Thinking about applications of artificial intelligence (AI), my first thoughts as a robotics student revolve around autonomous systems, robots and computer vision systems – banking surely is not one of them. Until I started an internship abroad and my credit card got blocked. My bank detected fraudulent transactions, blocked my card and I was suddenly proving my identity through biometric data to have my new card shipped abroad. I first hand experienced that banks were also using AI.
Apart from the front office applications that a regular customer would encounter, there are many more applications for AI in Finance and Investment, such as analyzing the credit worthiness of clients with a limited credit history and thereby enabling their financial inclusion. Furthermore, AI based algorithmic trading solutions have proven themselves as worthy in the field of quantitative finance. By leveraging the potential of Big Data, AI can be used to analyze financial markets with great efficiency and detect anomalies which can then be used to implement meaningful indicators. The organization for Economic Co-Operation and Development (OECD) analyzed the use of AI, machine learning and big data in finance and categorized it by its application areas and distance to the customers.
An overview is displayed below .
Examples of Real-Life Use Cases
Datavisor – Fight fraud and financial crimes with AI:
Their solution enables real time detection of fraud patterns and can take proactive steps to fight them. It uses unsupervised machine learning algorithms to identify known and unknown malicious actions .
Upstart - Credit profiling using AI:
Traditional lenders rely on a FICO based credit score, which leaves out much of the available information on a borrower. Upstart uses additional information such as education and job history in their AI models to calculate credit profiles and thereby enables borrowers with no or limited credit history to get a loan. They offer loans and also sell their credit profiling solution as SaaS to financial service providers .
Challenges for AI in Finance
The inadequate use of data can lead to issues with algorithmic fairness and poses the risk that credit decisions are biased, unfair or discriminatory .
Regulations and Governance
The interpretability and explainability of AI systems is a very complex matter. This makes the adherence to regulations and internal governance very complicated. A general lack of explainability can also lead to a lack of trust in AI and hence limited its adoption .
Risks of AI based financial decision making
The use of similar models by multiple financial institutions could lead through the convergence of actions to a herding behavior which may raise risks for the stability of the system, especially during times of stress through a self-reinforcing mechanism and could be exploited by malicious actors .
AI is increasingly being used in Finance and Investment. It will impact us directly by changing our interactions with banks through personalized customer service, enabling the rise of new products and automation of processes. Through its use of data, it may enable financial inclusion for people with a limited credit history and identify fraud and money laundering. The reliance on data driven solutions and probabilistic algorithms also raises questions about how to properly regulate AI systems, to enable safe and fair decisions.