New report finds attitudes at board level changing toward AI as banks prepare to overhaul antiquated systems.
Unless you have been hiding under a rock for the last 10 years, you have probably been told that artificial intelligence (AI) is going to transform banking or take your job or end the world. Or all three of these outcomes.
For compliance professionals, wondering if their own firm will embrace the technologies that promise to unburden their roles, a new report by consultants at TABB Group sheds some light on the thinking of the top chief executives and budget holders at Wall Street’s biggest banks.
The research by fintech veteran Terry Roche has found that the industry’s ‘Tier 1’ names are finally prepared to invest in AI platforms, and there is a growing commitment to increasing the headcount for data analytics personnel.
Roche, head of fintech research at TABB, told Radar that firms who are resistant to adoption of newer technologies, for whatever reason, have their heads in the sand and they are likely to be punished “sooner rather than later”.
“Everyone is touched by machine learning and AI, and it’s only going in one direction.”
“If you are in capital markets or investment and you want to run an efficient organization, and would like to be able to gain insight and remain competitive, then you should pay attention.”
The report is titled Enhanced Bankers – The Impact of AI, and it involved extensive interviews with senior executives at the largest US banks on their projects to leverage machine learning and AI in order to obtain greater insight, along with a survey of the market to understand the broader trends.
Roche and his team found that compliance was the business role expected to be aided the most, with a third of respondents identifying it.
The lack of love for compliance, and the common reluctance of boards to spend on it, is a major roadblock to adoption.
A report from Juniper Research in October 2017 found that spending on regulatory technology will grow by an average of 48 percent per annum over the next five years, rising from $10.6bn in 2017 to $76.3bn in 2022.
It’s no surprise why; banks have been fined $340bn since the financial crisis, according to numbers released in February.
Keefe, Bruyette and Woods, the investment bank, compiled the list. Bank of America leads with $76bn in fines; JPMorgan Chase has been fined nearly $44bn. Many others have been hit for more than $10bn, and just 13 banks make up 93 percent of the total.
So far those banks have spent more than $1bn a year on compliance-related costs, and as an industry, more than $27bn a year. These sums often comprise more than 10 percent of most banks’ operating costs.
Compliance staffing has also doubled since 2012, and when compliance numbers match or dominate front-office staffing levels, the increased growth and spend is not sustainable.
Much of that is down to the legacy technology that firms use, characterized by out-of-date or neglected lexicons, and inferior surveillance technology, that does little to reduce false positives.
The respondents to Roche’s report say resources are slowly being diverted to better and more targeted data storage and management, and the advanced intelligent analytics needed to get ahead of problems.
“Regulatory compliance was always fighting the last war,” Roche said. “It would root out malfeasance and create swimming lanes for conduct, but it never stopped the next, new, novel malfeasance. I think that is a little bit different now.”
With machine learning applied to patterns, Roche said compliance can truly get a bigger picture, and understand whether something irregular is happening or not.
“We are also beginning to see so much regulation now requires advanced analytics to report to the regulator, particularly against risk-based compliance,” Roche added. “This is likely to result in a blurring of the lines of the risk tools that are used around trading or investment, with that doing the reporting. It’s likely to be a machine, as it’s not something humans can keep abreast of given the pace and the breadth of the data.”
Much of that data is unstructured, which must be “cleaned” before it can be put to use in a database. This can be from email communications, market commentary, even news articles, or call notes inside a Client Relationship Management (CRM) tool.
The information is typically archived in bank IT infrastructure and usually searchable only by date or keyword; it’s considered “unstructured” because it hasn’t been dissected so that typical statistical analysis tools can consume it.
“In fact, manually cleaning and classifying keywords from millions of email messages, for example, might be simply impossible,” Roche states.
That is until modern machine learning algorithms crunch the existing data, in bulk, without a separate pre-processing step.
Structure and schema are then inferred from the raw data using advanced techniques, without a person having to do manual analysis.
“This moves analytics from being hypothesis-driven to data-driven,” Roche said. “Data has structure and relationships to other data from which AI may derive meaning. Those structures and relationships can provide insight if the data can be examined holistically.”
The TABB paper found that the sell-side worked out early on that problems often start with the haphazard methodology behind data collection and storage, as they are pushed into a diminishing corner by the twin pincers of competition and regulation.
“A lot of the larger firms have been forced by these two areas to get their arms around enterprise data management, and have a much more deterministic view on how you digitize information,” he said. “They are in a much better position to leverage machine learning and AI.”
On the other side, Roche said, the buy-side recognizes the advance of AI and machine learning will be invasive and impactful but that they cannot leverage it until they get their digital data house cleaned up.
“We are seeing different levels of maturation around data management,” he said. “Point at any regulation and I can tell you how you are not going to be a happy camper unless you have your data house in order.”
Roche’s team found a 360-degree view of all data within the investment cycle and capital markets “is clearly the direction the market is taking”.
“There is little doubt that much of this effort will help to enhance the efficacy of the banker or investment professional,” said Roche. “There is also little doubt that all roles will be affected with some level of automation introduced to the workflow.”