A recent analysis of European startups revealed that a large number of those who purported to be using artificial intelligence (AI) were in reality not.  It seemed a nice metaphor for the hype that has surrounded AI in recent years, with expectations wildly exceeding reality.

A recent report from The Alan Turing Institute explores whether a similar gap exists between expectation and reality in the financial services sector.  The report reveals that financial services is the 2nd biggest spender on AI, after the technology sector, with high frequency trading firms (HFT) particularly keen on fully exploring the potential of algorithmic trading.

“HFT firms were the main users of AI in finance, but applications have now spread to other areas including banks, regulators, Fintech, insurance firms to name a few,” the author explains. “Within the financial services industry, AI applications include: algorithmic trading, portfolio composition and optimization, model validation, back testing, robo-advising, virtual customer assistants, market impact analysis, regulatory compliance and stress testing.”

AI in financial services

The report identifies three core applications of AI in finance today:

• Fraud detection and compliance – Fraud detection is one of the more interesting ones, as banks and other financial service providers spend billions combating financial crime each year, yet the last few years have seen a rise in cases of payment-related fraud. The idea is that by enormously increasing the data used to predict fraud, AI can both make the process more effective but also considerably faster. The report highlights examples from the likes of HSBC and NatWest where AI is being used to improve their fraud detection rates.
• Banking chatbots and robo-advisory services – Chatbots are perhaps the most visible of the AI technologies in use in financial services today, with most designed to assist customers in managing their money more effectively. Some of these apps are hosted on the financial company’s own platforms, others integrate with the likes of Facebook Messenger or Slack.
• Algorithmic trading – Algorithmic trading (AT) has rapidly ascended to the dominant force in global financial markets. It’s estimated that machines are making up to 70% of equity market trades, 60% of futures trades, and 50% of treasuries. The logic is clear: computers are able to trade faster, more accurately and in a way that ensures the best price is secured for both parties. All whilst removing the potential for human errors.

The paper then goes on to provide an overview of the various forms of artificial intelligence being used today, which as you might expect from the Turing Institute is an excellent summary.  The paper then explores some of the regulatory and policy hurdles that need to be overcome, not least of which is to ensure policies and regulations can keep up with the speed of change in the technology.

This is a field that the authors believe AI can also play an interesting role, with a burgeoning RegTech market aiming to bring new technologies to the market.

“RegTech utilises Big Data and ML and is an emerging field to reduce costs and increase effectiveness,” they say. “Researchers argue that ML techniques can provide fast, accurate and consistent judgements, and streamline operations with reduced error.”

This belief in the ability for AI to regulate AI perhaps explains why governments are investing so heavily in their AI capabilities, with Germany leading the way with a €3bn investment, but with both the technological landscape and the regulatory landscape in a constant dance, it’s a treadmill governments cannot afford to step off of.

“Artificial intelligence in the financial services industry is still in early days,” the authors conclude. “AI will become more ubiquitous in finance, and with that comes more challenges including legal, ethical, economic and social hurdles. AI will also continue to bring new complexities to the global financial ecosystem. As more and more data become available and computing power increases, AI programs will become more complex.”

As a starting point in helping to understand some of these issues, the paper provides an excellent grounding.  Worth reading for anyone with an interest in the topic.