Singapore Commands Attention as the Leader of Ethical AI

Published On January 28, 2019

With a reputation for openness and support, Singapore’s financial regulator is the darling of the fintech world. Radar tracked down the top data scientist at the Monetary Authority of Singapore to find out how it is tackling the big questions that surround AI.

Singapore’s emergence as a leading global innovation hub was no accident but was the result of a visionary plan to embrace new technologies and automation in response to slowing growth and an aging population.

The kingdom’s capital markets have been one of the biggest winners in this digital-first push, attracting major global FX dealers, as well as Asian emerging market currencies, eager to compete in a market where the smartest use of technology can give firms that vital edge.

According to the latest triennial survey by the Bank for International Settlements (BIS), the average daily FX turnover volume in Singapore tops US$520bn. This ranks Singapore as the third largest FX center globally, and the largest FX center in the Asia Pacific.

The Monetary Authority of Singapore (MAS), which oversees regulation of the finance sector, has been key to this transformation; it has been extremely vocal in its support of new technologies and has advertised its openness to attract new business.

“AI and data analytics have the potential to transform the financial industry,” said David Hardoon, chief data scientist at the MAS.”

The regulator has encouraged the use of blockchain, cloud adoption, data analytics and Artificial Intelligence (AI), providing a safe space for firms to road-test their tech without subjecting them to full regulatory demands and licensing or sanctions, while testing and deploying the technologies in a variety of parallel projects.

However, a growing number of data-related scandals through 2018, from now run-of-the-mill hacks and cyber-incidents to the Facebook-Cambridge Analytica scandal, have cast AI and big data for now in a more negative light.

In the same month, the world’s largest social network had billions wiped off its stock value, and MAS lost no time as the first global regulator to announce that it was creating a framework for the ethical use of AI and big data in financial services.

It formed the 10-member Fairness, Ethics, Accountability and Transparency (FEAT) Committee in April, and tasked the group with developing the guide for responsible and ethical use of AI and data analytics by financial institutions.

“By partnering with key industry stakeholders, we have benefited from the industry perspective and experience,” Hardoon said. MAS will then look to shape “the right conditions” for firms using AI and data analytics, “based on the principles of fairness, accountability, and good governance”, he said.

“We want to encourage the financial institutions to think about responsibility before things go wrong and take responsibility and be accountable for their decisions that result from their use of AI and data analytics,” Hardoon said. “Financial institutions which choose to use AI or data analytics in their business need to be aware of the underlying risks and take the necessary steps to mitigate these risks.”

Perceptions as to what separates ethical from unethical, especially in business, differ wildly depending on who is asked, however, Hardoon insists this is not something the regulator will take a view on.

“It is not MAS’ intention to set ethical boundaries for the financial sector,” he said. “The FEAT guide would serve as an advisory to the financial institutions and fintech firms in developing their own internal AI governance standards.”

Every firm has its own ethical standards, he said, which may be expressed through company values or codes of conduct.

“When using AI and data analytics, firms should ensure that the use of these technologies are subject to the same ethical standards as expected of human employees in carrying out the same tasks,” he said. Firms will be encouraged to use the FEAT as a guiding principle.

One of the most crucial aspects to best practice, Hardoon said, is having an understanding of the technology beyond human input and review. Firms cannot rely on so-called ‘black box’ solutions where algorithms execute decisions in a closed system.

“It is important to strike a balance on the appropriate level of transparency,” Hardoon said. “While more transparency is useful in encouraging confidence in, and understanding of, AI and data analytics, we need to also be mindful to prevent undesirable outcomes such as deliberate attempts to exploit algorithms.”

There will be no excuse for firms who try to avoid taking ownership of any negative outcomes from their software, and the punishments available to the regulators will also be familiar to US and UK firms.

Singapore’s government is creating a Deferred Prosecution Agreement regime for corporates. The legislative tool allows prosecutors to deal with alleged criminal conduct in a way that avoids formal prosecution over a period of time. At the end of that period, if the defendant has complied with the obligations, such as a financial penalty, amendment of compliance programmes or payment of compensation, then no prosecution is brought.

The Kingdom is also following the global regulatory trend of strengthening accountability, and earlier this year also announced the first steps of its own version of the Senior Managers Regime, which is now fully in effect in the UK.

The Guidelines on Individual Accountability and Conduct were drawn up as a response to growing incidents of misconduct and egregious risk-taking in the financial industry both globally and in Singapore – an unfortunate by-product of its increased success and importance as a financial hub.

It hopes to shift thinking away from the box-ticking compliance mentality and transform the industry from a ‘what is legal?’ standpoint to ‘what is right and ethical?’

“MAS can take regulatory action against a financial institution if it is found to be in breach of any of MAS’ rules and regulations,” said Hardoon. The safeguards included in the FEAT guide would also serve as an advisory to financial institutions and fintech firms on the dangers inherent in letting their AI projects run wild.

“Having said that, given the speed with which algorithmic tools and practices are transforming, we will continue to monitor developments and calibrate our rules as necessary to prevent abuses of these technologies and maintain the integrity of our financial system,” said Hardoon.

While financial institutions in Singapore are just as bound by legacy technology as their Western counterparts, there is a nimbleness about the banks, broker-dealers, hedge funds and asset managers not seen elsewhere, said Hardoon.

“Financial institutions here have started their tech journey earlier than many other places,” said Hardoon. “For instance, we have financial institutions with established data science and infrastructure teams that have reported success with collating their data from legacy systems into central repositories such as data lakes,” he told Radar.

The advancement of data lakes allows the banks to explore AI technology and develop regulatory technology, aka ‘regtech’, for transaction monitoring and Know-Your-Customer systems. Most of these AI projects are at an early stage of development, Hardoon said.

“There are certainly a number of advantages to having a more modern industry not bound by legacy tech,” he said. “For supervision, MAS is able to collect more data points from their current business processes if the financial institutions have better monitoring systems, as it improves on the accuracy and precision of our analyses.”

For data collection, cutting-edge technology gives MAS the opportunity to collect and explore granular data from financial institutions. By encouraging the industry to use AI, it can then tap into that knowledge and experience, allowing it to be the regulatory forerunner and thought-leader in data analytics-related issues, such as the ethics and usage of AI in finance.

The 2018 Global Innovation Index ranked Singapore as the most innovative economy in Asia, and fifth in the world, despite spending hundreds of millions less in R&D than other regions such as the US, which lies sixth in the overall chart.

Business sophistication and a supportive regulatory environment were the main reasons for Singapore’s advancement, and while MAS continues to blaze a trail with its approach to shaping the AI debate in financial services, everyone is taking notice.