DBS has had to navigate significant hurdles in its years-long effort to adopt artificial intelligence (AI), during which it has recognized that success goes beyond simply discovering training models.
Data, in particular, is proving to be a huge barrier, according to DBS Senior Analytics Officer Sameer Gupta. In 2018, Bank of Singapore embarked on its journey to leverage AI across four core areas which include Developing Analytics Capabilities, Data Culture and Curriculum, Elevating Data Skills, and Data Enablement.
Also: The next big AI threat may already be lurking on the web
“The vision here was to use data to bring greater benefits to the organization,” Gupta said in an interview with ZDNET. To do this, he said, the bank recognized the need to make access to AI pervasive across the company as well as deliver economic value from AI. There is also a need to continuously reduce the cost of providing AI solutions.
Efforts were directed toward developing use cases and the right talent, including machine learning engineers, and building a data culture that encouraged all employees to constantly think about how data and AI can help their work. It meant providing a training program that instructs employees on how and when to use and not use data.
Also: How to prevent OpenAI’s new AI training web crawler from ingesting your data
The Bank has worked to create the infrastructure to facilitate its adoption of AI, including a data platform, data management structure, and data governance. It has implemented a framework against which all use cases of its data must be evaluated. As PURE formulated, this is based on four principles – meaningful, unsurprising, respectful and interpretable – which DBS believes are essential to guiding the Bank in using data responsibly.
Its data platform, ADA, acts as a single central resource, enabling the bank to better ensure data management, quality, discoverability and security.
Today, over 95% of the data deemed useful and essential to facilitating AI-assisted DBS processes can be discovered on the platform. The platform holds more than 5.3 petabytes of data, comprising 32,000 data sets that include videos and structured data.
But getting to that point has proven to be a monumental task, Gupta revealed. In particular, organizing data and making it discoverable takes titanic work, involving mostly manual and human expertise. Painstaking hours have been spent defining metadata, with the tools to automate such tasks sorely lacking.
Also: Nvidia boosts its Grace-Hopper chip with faster AI memory
He added that the bank used many applications, each of which contains the data needed to support artificial intelligence initiatives.
With data spread across different systems, he noted that “a lot of heavy lifting” is needed to bring datasets onto a single platform and make them discoverable. He said employees should be able to extract the data they need, and the bank had to ensure that this was done securely.
Today DBS runs more than 300 AI and machine learning projects, which it says generated S$150m ($112.53m) in revenue last year and saved S$30m ($22.51m) in avoiding risks, for example, from Improve credit control. These AI use cases cover a range of functions, including human resources, legal and fraud detection, according to Gupta.
The bank’s AI initiatives are on track to drive more economic value and cost avoidance benefits this year, doubling to S$350 million ($262.56 million). It aims for this figure to reach S$1 billion (US$750.17 million) in the next three years. Singapore’s largest bank, DBS currently has around 1,000 data engineers, data scientists and data engineers.
There is no “magic bullet” with AI adoption
Asked if he was exploring the use of generative AI, Gupta confirmed that the bank already manages more than 10 pilots, but stressed that it was still early days. He said the various teams, including marketing, sales and IT, will need to have more conversations over the next few months to better understand from these tests how generative AI can benefit the bank.
Also: Microsoft’s red team has been monitoring AI since 2018. Here are five big insights
He added that it also needs to ensure that the use of these AI applications continues to adhere to the PURE Principles and Singapore’s FEAT Principles that guide the industry’s use of AI. He said other known risks such as hallucinations and copyright violations would also need to be assessed.
DBS currently operates 600 artificial intelligence and machine learning algorithms, which collectively help power interactions with its five million customers across the region, including China, Indonesia and India.
Its use of 600 AI models, however, is insubstantial, said Gupta, who instead emphasized the goal of achieving optimal efficiency and accuracy from the fewest number of AI models.
He highlighted the misconception that the model itself is everything, and noted that it actually plays a small role in ensuring that companies benefit from the use of AI.
Also: ChatGPT Plus can mine your company data for powerful insights. Here’s how
Instead, they need to work through all the technical elements, which should include building in mechanisms to monitor their use of AI and continually collect feedback to identify areas for improvement. He will ensure that the organization learns from its application of AI and makes changes where needed, including in AI models and operational processes, as it works out kinks and fills gaps.
“You have to persevere to get the full benefit,” Gupta said. “There is no magic bullet.”
Asked if DBS was using artificial intelligence to better anticipate disruptions, such as the ones it experienced last year, he said the bank was working out how it could work better, including leveraging data analytics. Noting that many factors can cause spikes in demand, he said there is potential to take advantage of artificial intelligence, for example, in operations to detect anomalies and determine the next course of action.
He couldn’t comment specifically on the outage, but said a special panel made up of four members of the bank’s board of directors is leading a full review of the company’s technology resilience. He said that outside experts have been called in to assist in the review, adding that more details will be provided once this is complete.
Also: train AI models using your own data to reduce risk
Last month, it was revealed that human error was the cause of the DBS outage in May but had nothing to do with the outage in March. Singaporean Minister and Minister Responsible for MAS, Tharman Shanmugaratnam, said in a written parliamentary response that the bug was found in software used to maintain the system and had resulted in a “significant reduction” in system capacity.
This affected its ability to process online and mobile banking, electronic payments, and ATM transactions, Tharman said, citing the bank’s initial investigation.
Funds to help the sector adopt AI
Singapore said on Monday it has allocated 150 million Singapore dollars ($112.53 million) over three years to support the financial sector’s efforts to innovate through the use of technology.
The Financial Sector Technology and Innovation (FSTI 3.0) blueprint will continue to facilitate capacity development and adoption in key areas such as artificial intelligence and data analytics as well as regulation technology or regtech. Specifically, the Monetary Authority of Singapore (MAS) which is the industry regulator will look at promoting the adoption of artificial intelligence and data analytics among smaller financial firms.
FSTI 3.0 also includes new pathways whereby funds will be expanded to corporate venture capital entities and ESG (Environmental, Social and Governance) projects. MAS will also hold open calls for use cases in emerging technologies, such as Web 3.0, with grant funding going for trials and commercialization.
Also: The AI gold rush makes basic data security hygiene critical
For DBS, the focus is now on ensuring its AI projects can scale, and access remains spread across the organisation, Gupta said.
“We need to make sure that we industrialize how AI is developed and deployed to the bank, so that we can reduce the effort it takes to implement it. You can’t do that if every use case is implemented in an ad-hoc way.” .
He also stressed the importance of ensuring continuous measurement of artificial intelligence, so that the bank can determine if it is achieving positive results. “We need to ensure that there are benefits for both employees and customers,” he added.