Skip to content Skip to sidebar Skip to footer

Source: Finextra as of 02-07-2020

When will artificial intelligence really have ‘arrived’? For a long time, this was a question for philosophers and computer scientists, pondering over whether passing the Turing test truly indicates intelligence, or debating about how broad our definition of artificial intelligence should be. Over the last several years, however, this question has changed considerably: with the advent of consumer AI tools such as virtual assistants and the increasing availability of off-the-shelf solutions offering to bring the power of AI to business operations, the issue has become less philosophical, and much more pragmatic. Now, for business leaders, it is often a matter not of whether to respond to the arrival of AI, but of how to respond to the arrival of AIThe promise is, of course, huge. It’s hard to think of an area of the economy which won’t be changed, and it’s hard to think of a short summary which properly encapsulates what those changes will be. It’s probable that for every business, there is at least one thing which will be done faster, or made available to far more people, or be significantly more accurate, or will simply be possible for the first time, through the use of AI. Its importance has been compared, convincingly, to electricity itself. Nowhere is this more true than in financial services. Whatever your definition of artificial intelligence is, or whatever particular aspect of this broad technological category you happen to be looking at, it is clear that AI relies on data – and, often, vast quantities of it. While all sectors generate their fair share of data, and are finding innovative ways of gathering more, higher-quality data, finance is and always has been a data industry right down to its core. Where retail has goods, agriculture has food, and entertainment has people’s emotions, finance deals with information.

In some ways, this makes AI an easy fit for the financial services industry – and, indeed, a huge amount of progress has been made. A report published last year by The Alan Turing Institute details some of the sector’s significant applications of this latest wave of development, from the AI-based algorithmic trading which now accounts for around half of all trading activity, to automated fraud detection working to minimise false positives, to chatbots which help consumers manage their money in a naturalistic, conversational setting. At the same time, however, the world around financial services is changing fast, and when expectations and assumptions about how things should work are being set in other sectors, finance risks falling behind. Whether it’s a music recommendation algorithm or an automated transcript of a conference call, consumers and businesses alike are becoming habituated to systems which anticipate needs in a personalised way. Bringing finance into line with this reality is challenging.

One reason for that is simply the accuracy and security which this industry demands. If a virtual assistant (or, indeed, a bank’s customer service chatbot) slightly misunderstands your query, it’s unlikely that much damage will be done. When the information at hand is money, however, things are more serious – as when a trading firm lost nearly half a billion dollars due to a computer glitch. Perhaps the bigger challenge for financial services, however, is that AI is an architectural innovation as well as a component innovation – which is to say, it requires not just introducing new technology and ideas, but joining up old technology and ideas in a different way. AI, remember, requires massive amounts of data: this is how it learns how things work, and it’s how it makes inferences about how those things will behave in the future. For many businesses, introducing the systems to manage this data will mean building up entirely new computing capacity alongside innovations like internet of things monitoring to gather the information which is needed.

In financial services, however, where information has always been the heart of the business, there is a tougher problem of transforming existing systems to speak a language which AI understands. Legacy systems in finance have been built up over the course of decades, and changing existing systems which are currently delivering value is a bigger, riskier job – in a highly risk-averse industry – than starting from scratch. The change is coming. One option for operating with legacy systems in a digitalised, intelligent context is to use robotic process automation. RPA uses AI to learn how a system functions without altering that original system in order to provide a bridge between it and the new infrastructure which has been designed to deliver the automation and insights which modern financial services need. A recent NetApp survey found that RPA is already enjoying broad adoption in the industry, particularly for areas such as customer services, fraud prevention, and portfolio management. Ultimately, however, we will see that significant architecture changes will really open up the horizon of possibility for this sector. The move to cloud computing, with its elastic response to demand that can handle the intensive computation that AI training requires without the capital expense involved in building that capacity in-house, is a key part of this. While, in many ways, financial services is a sector already at the leading edge of AI, the availability of architecture which is designed from the ground up for AI-driven operations means that much more change is set to arrive.

Show CommentsClose Comments

Leave a comment

News ORS © 2020. All Rights Reserved.