Seizing the cognitive opportunity
Sibos 2016 explored the potential of artificial intelligence, machine learning and cognitive systems to help banks cut costs, add value and meet multiple challenges
Last year, much was made of the ability of a computer programme, called AlphaGo, to beat one of the very best players in the ancient Chinese game of Go. Despite its relatively simple rules, Go is a very complex abstract strategy game. The programme’s victory, which was clear-cut, was seen by some as a ‘breakout’ moment for machine learning and the prelude to its more widespread use in many fields, including financial services. Sibos 2016, through a series of panels and one plenary session, sought to explore the possibilities and opportunities in a number of aspects, including compliance and cybersecurity.
Ginni Rometty, chairman, president and CEO of IBM, struck an optimistic tone in Wednesday’s big issue debate, ‘Cognitive business and the future of financial services’.
Her optimism stems in part from a belief that financial services “can and will lead the way in the adoption and exploitation of cognitive computing”.
IBM has seen its business evolve from tabulating machines, through programmable machines, and is now entering a world of cognitive or learning machines. By augmenting the capabilities of key individuals in any business, these machines can help generate better decisions, whether in detecting cancer or spotting financial fraud. To justify her prediction that banking would be a fast adopter of cognitive systems, Rometty highlighted two particular strengths enjoyed by the industry.
First, banks are naturally digital in the way they operate and, Rometty believes, are about to reinforce their digital credentials through application of distributed ledger technology (DLT). Second, banks have always been at the forefront in the deployment of new computer technologies, precisely because their business is digital. By adopting cognitive computing early, banks have the opportunity to derive significant financial benefits, especially in the relatively untapped area of middle- and back-office processing.
By working with DLT directly, through an initiative within its global financing business, IBM understands better than most the scale of operational benefit that the technology can bring. As Rometty sees it, “Blockchain can create an environment of trust and efficiency in the exchange of almost anything, offering traceability as well as confirmation of authenticity to market participants.”
But IBM understands that for all its advantages, wide scale adoption will not occur without standards, governance and proven robustness. If these pieces are put in place Rometty is confident that “Blockchain can do for transactions what the internet did for information.” IBM’s continuing and significant support for the Hyperledger Project is testimony to their commitment to the use of distributed ledgers. The breadth and depth of blockchain discussions at Sibos 2016 suggests Rometty’s assumption around early adoption is not misplaced either. More practically, the recent announcement between IBM and CLS Group, as well as initiatives around smart contracts, identity management and trade finance, involving SWIFT members, IBM and Hyperledger, are indicative of real progress moving towards implementation at scale.
Ripe for innovation
If everyone has a robust digital platform operating with the efficiency enabled by blockchain, where will competitive differentiation come from? Joining Rometty on stage following her keynote, Sergio Ermotti, group chief executive officer of UBS, noted that, “Few banks are earning enough to cover their cost of capital and all are looking at ways to cut costs and also generate more revenues.” Both speakers agreed that the middle- and back-office operations of banks offer enormous scope for cost reduction through use of smarter technology. Whereas banks have been quite successful at deploying technology in the front office, other areas remain ripe for greater process efficiency through improved deployment of technology innovation.
In the long run, of course, banks have to grow revenues by winning market share and adding value to customers. To that end, Ermotti sees the ultimate difference between banks as being in “the quality of their people, the relationships they enjoy with clients and especially the level of trust that exists both personally and institutionally”. Those firms that put together the best quality learning machines with the best personnel as ‘teachers’ to guide their development will generate both cost savings and the opportunity to create new products. “This process is already underway,” Rometty noted, “with IBM’s Watson system being deployed in areas as diverse as customer service, clinical decision support and legal businesses.” IBM’s stated goal is to build commercial revenues from Watson to more than US$10 billion per annum over the next ten years. Financial services offers interesting opportunities in that regard.
Subsequently, at the Innotribe session ‘AI in financial services’ on Wednesday, three institutions presented their current work. Eric Rosenblum, an executive at data analytics solutions provider Palantir, described an application originally designed to give an institution a clear view of its overall relationship with any particular client. Once created, the tool could be used to understand all interactions between the firm and each client, which improves users’ ability to target new product initiatives. As Rosenblum noted, Palantir is far from a single-minded advocate of artificial intelligence (AI) solutions, believing that much of the opportunity comes from better structuring of underlying data. Nevertheless, Rosenblum’s example, together with use cases from US-based robo-advisor Betterment and BNP Paribas, reinforced the sense that AI technology is beginning to gain traction. BNP Paribas’ use case also involved a view of client connections and interactions, in their case interfaced with natural language programming to provide an easy to use internal tool.
Some speakers tempered the general level of enthusiasm, with those speaking in the session ‘Machine learning – The future of compliance?’ suggesting that new innovations must find their place alongside existing processes and techniques, rather than replacing them wholesale. As Anthony Fenwick, global head of AML at Citi, pointed out, unlike Go, “Financial crime does not stand still. The rules of the game, already complex, are constantly changing as criminals respond to current techniques aimed at their detection.” But there is no clear distinction between black and white. “A typical financial criminal these days looks very much like a normal legitimate client, whether representing themselves as a high net worth individual or a representative of a corporate entity seeking to open accounts,” Nick Ryman-Tubb, CEO at the Institute of Financial Innovation in Transactions & Security, observed in the same session.
Similarly, artificial intelligence should work in tandem with existing cybersecurity defences, suggested Kalyan Veeramachaneni, principal research scientist at the Massachusetts Institute of Technology’s Laboratory for Information and Decision Systems, who gave a primer on the machine learning and AI before exploring the latter’s use in information security defence and response on Thursday’s ‘Cyber 101 – Artificial intelligence for information security’ session.
Time to fulfil potential
Such caveats do not of course render machine learning and artificial intelligence ineffective in compliance or security, but it does perhaps place a potential limit on what even its supporters might consider its impact to be. Indeed, some observers with longer memories noted that chess grandmaster Gary Kasparov’s 1997 defeat by a computer (IBM’s Deep Blue) was heralded as a breakthrough for AI and led to an outpouring of prophecies about the growth of intelligent machines which nearly 20 years on remain unfulfilled.
When discussing the use of AI in compliance, speakers were quick to recognise that the regulatory burden on banks and other financial institutions continues to grow. As such, anything that can have a material impact in improving efficiency deserves serious consideration. “Large global banks like UBS face 40,000 alerts every year of changes in regulation that might affect them,” acknowledged Ermotti. “Even employing 10,000 people in compliance, as some banks are said to do, may not be enough.”
The last word should however go to IBM’s Rometty and her assertion that, when the history of the present time is written, it will be seen as the era of data. Those companies that have done most to collect, analyse and gain insight from that data will be the businesses that are remembered, along with the technology companies they relied on to help make it happen.