- January 23, 2020
- 4189 Views
The race for ideas to integrate AI into financial services
By Ellen Hardy
By 2022, banks will be spending as much as $12.3 billion on AI and cognitive technologies with the race underway to integrate the latest capabilities into financial services
- Globally, finance is believed to be outpacing all other industries when it comes to introducing AI, with Chinese banks and fintechs leading the way
- There is a strong trend for banks entering joint ventures to make the most of AI, but they learn from the mistakes of the big tech unicorns and communicate their progression
- While a lot of exciting innovation is occurring, regulation surrounding data protection and privacy, and an increased focus on liability will present ongoing concerns for financial institutions
In an era of low interest rates and wavering market confidence, putting in place the right efficiency measures is as critical as ever to banks’ bottom lines. Add to that the ongoing disruption from fintechs, as well as the opportunity to gain insight into the future of workforces and everything points to tech solutions in artificial intelligence (AI), machine learning and even robotics.
Chinese banks leading the way
The view in the global technology community is that Chinese banks and fintechs are leading the way when it comes to machine learning and building a data science workforce. At the same time, many European and US banks are slightly ahead in terms of deployment — but Asia is moving fast and expected to overtake the West in the coming years.
Leading the way is China CITIC Bank, which developed its ‘brain platform’ in conjunction with Tsinghua University. The project has some 15 machine learning models applied in different parts of the bank, marketing, automation, AML and anti-fraud businesses. The project includes construction of an artificial intelligence platform and a blockchain-based trade finance business model in partnership with Bank of China and China Minsheng Bank.
For CITIC, the opportunities are in targeting individual needs of different operations within the bank. “The platform aims to intelligently empower each business line,” a spokesperson explained about the platform’s self-developed architecture.
“The platform implements intelligent full-process services from artificial intelligence model training to application deployment. The platform is easy to use and tailored to fit the specific needs of banking business, provides real-time interfaces and batch processing interfaces,” the spokesperson added.
With CITIC’s AI operations firmly in place, it is turning its attention to aiBank, its digital banking joint venture with Baidu. aiBank spent 2019 seeking strategic investment partners for as much as $1 billion (RMB 7 billion), as well as announcing a partnership with credit card company 51 Credit Card to create a “state-level innovation trial” fintech ecosystem. aiBank will reportedly focus on smart risk control and big data applications, as well as finding solutions in more traditional fields such as consumer finance, credit payments, escrow operations and fintech.
ICBC, on the other hand, has its smart banking construction scheme focused on improving services for its more than one billion retail customers. The bank is concentrating on intelligent customer service and building an operation support system that integrates all channels and prioritises customer experience, using tools such as voice bots, seamless connection across AI and manual platforms, and scenario-embedded smart Q&A.
Making rapid advancements while cutting through hype
The most advanced players are already way ahead in their vision of the tech future.
None more so than China’s Ping An Group, which is forging a distinct path with a comprehensive vision. Chief innovation officer Jonathan Larsen recently revealed that in spite of the $1 billion innovation fund under his watch, the bank has “no illusion that we have any monopoly” on transformational technology.
“We have found that the capabilities it has acquired in areas like visual artificial intelligence, natural language processing and the ability to create integrated cloud services that can be sold as verticals to other financial institutions,” he said.
While Ping An is investing in tech integrations, connect technologies and building a proprietary tech arm that is a provider to other banks, part of Larsen’s remit is to use the fund to scout businesses who have technology that is ahead of the game. Larsen shared that they are finding broad applicability to these concepts, which are also have significant horizontal scale.
“Smart city is one of our five ecosystems that we’re focused on. We provide a single app that allows every citizen in Shenzhen to access pretty much every government service with the same ease of use as you can access an online service,” Larsen explained. “What we’re finding is that a lot of our analytics, AI and blockchain solutions can be used for government recordkeeping, property registers, traffic management and pollution management.”
In 2018, Ping An’s IT capital expenditure was up by 82% year-on-year, and it increased its technology staff by over 44%. But Larsen stressed that a big part of his remit is applying a healthy skepticism to cut through the hype to find real capabilities and technologies with a unique business mode.
It can be argued that China is leading the way in AI in finance due to an early appetite to merge tech and finance, such as Tencent-led WeBank and Ant Financial’s Alipay, which dominate China’s third-party mobile payments sector, estimated to be worth around $7.17 trillion (RMB 50 trillion). The two firms specialise in building full psychographic profiles of customers through personal, social, financial and commercial data.
WeBank recently announced the establishment of Retail & AI Joint Laboratory with retailer BKK Group, to “break through the difficulties of traditional retail industries to help the increase of the economic growth.” The firm will focus on three core pillars: smart labor, smart operation analysis, popular products forecast, with bold aims such as to “improve 50% of efficiency in the cashier position”.
The rest of Asia is taking advantage of late arrival
Other Asian nations, which are not yet as advanced as China, are starting to pick up on robotic process automation, chatbots and machine learning in credit analytics.
Some firms, such as Myanmar’s Yoma Bank, have showed that there are advantages from entering late into the market and leaping ahead, taking decision analytics platform with Experion and building a leading credit analysis by jumping on the latest advances in machine learning.
