By Raisa Ostapenko
From the discovery of an eighth exoplanet circling distant star Kepler-90 to Microsoft’s ambitious new project to map and decode the human immune system, Artificial Intelligence (AI) and especially its subset Machine Learning – a field of Computer Science that gives computers the ability to “learn” based on past data – have seen a promising boom in application across numerous industries over the past decade.
The investment banking sector is amongst those. Opportunities abound, from the basics – like relieving employees of time-consuming, menial tasks, such as cleaning their inboxes or resetting passwords – to more consequential services – such as fighting money laundering, rogue trading, and cybercrime.
Even more, the technology promises to protect employee rights through unprejudiced recruitment.
Over the past few years, banks, from HSBC to Credit Suisse, have been partnering with financial technology companies to integrate AI into a wide range of operations.
These ultra-efficient “intelligent agents” optimise data analysis and resource distribution, reveal complex patterns with facility, and have generally proven themselves capable of doing in seconds what at one time often required teams of thousands of humans and sometimes thousands of manhours.
In automating superfluous activities, such applications also help banks serve their clients with greater efficiency, more personal attention, and greater security.
More likely than not, AI is involved each time a bank blocks a debit card that a customer used in a way that he or she does not do so habitually.
Some companies use voice-recognition to determine whether people calling a bank are indeed the customers they claim to be.
First direct, for instance, uses telephone technology that measures 140 different aspects of a customer’s voice in just around 2 seconds, updating itself for natural changes, such as those resulting from aging.
Anti-money laundering and misconduct
Of growing importance and interest is the application of AI in compliance and anti-fraud sections of banks.
Just recently, in April 2018, HSBC teamed up with start-up Quantexa to use AI-based anti-money laundering technology in a bid to clean-up its act on the back of a string of misconduct scandals in recent years, including laundering the money of drug cartels in Mexico.
Other banks may have turned to AI prophylactically, i.e. in order to avoid future scandal from as yet unidentified conspicuous behaviour.
Two years ago, for instance, Credit Suisse launched a joint venture with big data start-up Palantir Technologies – known for its work for security agencies such as the FBI and the CIA – to tackle rogue traders – authorised traders who make unauthorised trades.
The bank started working with Palantir after equities trader Kweku Adoboli at Swiss competitor UBS lost over US $2 billion in rogue trading.
The joint project – Signac – learns patterns of lawful human behaviour in finance by analysing both internal and external data and flags any anomalies it identifies in the bank’s operations.
This efficient tool has since been applied beyond trading to international asset management.
Analysts, however, indicate bank bosses are often faced with some tough questions, such as deciding to what extent AI should replace human capital.
What path they choose could impact various areas of the bank’s network, including to what extent anti-money laundering (AML) processes are handled by machines or by humans.
‘Reservations’ about AI
Whilst investment banks are clearly seeing the benefits of AI across the board, some believe that they have reservations about embracing this new technology in applications that go beyond simple automation.
One reason for this may be the importance of human capital to banks, which tend to want to seek out and cultivate the best and sharpest talent fresh out of the world’s top universities.
“If we’re honest, machine learning is synonymous with job loss,” said Alexander Fleiss of start-up investment management firm Rebellion Research in an ai in industry podcast on Tech Emergence.
According Fleiss, machine learning and resulting automation is a big enough threat to job security to make many people in “paper-pushing” positions redundant in the next five years, which could possibly affect compliance departments.
Striking a balance
Fleiss’s assessment, however, may be premature.
In fact, replacement may not be the logical outcome of automation.
Instead, according to “Turning automation into Intelligence,” a report by Accenture, investment banks may intelligently transition towards a hybrid workforce, which, “part human, part machine … unleashes each element to do its best work.”
“Companies that adjust their organisation and culture to incorporate intelligent automation as co-workers, rather than people replacements,” the report concludes, “could reap important rewards: more reliable performance and insight, extension of services to previously unprofitable markets (such as lower-end retail markets and smaller institutions) and continuing cost reductions.”
Banks realise that AI is a critical key towards a successful and competitive future.
Furthermore, advancements in technology have not gone unnoticed by online offenders and, “as malware and other technology powering online fraud becomes increasingly sophisticated,” read a Credit Suisse article, companies have no choice but to seek solutions in AI and to do so with the least damage to the financial market’s structural integrity, at least when it comes to employment.
It may be that a challenge to integrating machine learning into fighting financial crime is, therefore, nothing more than a pre-emptively anticipated threat of regulation of AI applications in banking, as companies already rush to comply with increasing complex anti-fraud regulations.
In November 2017, the Financial Sustainability Board – an international body that monitors the global financial system – said that, AI, if unmanaged, could lead to serious complications, such as a potential dependency amongst banks and insurers on the few tech firms specialising in machine learning.
Ultimately, it appears investment banks will remain in a strong position if they embrace the benefits of AI and machine learning, all the while sustaining their commitment to human capital, embracing the hybrid workforce, and retaining a sense of autonomy in AI application by forming their own internal tech teams, instead of surrendering all intelligence innovation to outside players.
About the author: Raisa Ostapenko is a writer and political commentator. Currently based in Paris, she spent three years as a journalist in Moscow and has written for numerous outlets including the BBC.