Back in October, US research and advisory group Gartner listed artificial intelligence (AI) and advanced machine learning (ML) as one of its ‘top 10’ strategic technology trends for 2017. “AI and machine learning have reached a critical tipping point and will increasingly augment and extend virtually every technology enabled service, thing or application,” the group reported.
“Creating intelligent systems that learn, adapt and potentially act autonomously rather than simply execute predefined instructions is primary battleground for technology vendors through at least 2020.”
AI and ML include everything from deep learning and neural networks to techniques like natural language processing (NLP) and its newer cousin natural language generation (NLG). NLP and NLG use an understanding of human language to process large amounts of data and generate natural language text – in the case of NLP – or to review unstructured data and give meaning to it with linguistic text (in the case of NLG). The opportunity for this technology to automate processes and bring new meaning to data in financial services is significant.
In fact, in a world where 80% of data is unstructured – according to ‘The Text Mining Handbook: Advanced Approaches in Analysing Unstructured Data’, by Ronen Feldman and James Sanger – NLP could be a powerful way to automate tasks, where it is well-known that machines makes fewer mistakes than humans when working on repetitive tasks.
What’s more, according to The Economist magazine, given the advent of digital neural networks (DNNs) over the past five years, DNNs are “helping to improve all manner of language technologies, often bringing enhancements of up to 30%. That has shifted language technology from usable at a pinch to really rather good. But so far no one has quite worked out what will move it on from merely good to reliably great.”
Four points to consider
For financial institutions looking to take advantage of the opportunity, keep the following four points in mind:
- Define the business case
While it might be tempting to try to find an out-of-the box solution, resist the urge. The NLP/NLG engine will need to be applied to a business case and technology framework to be effective; moreover, it will also need to be trained to understand the unique lexicon of that industry.
For example, which meaning of ‘trade’ am I using given the context and what action might therefore be needed? As the amount of data that financial industry generates is set to grow exponentially, in what areas is it becoming impossible for humans to process that data where natural language summaries can play an increasingly critical role to ensure the data is useful?
- Set a methodology
Companies should ask themselves what their plan is for the lexicon and sentiment analysis where intention matters and what languages matter for their business. There are now open source algorithms available that each takes a different approach to analysing and processing data. It requires an understanding of the business context to determine the best algorithms and techniques to use in that context and if NLP alone is needed or if multiple AI approaches need to be employed together to address the challenge; for example NLP + chatbots, or NLP + the Internet of Things (IoT) + payments.
- Develop the engine
Companies such as Google, or fintechs such as the multinational Yseop offer enterprise NLP and NLG toolkits that can used for the custom development work required for financial services. As with any project, firms should conduct a technology audit to understand which tools and technologies meet their technical and business requirements. The technology methodology for financial services and having the right engineering requires both domain knowledge and technical knowledge.
- Iterate, develop and evolve
Many large financial institutions have set up innovation groups that run alongside their daily business operations. This enables them to take an agile approach to technology development and continue to iterate and evolve their solutions in a ‘sandbox’ testing environment, isolating untested code changes and outright experimentation from the production environment. We believe this is a prudent approach, so new technologies can evolve alongside legacy before going fully into production and impacting global operations.
Two recent market research reports predict respectively that the AI market can expect a compound annual growth rate (CAGR) of 62.9% over six years to reach US$16bn by 2022 and that investment in AI will top US$5bn within the next three years (against just US$420m in 2014), with NLP an important part of that growth. Financial services firms can use the technology to automate manual processes and augment and elevate the role of the trader, the compliance offer and other important roles across the organisation.
The BoE launching digital currency on blockchain tech would be international payments ‘game changer’
The Bank of England may be the first major central bank to put its currency on a blockchain which would be a ... read more
For UK treasurers looking to manage financial risk within their organisation, recent news of an interest rate rise from 0.25% to 0.5% was a mixed blessing.
The industry consensus is that the best way to select a treasury management system is by analysing the 'Requests for Proposals' criteria. But is it? Does this lead to the best solution? If you adopt this methodology will you have a TMS that meets your operational needs?
Despite their importance to the world economy, SMEs often face problems accessing credit when and where they need it. Their banking needs are often more complex than the usual retail banking customer and they don’t offer banks the revenue potential of larger corporations.