Artificial Intelligence in Risk Management: Is it here to Stay?
The excitement around Artificial Intelligence has existed in our world for decades, the idea introduced through sci-fi books and films that portrays a world in which humanity is traded for machines. While we can’t say for sure what the future of Artificial Intelligence holds, we are beginning to see a rise in its implementation, especially in banking. In recent reports, 75% of banks with over $100 billion in assets have begun implementing Artificial Intelligence, predicting that it will save up to 447 billion dollars by 2023.1 According to a 2018 study by GARP/SAS, AI is being applied to various areas of banking: 52% in Automation of Manual Processes, 45% in Credit Scoring, and 43% in Data Cleansing. In Canada, the use of AI in banking is still new, but banks such as TD are beginning to see its merit; various reports and models have been created to understand how to implement it and what the improvements would look like.
What is AI?
In order to understand the impacts of Artificial Intelligence, it is important to understand what AI means. At its most basic definition, artificial intelligence is a technology that is able to perform complex tasks, in which it is able to think, learn and understand on its own, much like a human brain. In other words, AI is a technology that is able to analyze and comprehend massive amounts of data, is also able to take action on the data it has received and can learn and improve from those actions for future uses. Falling under the umbrella of AI is machine learning. Machine learning focuses on allowing a system to learn new things from data and thus, come up with its own predictions and conclusions.
How will AI disrupt Risk Management?
So how would this technology disrupt risk management in banking as we know today? A TD bank report analyzed the use of machine learning and its ability to make successful risk predictions by using low volatility strategies. By comparing traditional models to machine-learning models, TD’s report reveals the limitations of the traditional models. According to the report, the traditional models are unable to capture all sources of risk because of its reliance on a fixed number of risk factors. As a result, new risk factors go uncaptured because the models are based on strict assumptions of the input data (normal distribution, linear relationships with outputs, non-collinearity). An analysis of machine-system models reveals that its flexible formula is more advantageous for risk management. The machine-learning system is not only able to learn on its own using algorithms, imitating human thinking, but its flexibility allows room to remove or add in new risk factors to output the most accurate risk prediction.
Potential Problems with AI
Even with the variety of evidence proving AI as the next radical step in transforming banking for the better, many still carry reservations. Despite the obvious benefits AI brings, specifically its ability to predict risks accurately, there are still issues that need to be considered. In a Deloitte report, the biggest threat of AI in risk management is that existing risks become a challenge to identify quickly and display themselves in unfamiliar ways. Outside of that, there are also risks that can come with the model (wrong algorithm, biased results, etc.), problems with the technology (hacks, compatibility, etc.), regulatory and compliance (ie. legal data rights), and people (lack of concrete responsibilities, lack of skills, etc.). Taking these issues into consideration, the use of AI becomes a lot more complicated. Yet despite these limitations, it is important to note the clear benefits that AI brings to risk management. GARP/SAS found that the implementation of AI in risk management promises 78% faster insight from data, a 77% reduction in manual tasks, an upwards of 77% improvement in decision-making, 73% higher productivity, a drop of 66% in operating costs and an increase of 66% in customer service.
Artificial Intelligence is still a relatively new technology that is continuously growing and improving, but in recent years, businesses have begun to realize its potential. In risk management, particularly, AI allows for new avenues in which accurate and reliable results are found. This not only changes the way banks will navigate risk management but will allow for a more immersive and effective customer service. The banking industry is evolving with AI, and its future is one that is bright and promising.