A history of global financial services will almost certainly record the past decade as its most tumultuous ever. In just 10 short years we’ve had a global financial crisis that claimed institutions once thought too big to fail. Disgruntled customers have been wooed away by emerging non-traditional players offering lending, payments, deposit-taking and many other financial transactions.
Regulators have taken unprecedented steps against financial institutions and their leaders, holding them to account for poor service, excessive charges, and even – such as the case of Iceland – jailing them for damaging the national economy.
Demands for ever-more capital to buffer any new systemic risks are eating into margins. And stringent rules have come into play around data privacy and money laundering, just two issues that are creating a huge compliance burden.
These are just some of the many forces that have created a perfect storm in the financial services industry. The fallout has included rising costs, painful fines, steep remediation payments, declining returns, growing regulatory and media scrutiny, reputation damage and unrelenting pressure to be more transparent, more competitive and more responsive to increasingly demanding customers.
And as if this weren’t enough, financial institutions must keep current with the very latest technology to ensure systems and service delivery meet the requirements of both customers and regulators. The halcyon days of financial services are, it’s fair to say, a distant memory.
Fortunately for the financial services sector, the rapid evolution of artificial intelligence and automated machine learning is offering institutions viable ways to tackle these many challenges.
The appeal of AI is evident in the hefty budgets being dedicated to integrating it into the large and complex technology systems and people operations that characterise the financial sector. The International Data Corporation estimates that global expenditure on AI will reach US$35.8 billion in 2019, up by almost half on last year’s spend, with retail and financial services being the biggest spenders. The banking sector alone is expected to allocate US$5.6 billion to AI-enabled solutions such as automated threat intelligence and prevention systems, and fraud analysis and investigation systems.
So promising is the impact that we’re seeing AI deployed in almost every aspect of financial services. Processing customer applications for loans, for instance, can be expedited by new AI-powered functionality that cuts processing times from weeks to minutes. One of our clients, Harmoney, now automates credit risk assessment, using machine learning on 300,000 loan applications to to calculate defaults with such accuracy and speed that there’s been a tangible impact on bottom-line margins. It’s fast, reliable and affordable.
Where traditional and mainstream financial institutions – banks, assurers and so on – have an advantage over newcomers and fintech is their data repositories. They’ve been gathering big data for decades. This wealth of information injected into AI offers them the ability to refine products and services to an enviable granularity. The bigger the data resource, the more effective the AI.
AI has also become a substantial tool in ferreting out illegal transactions, money laundering and the movement of cash around the world to finance terrorism, counterfeit trade and trafficking. One of our Canadian clients, a large bank, integrated automated machine learning to deliver an exponential improvement in the review and detection of suspicious transactions. This allowed the investigations team to process many more transactions, and help stamp out criminal financial activities.
In Australia, even the regulators are embracing AI to help drive better compliance in the financial services sector. The Australian Securities and Investments Commission is looking to AI to help root out misconduct and improve regulation. As reported by the Financial Times, this is just one facet of a “wider global shift towards the use of machine learning technologies and AI in the regulatory, compliance and finance sectors.”
Various AI technologies are being considered by Australian authorities to ensure that customers receive appropriate financial advice. They’re testing technology that scrutinises documentation and analyses recorded sales conversations.
Moves like this promise a new dawn for the financial services sector. Long beset by a never-ending litany of challenges, the industry has realised that AI and automated machine learning offer the means to tackle them head on in a way that ticks all the boxes. New AI solutions not only improve efficiency, costs and accuracy in many financial operational processes but can also meet the high compliance requirements of regulators. By replacing error-prone human activities with binary processes that leave no room for interpretation, observance and enforcement of the rules becomes standardised.
While AI-powered processes can clearly support financial institutions with carrying out operational tasks, compliance and risk management, there’s also significant potential to enhance front-office activities. The mantra of good customer experience has become the defining benchmark in the financial sector, with metrics like net promotor scores becoming as important as returns on equity and investment. Using AI, financial institutions can improve the way they use customer data to deliver refined products and highly segmented marketing. This offers the potential for greater cut-through on outreach and selling activities, while also providing customers with the financial products they want.
AI is not without its own challenges. It is very complex and requires detailed knowledge of machine learning technology, data platforms and the business to which it will be applied. Consequently, the availability of people with these skills in such a nascent field is low. This makes them expensive and hard to retain. Where these obstacles can be overcome is the application of automated machine learning, which effectively uses AI to build AI.
It automates common challenges such as fitting models to data, testing models for performance and accuracy, creating documentation for auditors and regulators, deploying and monitoring models and retraining them when they become old and stale. This empowers a broader range of employees to build AI, making it more affordable and accessible, and freeing up data scientists to be more efficient.
We’re very excited by the opportunities to enhance the entire spectrum of financial services activities with automated machine learning. More than that, it can also help pave the way to restoring financial institutions to the role of valued and appreciated players in our economies.
Tim Young, general manager APAC, DataRobot
