AI is suffering from its own success. As hype balloons, misunderstandings about underlying processes fuel expectations and bizarre apocalyptic predictions. At the same time, an increase in companies jumping on the AI bandwagon has sparked uncertainty around whether solutions offer real innovation, or are just more “AI washing”.

There is, however, one sector where AI has long powered solid results: finance. Two years ago, seven in ten financial firms were using machine learning to streamline tasks, with 83% even developing custom models for internal functions. Today, half of UK finance firms have invested in AI and almost the same number (43%) are set to do so in 2024.

Cynics looking for proof of AI’s practical value need to look no further than the innovative finance space, particularly in three areas of implementation.

Powering Streamlined Processes 

Although gains from AI-accelerated efficiency vary by firm, arguably the clearest come from small refinements that deliver vast time savings. Take, for example, invoice processing.

Previously, manual data entry and calculation were slow, arduous and prone to expensive errors, especially as AI document fraud grows.

Developments in intelligent document processing (IDP) have automated data collation from documents, invoice classification and overall processing. This means finance teams benefit from more time to use their expertise elsewhere, lower risk of errors and better scope to unlock efficiency advantages, such as early payment discounts.

Banks have also begun using large language models (LLMs) to process diverse transactional documents with accuracy, speed and scalability. An upgrade from optical character recognition (OCR) solutions — which struggle with complex layouts, page breaks and languages — self-learning, adaptative LLM tools minimize human-in-the-loop intervention while sustaining security and data integrity.

Easily integrating with robotic process automation (RPA), engines trained on real, annotated multi-source data enable teams to assess documents against unique criteria in minutes, not hours. A McKinsey report estimates that in the banking sector alone, generative AI-increased productivity will bring an extra $200 billion to $340 billion in value globally.

Fighting Cyber Fraud

With cybercrime costs hitting the finance sector hard, reaching $5.9 million per breach, AI is taking the fight from reactive to proactive. Detection methods typically entail training models on data about past payments and scams to enable pre-emptive identification of abnormal patterns and potential threats.

Earlier this year, Mastercard reinforced its Decision Intelligence (DI) technology with generative AI scanning that can evaluate the relationships between one trillion data points to gauge whether a transaction is likely to be fraudulent, an approach that has enhanced its fraud detection rates by as much as 300%.

On the business side, AI also offers versatile protection from malicious schemes, including supplier invoice fraud. Although approaches such as three-way matching have long been used to ensure funds don’t end up in the hands of bad actors, rigidity around variations in line items frequently results in a blanket stop on legitimate and scam payments.

In contrast, situational AI evaluation fosters nuanced consideration of each deviation to reduce the chances of false positives. It’s predicted that generative AI will curb false positives in threat detection by 30% over the next few years.

Simplifying Good Governance

The agility of AI is perfectly suited to achieving multiple goals at once, such as establishing smoother workflows and ensuring rigorous governance. In addition to enabling finance companies to process complicated documents quickly, AI analysis makes it simpler and faster to mine them for granular insights that can be used to bolster business value and deliver accurate forecasts of strategic outcomes.

Simultaneously, robust validation can be deployed to check whether documents follow strict industry rules. At a basic level, using AI to automatically reconfigure documents according to global specifications boosts compliance and saves teams from the hassle of wrestling with varying formats – especially when solutions identify and resolve discrepancies.

Smart technologies are also playing an integral part in minimizing money laundering risks. Google’s new Anti-Money Laundering (AML) AI is one of the latest offerings using risk scoring informed by transaction, account and customer information to highlight suspicious activity, with its HSBC trial seeing up to four times higher positive alerts.

Cutting through the hype is getting harder as AI buzz increases. However, fixing sights on finance gives businesses clarity about what it truly offers. Real use cases show that there is plenty of potential for existing and emerging tools to deliver practical benefits for human workforces, enabling them to increase their efficiency and productivity. In short, AI collaboration empowers people to work better, faster and, crucially, safer.

LeackStat 2024