By John Manning, International Banker
It’s an exciting time for financial services. A veritable smorgasbord of new, interrelated technologies are brewing up a perfect storm of disruption in the industry, including blockchain, data science, cloud computing and biometrics. Arguably, however, it is the significant advancement being achieved in the world of artificial intelligence (AI) that is having the most transformational impact on banking. In turn, AI is expected to permanently change the industry in profound ways during the coming months and years.
Few business sectors are currently more focused on developing AI for its own betterment than banking, as financial institutions seek to gain a competitive edge on their peers by implementing the technology to achieve improvements in speed, cost, accuracy and efficiency, as well as meet customer needs in an altogether more comprehensive manner. Much of the anticipation surrounds the ability of machines to replicate, and often exceed, what humans are able to do in banking. By collecting and utilising data to identify a whole host of patterns, moreover, those machines can become more adept at predicting activity that will ultimately make banks more efficient, from the front office to the back end.
The banking sectors being influenced by AI
In terms of improving the customer experience, it is perhaps the chatbots that are currently the most visible form of AI being adopted across the sector. These automated service assistants are providing customers with the convenience of resolving their queries via an online messaging system, perhaps using their laptops or smartphones, instead of having to visit a branch. Through machine-learning techniques, more importantly, chatbots are improving consistently with regards to their ability to accurately identify the customer’s issue(s) and respond with the appropriate solution(s). As such, today they can recognise tens of thousands of variations on common questions a customer is likely to ask.
Chatbots are now practically an industry standard, and while they may not be able to perform more complicated tasks, they are undoubtedly improving their capabilities. In January, the Commonwealth Bank of Australia (CBA) launched its in-house bot Ceba to more than a million customers, and which at last count could boast of being able to perform 200 tasks. But CBA expects that by the end of the year, this number will hit the 500-mark, while Ceba will also be able to successfully discern 500,000 different ways customers may ask for different banking activities. And then there’s Nina, Swedbank’s AI chatbot, which has been in operation for around two years. In addition to answering customer questions on an online chat platform, Nina can also pass on more complex calls to members of staff. The bot has managed to achieve a 78-percent first-contact resolution within the first three months, in addition to a customer adoption rate of 30,000 conversations per month during this time.
But in terms of AI development and what could ultimately be achieved in banking, chatbots appear to be just the beginning. At the other end of the complexity scale is the usage of bots on the trading floor, primarily as a way to help traders perform more competently. In sectors such as algorithmic trading, AI algorithms can be developed in order to produce highly refined investment strategies. As opined last year by Michael Harte, Barclays’ group head of innovation, the clearest use case for AI in banking is “in large algorithmic trading”, which means “using vast amounts of high velocity data to outsmart the competition and to provide better instruments and value to customers”. AI is proving especially useful in high-frequency trading, which is a specific type of algorithmic trading characterised by high order rates and often ultra-fast trade execution. Given the short window of time in which trading opportunities present themselves in such markets, robots can be programmed to more efficiently capitalise on such opportunities than humans.
Swiss bank UBS last year debuted its two new AI systems on the trading floor, one of which is a system that analyses reams of market data to identify trading patterns, and thus formulates new strategies for trading volatility for the bank’s clients. Through machine learning, therefore, the system can improve its trading performance, which should ultimately translate into greater market returns for UBS’ trading division. The second is a relatively more straightforward protocol that was jointly developed with Deloitte and which addresses the post-trade allocation preferences of the bank’s clients. The system scans through emails sent by clients to identify their requests for splitting up large block trades among their various funds, before executing the transfers. By doing so, a 45-minute task can now be accomplished in just a few minutes, thus freeing up time for bankers to focus on more important jobs.
Indeed, management of customer data appears to be one area in which AI adoption is progressing comparatively quickly. The UBS emails are just one form of such data; others include news articles, recorded phone conversations and legal documents, with JPMorgan Chase’s contract intelligence (COIN) being an example of the latter. While manual review of the approximate 12,000 commercial credit agreements per year is estimated to take 360,000 hours, or 173 years, the use of COIN reduces that time down to just a matter of a few seconds. Clearly the time and cost implications of such a powerful analytical tool can be substantial.
Similarly, AI is making considerable inroads on the compliance and security aspects of banking. With money laundering continuing to be a persistent problem for the global banking industry, there is now much anticipation over what technologies such as machine learning, deep learning, data mining and analytics can do to combat this threat, especially now that banks are being fined more frequently due to failings/inadequacies in their anti-money-laundering infrastructures. HSBC is among the keenest to explore such options—in April, it was revealed that following a successful pilot, the bank had partnered with big data startup Quantexa to utilise AI software to counter money laundering. According to Quantexa’s press release, “The technology will allow HSBC to spot potential money laundering activity by analysing internal, publicly available, and transactional data within a customer’s wider network”.
Moreover, the new venture follows another AI-based HSBC partnership last summer with startup Ayasdi to automate anti-money-laundering investigations that were previously being conducted by thousands of humans, according to the bank’s COO Andy Maguire. The aim of the initiative is to improve efficiency in this area, especially given that the overwhelming majority of money-laundering investigations at banks do notfind suspicious activity, which means that engaging in such tasks can be incredibly wasteful. In the pilot with Ayasdi, however, HSBC reportedly managed to reduce the number of investigations by 20 percent without reducing the number of cases referred for more scrutiny.
