Ready for the real world?
Artificial intelligence and other disruptive technologies can have a transformational impact on banking, but innovation must be aligned to customer need
Banks face disruption from external parties, but also from their own adoption of new technology and the process of change. Managing this process requires a well-considered strategy that exhibits real awareness of what technology can and cannot do.
Speaking at the Sibos 2017 plenary session ‘The growing significance of disruptive innovation and artificial intelligence’, Alex Manson, global head of transaction banking at Standard Chartered, said that, as banks embrace disruptive technology, they must avoid building solutions looking for a problem. With pressure to innovate in the face of competition and change in customer expectations, there is a risk that banks will follow the crowd before client needs and objectives are well understood.
“The definition of relevance is whether the customer cares,” Manson said. “If no customer cares, your solution may well be new, but it is irrelevant.”
One of the most transformative technological innovations – the application of artificial intelligence (AI) and machine learning – is inevitably surrounded by a lot of hype. If financial services providers are to gain real advantages from AI they need to filter out that noise in order to extract real value from technology.
“Today we see AI is the buzzword of the year, but I do believe it will fundamentally change the way we work and the way we operate,” said Axel Lehmann, group chief operating officer at UBS.
Lehmann noted that the timeframe for realising the promised returns and the potential cost/income ratio benefits of AI was not certain, likely between four to eight years to generate structural benefits, but suggested the technology could impact the business on two levels. First, it could enable the automation of more mundane, repetitive processes, thus impacting the scale of workforce necessary. second, it would also change how more value-added work, which directly drives revenue generation, is conducted.
“It will go right into the core,” he said. “In portfolio composition, we may soon not only be trading on price discovery: it may be that predictive pricing is possible. The trader of today will not be the same trader in ten years; most of what a trader is doing today can be enhanced and done with technology.”
A fertile bed
The application of AI to banking processes depends upon the type of technology being used; the term can be applied to a wide range of smart systems, from ‘narrow’ AI which is designed to handle and automate very specific problems to more flexible predictive and cognitive systems.
Speaking at the session, ‘The impending technological revolution in FinTech and artificial intelligence – Are you ready?’, Richard Nesbitt, president & chief executive officer of the Global Risk Institute, a Toronto-based research body, predicted that some of the more advanced technologies will not be a concern within the lifetime of most finance professionals. Yet the narrower applications available now are already incredibly useful for an enormous range of tasks.
“With narrow AI you can have self-driving cars; these are possible without general AI or super intelligence, and what that means for industry and banking will be quite profound,” Nesbitt said.
The increased electronification and digitisation of transactions and trading in recent years is yielding exponential growth in data available to financial service providers. The capacity to easily record, monitor and review trades has applications that range from revenue generation to regulatory supervision to cost management. However the volume of data that is available, the speed at which it is produced and the way it records information do not make it easy for humans to process, even when visualisation tools are available.
“AI has been taking off for the last couple of years due to three things,” says Nesbitt. “Firstly, huge amounts of data are being generated and stored. Secondly, advanced hardware is allowing cheap storage of and access to that data. Finally, advanced algorithms allow machines to learn independently from their programming.”
Rules-based engines are currently used to identify patterns within data, to scale up the capabilities of teams within banks, however they are constrained by the strict application of those rules to shifting patterns in real world activity, such as the changing modus operandi of cybersecurity threats. In contrast, noted Nesbitt, systems at the more sophisticated end of the AI spectrum can be trained on a dataset to learn patterns that help process the big data sets found in finance, but can also follow variations in those patterns and adapt to changes within them.
The future now
To illustrate specific use cases within banking, Dermot Canavan, head of trade finance services product management at ING, and Marc Smith director at technology provider Conpend, presented a case study, ‘Automated trade finance compliance screening using artificial intelligence’, focused on use of AI to automate a previously manual sanctions screening process. The system they developed captured data for analytics and audit, leveraging existing technology used by the bank for document management and optical character recognition.
The AI element improved several parts of the process, including document capture, which had proved challenging historically due to format variations. Conpend used the big data approach of taking all the raw data from the documents then applying algorithms that could select relevant data without creating a rigid template to match the layout. Once a proprietary engine captured and extracted data from the required fields within the trade documents, the system then interpreted the data and translated them into a common format.
The bank can then use the data to perform any checks necessary, processed via existing sanctions screening facilities. Machine-learning algorithms – acting in a similar fashion to a spam filter – are then used to assess the alerts raised, identifying false positives based upon the user behaviour. “In some cases that has reduced output down to zero, which allows straight-through processing,” said Smith.
Disruption and defence
The Global Risk Institute’s Nesbitt also highlighted the potential role of AI in addressing cybersecurity threats, pointing to the emerging threat posed by quantum computing, which is predicted to enable processing at faster speeds than binary-based systems. These could be employed to unpick their cybersecurity defences if banks do not invest in quantum cryptography and artificially intelligent defences. As increasingly powerful computing resources are made accessible via commoditised commercial models, including cloud accessible platforms, the ability of bad actors to deploy them for illicit purposes grows.
But employing AI can allow very rapid responses to cyber-attacks, said Nesbitt, where slight variations prove challenging for rules-based systems, helping to counter the speed at which such attacks can develop.
Recent technology innovations are already being used to tackle a range of specific challenges within banking, but at a strategic level they will also provide a way of helping businesses across the financial sector to achieve long-term viability and efficiency. For example, the application of disruptive technologies such as AI can make firms more resilient to operational risks by reducing the need for multiple semi-manual processes and parallel, but not integrated, siloes, simplifying the current organisational models found within banks.
“Right now there are complicated structures with wasted code and wasted interfaces,” said Amber Case, fellow at the Berkman Klein Center for Internet and Society, Harvard University, speaking in Thursday morning’s plenary session. “Each new feature increases the complexity of the system and the surface area available for attack.”
Replace and enhance
Allied to a customer-centric mindset, new technology can help banks avoid disruption from competitors, but they must also be careful not to disrupt their own organisations by changing too quickly. The effective adoption and integration of digital technologies which replace and enhance rather than mimic existing structures will be crucial, but the success of this effort will be dependent upon senior management imposing a strategy that manages changing process and culture.
“The culture of innovation has to be client-centric, it has to be focused on the human,” said Standard Chartered’s Manson. “It has another element, which is the need to have the courage to discard certain current ways of doing things, as well as trying things that may not work: learning is not failure, learning is a success and we are encouraging this within our organisation, in day-to-day operations, but also in the way we engineer and the way we innovate.”
As we were reminded in the ‘Surviving disruption in financial services’ panel, adoption of new technology and adjustment of business models is not only about focusing on the human customer, but also addressing the people problems that can exist within a team.
“When you have four or five generations working together there are challenges,” said Paul Francisco, chief diversity officer at State Street. “How do you bring those generations together? A Millennial may think he or she is more prepared because they have easy access to information and technology versus someone in the firm with 25 years’ experience. Companies that are really figuring this out fastest get to market the fastest and that is a competitive advantage.”