Modern risk management is increasingly data-driven, proactive and careful in approach. However, current approaches to risk management may be leading banks and other lenders to miss opportunities that appear risky due to a lack of conventional data. As a result, risk management may be slowing down financial processes and driving up risk management costs.
Machine learning, which can be used to uncover subtle relationships in vast datasets, process new information in real time and automate processes, has quickly emerged as a valuable experimental tool in risk management.
Here is how machine learning is being used to improve risk management — and how the tech may define the future of risk management.
1. Ensuring Compliance With AI
Compliance is key in risk management. Failing to comply with anti-money laundering (AML) and know-your-customer (KYC) compliance guidelines can leave banks and other financial institutions vulnerable to fraud.
At the same time, however, regulatory compliance spending has ballooned over the past few decades — to the point that expenditures are on a course some industry leaders consider unsustainable.
The burden of compliance may also be bogging down banks’ customer onboarding processes, potentially leading to increased account abandonment. According to data from Jumio, 38 percent of millennials have abandoned mobile banking activities because the client onboarding process took too long.
Advanced automation with machine learning may be the solution. AI and ML technology can analyze atypical and heterogeneous data sets — like descriptions of transactions, phone or email conversations and network relationships — to detect fraud, manage risk and automate compliance in ways that traditional algorithmic approaches can’t.
For example, some specific applications of machine learning, like natural language processing, enable a computer to effectively “read” vast amounts of information and plain language communication. As it does so, the computer can automatically extract risk-related information that can be used to make judgments related to KYC and AML compliance.
2. Using AI to Identify Less Risky Investments
In the same way that AI can process atypical data to manage compliance, it can also be used to identify less risky investments.
Another example of AI risk management tech in action is ZestFinance, which partnered with Chinese tech giant Baidu in 2017 to improve the company’s lending strategy. Baidu, which operates China’s largest search engine, was primarily interested in offering small loans to retail consumers who use the company’s platform to buy products.
Lending to consumers in China comes with some unique risks. Less than 20 percent of people in the country have credit ratings or profiles — and lending to those with “thin” credit profiles, or no credit data at all, is inherently risky as they have no borrowing history to predict risk.
Baidu has much less standardized, more conventional financial risk information to go off of — but as a tech giant, the company still possesses vast stores of data on these retail consumers. This data can likely be analyzed with AI to identify high- and low-risk lenders, even without credit ratings.
Just two months into Baidu’s trial of ZestFinance’s AI-powered tech, the company was approving 150 percent more small-item loans without seeing an increase in credit losses.
ZestFinance hasn’t disclosed the exact combination of tech the company employed, beyond confirming the use of AI. Even without knowing the specific tech behind their success, however, the project is a powerful example of how machine learning can be used to identify less risky investments — even in markets with significant uncertainty and when traditional methods of risk evaluation aren’t possible due to atypical available data.
3. AI/ML May Soon Transform Risk Management
Risk management is currently undergoing a transformation from traditional strategies to ones that are highly data-driven and meticulous in approach.
This transformation is helping banks, lenders and other financial institutions avoid fraud and risky investments — however, it’s also bloating the costs of regulatory compliance and potentially steering lenders away from investments with risks levels that can’t be quantified by traditional analytical techniques.
Artificial intelligence and machine learning may provide a way out for the industry. The tech’s ability to work with unconventional data types — like written information — allows it to effectively automate processes that were previously impossible to automate.
It also enables companies to more accurately identify high- and low-risk lending opportunities, even when traditional data isn’t available.