In addition to having an impact on specific divisions with the bank, AI is also playing a crucial role in changing banks’ inner organisational structures. For instance, US banking giant Wells Fargo announced last year the establishment of a new AI Enterprise Solutions team, which would operate under a new Payments, Virtual Solutions and Innovation group. According to the bank’s official announcement, the group aims “to help increase connectivity for the company’s payments efforts, accelerate opportunities with artificial intelligence, and advance application programming interfaces to corporate banking customers”, with the ultimate goal of helping the company “create innovative digital banking experiences”, as well as make it easier for customers to achieve their financial goals.
A profound impact on banking-industry employment awaits?
While there is much expectation over the positive change that AI can bring, there is also considerable trepidation surrounding the technology. One of the most frequent concerns being raised is the high degree of substitutability bank employees are likely to have with robots. As machines continue to become more advanced, those jobs that involve significant repetition such as bank tellers are likely to be at risk first, before jobs that require more complexity begin to be threatened. According to Lex Sokolin, who is the global director for fintech research firm Autonomous Next, AI adoption across the financial-services industry could save US companies up to $1 trillion in productivity gains and lower overall employment costs by 2030. Sokolin also predicts that such changes in financial services will only be gradual until 2025, before accelerating from then until 2030.
And according to a recent Autonomous Next report, banking and lending will see the biggest transformation, with $450 billion in savings being potentially achievable—as well as a whopping 1.2 million jobs being at risk; followed by insurance with $400 billion in savings and 865,000 jobs under threat; and finally the investment-management sector, in which 460,000 jobs could go while $200 billion in savings could transpire. What’s more, of those AI-induced banking-job losses, 70 percent will be in the front office—485,000 tellers, 219,000 customer-service representatives and 174,000 loan interviewers and clerks; and they will be replaced by chatbots, voice assistants and automated authentication and biometric technology. In other areas of the bank, 96,000 financial managers and 13,000 compliance officers will be let go in favour of AI-powered anti-money-laundering, anti-fraud, compliance and monitoring tools. AI-based credit-underwriting and smart-contracts technology will also be responsible for the shedding of 250,000 loan officers.
Moreover, such fears regarding employment are confirmed by some of the world’s most reputable banking leaders. Ex-Citigroup head Vikram Pandit, for example, expects that AI could render as many as 30 percent of banking jobs obsolete in the next five years, asserting that AI and robotics “reduce the need for staff in roles such as back office functions”. Japan’s Mizuho Group plans to replace 19,000 employees with AI-related functionality by 2027, and recently departed Deutsche Bank CEO John Cryan had once considered replacing almost 100,000 of the bank’s personnel with robots.
At the same time, however, conflicting data suggests that the proliferation of AI will be accompanied by a rise in banking jobs. A recent study from Accenture found that by 2022, a net gain in jobs of 14 percent is likely to materialise among those companies that effectively utilise AI, along with a 34-percent increase in revenues. With complex algorithms likely to be employed by AI, banking shareholders, regulators and other parties with vested interests are unlikely to be assured if there is no one around to explain the workings and methodologies of the machines. Such a view seems to be demonstrably supported by Mr. Ken Wong, head of AI Lab (Fintech and Innovation Group) at Singapore’s OCBC Bank, who recently acknowledged that his bank has made more hires related to AI since last year: “Besides data scientists, who develop AI algorithms and models, we are also hiring more engineers to manage these AI applications, and business managers who understand how to leverage AI within the business to increase efficiencies and improve customer experiences.”
It may also be the case that only the most mundane jobs such as data entry will be sacrificed for machine superiority. New jobs are likely to be created around the AI software in such fields as compliance to ensure regulators are provided with sufficient communication and explanations of the AI being implemented. What’s more, Accenture sees AI as being a likely catalyst for liberating humans to be able to work on more interesting and complex jobs. Its survey of 1,300 bank employees found that 67 percent of respondents expect AI to improve their own work-life balance, while 57 percent expect the rise of the machines to actually help their own career prospects.
Whilst a rise in job numbers associated with higher AI-adoption rates seems ideal, however, some evidence suggests that most financial institutions are not yet fully confident in how to effectively apply the technology for the best results. This may go some way towards explaining the ambiguities of the aforementioned numbers and trends. New research from capital-markets research and consulting firm TABB Group, conducted on behalf of augmented-intelligence-solutions provider Squirro, for example, found that 67 percent of banks are actively using AI today, despite the fact that 87 percent of those exploring the technology remain unsure of how to deploy it effectively. As Miguel Rodriguez, vice president of customer success at Squirro, observed, the report “has shown that banks are increasingly receptive to using AI and machine learning in their organisation, but it also highlighted what needs to be done to ensure maximum value is gained”.
AI has become something of a buzzword, not just in banking but for many industries. But behind the hype lies the potential for genuine transformation in the way the banking industry—and indeed, the world—functions. And with specialised hardware being equipped with ever-increasing processing power to enable more robust AI systems, as well as vastly improved infrastructure to mine and feed data to such systems for even greater capabilities, the banking industry is now realising the extent to which machine intelligence can positively redefine how it operates. Of course, several questions remain, such as how exactly banks can co-ordinate AI’s awesome power with the customer’s need for human interaction when needed. Nevertheless, some dramatic changes now loom on the horizon for banking.
Thank! There is a serious misconception that AI in finance means forecasting the stock market and that “everyone does it” when in fact, operations, marketing and security have the greatest impact. There is also a serious problem when financial companies seek to hire someone to improve their work and gain machine learning experience. As a result, they have a scientist who spends most of his time with Excel macros! Not an ideal situation for either side.
Check some tips for the AI in Banking