud-first AI approach.
But in the highly regulated financial services industry, some data can’t move to the cloud due to compliance requirements, data sovereignty laws or other international regulations. Banks are also finding some workloads to be cost prohibitive in the cloud, including training Large Language Models (LLMs) that form the foundation of many of a bank’s most important AI applications, from virtual assistants and chatbots to sentiment analysis of earnings calls and monitoring for trading floor compliance.
The industry requires operational resiliency, a term that means your systems can absorb and survive shocks (like a pandemic). To this end, in 2023 banks will look for open, portable, hardened, hybrid solutions – including AI software that operates across on-prem and cloud environments to offer a single pane of glass through which they may manage data science experiments and infrastructure utilization. As more AI applications are deployed into production, banks will be obligated to purchase support agreements for their AI software rather than rely on community-supported, open-source solutions.
Banking on Generative AI
This year, the world witnessed the sensation of ChatGPT from OpenAI, which showcases the power of LLMs and generative AI for a variety of use cases from story writing to writing software code. In 2023, banks will ride this wave to enable a range of applications with LLMs. Generative AI can be used to write emails personalized to an individual customer’s profile or lifestage, help software engineers improve their code, and help marketers create better cross-sell messaging.
As stated earlier, banks will not be able to rely on open-source solutions trained on uncurated corpuses of data like the one used in ChatGPT. They will need reliable frameworks for building, training, and fine-tuning GPU-accelerated speech and natural language understanding (NLU) models. LLMs are also transformer-based, requiring massive compute and data sets to train. Without accelerated computing, banks will incur surprisingly high unbudgeted costs and longer times to value.
Putting a Face to Digital Banking
The chatbot has been a mainstay in the customer service experience across the financial services industry for many years. However, many customers express frustrations with this experience as it can feel cold and disconnected. The problem is that chatbots aren’t empathetic. In 2023, banks will invest further in digital avatars to improve customer experience.
Digital avatars go beyond chatbots and rules to bring eye contact, gestures, and audio to face to customer service interactions. They also leverage LLMs, so the number of questions and tasks they can manage is not limited to pre-written scripts, but still bound by the guardrails dictated by the data scientists use for fine-tuning the models. Digital avatars will change how virtual assistants look and feel, ultimately building trust with bank customers.
New technologies like this promise to make 2023 an exciting year of innovation across the industry. But all of these improvements are built on a foundation of data science, which requires data scientists.
The new State of AI in Financial Services survey report to be released in January shows the number-one challenge financial institutions are facing is hiring and retaining data scientists. To attract these top talents, banks will need to provide them with the accelerated compute necessary to perform their best work and deliver on the promise of these innovations in 2023.
