AI in Finance: Exploring the Leading Edge

Artificial Intelligence (AI) is transforming the Finance function by enhancing efficiency and improving decision-making capabilities, with Generative AI driving an even greater shift than previous technologies. AI-powered financial planning and analytics can drive growth, streamline operations and add value, but CFOs must balance innovation with control and risk management. At a recent India CFO Forum session in Bangalore, Dr Phaphul Chandra, Professor of Artificial Intelligence and Director at Atria University, explored real-life applications and best practices for implementing AI in Finance, and shared some key considerations around change management.

The AI Revolution

The buzz surrounding AI, particularly since the launch of ChatGPT in late 2022, has been unprecedented. ChatGPT’s rapid adoption, gaining a million users in just five days, reflects the scale and excitement surrounding AI’s potential. This surge is driven by the convergence of three key factors: exponential growth in computing power, an explosion of data availability and significant advancements in AI algorithms. As AI continues to evolve at an extraordinary pace, it is reshaping industries and functions, including Finance. The technology is no longer a novelty but a viable solution, ready for both consumer and enterprise use. However, the question remains: Will AI replace humans in decision-making? While AI promises productivity gains of up to 50%, surpassing the Internet's 10%, it also brings inevitable challenges, such as job displacement. AI-led automation will streamline operations, but the impact on employment cannot be ignored.

AI Case Studies in Business and Finance

AI has revolutionised various sectors and offers numerous compelling case studies. One of the earliest AI use cases emerged in the 1990s when Fico (then HNC) implemented neural networks to detect fraud in credit and debit card transactions. This system, which continues to power 70-80% of fraud detection today, exemplifies the power of AI in improving security and operational efficiency in financial services. Moving into the 2000s, AI began to transform the way banks approached credit scoring. By analysing a variety of structured and unstructured data, such as credit history and transaction statements, AI models predicted the likelihood of a customer repaying a loan. This revolutionised how financial institutions assessed risk, with AI becoming integral in shaping the terms of credit for retail customers.

In the 2010s, the use of satellite data by financial institutions like UBS to predict revenue based on traffic patterns at Walmart’s parking lots highlighted a novel AI application. By leveraging public satellite data, UBS was able to forecast quarterly revenues and make informed stock trading decisions, demonstrating AI’s potential in market analysis. This trend expanded into anti-money laundering (AML) and compliance efforts, where AI is now reducing the number of false positives in fraud alerts, significantly improving the efficiency of compliance teams. More recently, ESG compliance has emerged as another crucial use-case. AI has been pivotal in collecting and analysing data across company value chains to support sustainable practices. The latest breakthrough is JPMorgan’s contract intelligence tool, which has saved the company 360,000 man-hours annually by automating contract review and reducing human error.

 The trend of deploying AI across the business continues to deepen with tools like Microsoft Copilot, which helps CFOs generate financial reports, streamline processes and save time. Additionally, AI is being applied to tasks as simple as scheduling meetings across enterprise boundaries, showcasing its ability to handle even the most mundane tasks efficiently. These case studies underscore the profound impact AI is having across finance, compliance, operations and even sustainability.

Leveraging Data for AI in Finance

For CFOs looking to leverage AI and machine learning, three main types of data are crucial:

1. Structured Data: Traditional spreadsheets, ERP, CRM and financial accounting software contain structured data that has long been used in Finance. AI now enhances how companies extract value from these datasets, automating insights and improving decision-making.

2. Unstructured Data: Contracts, financial reports, compliance documents and even public filings contain valuable insights. The rise of generative AI post-2022 has made processing unstructured data easier, enabling AI-driven analytics in finance and algorithmic trading.

3. Semi-Structured Data: Data from IoT sensors, web activity and alternative sources like satellite imagery provide real-time insights. Though not yet widely used in Finance, applications like supply chain tracking and market forecasting are emerging.

While structured data remains the low-hanging fruit, the ability to extract insights from unstructured data will define AI-driven finance in the coming years.

The End-to-End Process of Deploying AI Solutions

It can be a challenge to deploy AI or machine learning solutions within an enterprise, particularly when building solutions internally. The process starts with setting up the development environment, where developers work with domain experts (like CFOs) to prepare enterprise-specific data. This preparation can be time-consuming. After the data is ready, the next step is training the AI model, whether built from scratch or customised from open-source solutions, followed by packaging, serialising and deploying it into the enterprise system. Even after deployment, the model requires ongoing retraining as new data, such as updated financial reports or competitor shifts, comes in. The process also involves managing the large computing power needed to scale AI solutions. Beyond deployment, continuous monitoring of the model's performance and the associated running costs is essential. While these complexities are part of building custom AI solutions internally, using AI from vendors like ERP or CRM providers simplifies the process. Nevertheless, understanding these steps highlights the challenges and resources required to implement AI at scale within an organisation.

Where Should Finance Leaders Begin with AI Integration?

Finance leaders should start by focusing on existing tools and platforms, particularly those provided by their ERP and CRM vendors. This is the low-hanging fruit for a quick and impactful start. AI technologies are advancing rapidly, with new tools, algorithms and data sets emerging weekly. While the hype surrounding AI can sometimes make it seem overvalued, rapid progress in this area makes it clear that AI is a long-term play with significant potential. Therefore, by leveraging current tool vendors, Finance leaders can integrate AI solutions without the need for extensive internal development or facing immediate regulatory challenges.

A Double-Edged Sword

AI has reached remarkable levels of cognitive capability, with models like GPT-4 achieving an estimated IQ of 152, outperforming 99% of human test-takers. These advancements enable AI to excel in complex assessments such as the SAT, GRE and even professional certification exams. However, this raises a philosophical debate: as AI becomes more capable, does it make humans less intelligent by reducing the need for critical thinking? While historically, automation has given rise to shifting cognitive behaviour – from calculators simplifying arithmetic to smartphones replacing memory recall – it also expands opportunities for higher-level abstract thinking. The real challenge lies in whether individuals leverage AI to enhance their capabilities or instead, become overly reliant on it. Ultimately, AI’s impact is not inherently good or bad, but depends on how societies and businesses choose to harness its potential.