Deepak Sharma, chief digital officer at India’s Kotak Bank, said that his firm began its AI journey two years ago. The brief is to reduce cost and improve efficiency, not simply in man hours (although they’ve saved around 15,000 in 2019) but through reducing error rate and decreasing turn-around time.
Kotak can be viewed as a fast follower of international trends, looking at productivity, personalisation, and fraud detection like most leading banks. But they are also tackling unique challenges, such as the diversity of languages in India. Kotak is the first Indian bank to do voicebot on interactive voice response and in two languages, so far, that cover 70-80% of the population, with answers in real time.
“We have a couple of million customers communicating with us via WhatsApp, and we are building full integration into that,” Sharma added. “In terms of voice and vernacular, the challenge of being in a country with so many languages is that we cannot train employees to deal with them all, so we are building automated video generation and language support, which will pick up a lot more in the next 12 months.”
Sharma noted that talent in the field of AI and machine learning is still at a premium, but banks in India have been successful in attracting and retaining top talent in the AI field, second only to the big tech firms. This is due to their large and rich data sets which are getting used in relevant and impactful use cases. He added that the challenges in keeping pace are modernisation of infrastructure and maintaining the skills and knowledge to put together the tools for the future.
“We think that the next generation of AI will be looking at things including roboadvisory, overall customer engagement level, hyper-personalisation, both direct to customer and employees who serve customers in real time. In terms of risk management, there are inroads to be made in alternate credit score, underwriting and creating new products, and risk pricing using alternative data that is non-bureau,” he added.
Regulatory and privacy issues remain a critical concern
The Australian Human Rights Commission recently proposed that legislation should be introduced so that a person is ultimately responsible and legally liable for the decisions made by AI. It’s just one hurdle that banks will have to consider as AI comes more into public consciousness, and they face regulatory responses from countries across the world.
In Europe, regulation of data use has become a hot topic. Gorkem Koseoglu, chief analytics officer at Dutch-based ING, said that its newly-established global analytics department is turning its attention to wholesale banking applications for AI, while strongly factoring in a changing regulatory landscape.
“Data is the fuel of all AI, and as such, ethics discussions have accelerated. The AI community is already working on modules to explain decisions taken by complex models,” he said.
Koseoglu shared that ING has been working on solutions to augment the decision making process by leveraging the power of big data and analytics with its cutting-edge AI tool Katana, which allows traders and investors to quickly make sound decisions when buying and trading bonds. It can aggregate vast amounts of data from multiple historic and real-time sources to predict the winning price and identify the most promising trades that might otherwise have been missed.
He further noted that as programs such as Katana become increasingly cross-industry and internationally focused, it is important to stress the compliance principle of ‘same services, same activities, same risk, same rules and same supervision’ across geographies, as well as building alignment across regulators.
One example of the challenges of operating AI in different jurisdictions is deep learning algorithm provider SenseTime, which was added to the US government blacklist of Chinese companies in October. In spite of this ruling, the world’s first AI unicorn is seeking to position itself as a responsible player in the field.
“As the researchers and creators of potentially powerful technology, we recognise that companies such as ourselves, have an important role and duty of care in data privacy and protection,” a SenseTime spokesperson said. Still, SenseTime vice president Leo Liu highlighted the tension for AI innovators, saying that “you must push the technology to the limit,” and that the everyday conveniences of AI will win out over concerns about personal data collection.
Prema Varadhan, chief architect and head of AI at banking software company Temenos, said that it is critical that financial firms learn from the mistakes of the big tech unicorns and communicate their progression when it comes to letting machines do the driving. A key issue in the evolution of machine learning is moving from operating in ‘black box’ to ‘explainable AI’.
“The biggest barrier to AI in banking is the lack of transparency into how decisions are made in order to be able to show regulators and auditors that the decisions are not biased,” she said. “For example, Apple recently made headlines for allegations of discrimination against women who applied for the Apple Card, with some men receiving 10-20 times higher credit than their wives.”
Varadhan added that such situations may result to the emergence of new paths that better empower consumers, in the crypto mold of autonomous and decentralised organisations that allow users to take control of their finances and data through open source platforms.
Ultimately, while financial institutions are set to benefit massively from adding behavioural data to personal information they hold on customers, they will face increasing burdens when it comes to how they access and monitor the masses of data involving a consumer’s digital footprint and how it is stored and shared. The industry will also face greater calls for transparency around how it is analysed — not to mention where the responsibility for breaches lies, especially within the complex world of fintech joint ventures.
Permanently altering the banking landscape
In some respects, the race to implement AI is a tale of two banking sectors. Early evidence suggests that cutting-edge fintechs and big banks are in a technological arms race to develop and license their own AI programs. For many other small-to-medium sized banks and less agile firms, they will be heavily reliant on vendors for innovation and technology strategy.
To keep pace, financial services firms may wind up creating their own large tech subsidiaries and become, in effect, a new generation of tech firms. This could lead to further consolidation of clients by the powerful financial institutions who have heavily invested in technology, and lucrative gains for any fintechs who develop truly innovative programs and are willing to sell their ideas.
But while banks are seeking bragging rights when it comes to the best and smartest technological advances, it won’t all be smooth sailing. The need to manage public and regulatory perception of AI, especially surrounding privacy concerns, will increase alongside the rapid growth of the next wave of technological solutions.