daniel chigunwe
Elgin Chetsanga
AI, a new buzz word
AI! Nowadays it’s impossible to spend a day without hearing or saying this buzzing acronym.
Indeed, many of us would have first heard about AI when the wildly popular ChatGPT was released at the end of 2022.
The launch of ChatGPT undoubtedly led to the boom in consumer awareness and spurred the growth of many AI applications for everyday use.
Today, AI is now being increasingly recognised and embraced in many fields for its potential to transform day-to-day activities.
What is AI?
AI, as defined by an AI tool is “the simulation of human intelligence processes by machines, especially computer systems”. These processes include learning, reasoning, and self-correction. AI can be categorised into Narrow AI, which is designed for a particular task such as speech recognition or image classification, while General AI, aims to mimic human-like intelligence and abilities across a wide range of tasks.
AI is built on various technologies and approaches, which continue to evolve and grow in number. Some of the most common technologies which bring AI to life include Machine Learning, Robotics, Expert Systems, Natural Language Processing, Large Language Models, and a host of many others.
These technologies are sometimes loosely referred to as AI even though they are just mere components of the broader concept.
Embracing AI in risk management
Embracing AI in the Risk Management space has the potential to unlock efficiencies and ultimately value for your entity. In today’s fast-paced, data-driven world, AI potentially allows businesses to enhance their risk management capabilities.
Let us explore just a few examples of use cases for AI in Risk Management for the finance industry.
Fraud detection
One of the practical use cases of AI in Risk Management is in fraud detection. For financial institutions fraud detection is a critical area of focus. Financial institutions and their customers suffer losses such as card fraud, loan fraud etc.
For example, according to the Nilson Report, US$27,85 billion was lost in card fraud alone in 2018. AI can step in to detect fraudulent transactions and activities.
Technologies such as machine learning and anomaly detection techniques can be deployed to help flag and reduce the incidents of fraud.
Cyber security —The reliance on technology in today’s world has made protecting sensitive information a critical priority for any business. Another use case for AI in risk management is around Cyber Security.
AI can be deployed to enhance cyber threat intelligence analysis. Threat intelligence analysis provides essential data on sources of cyber-attacks and other indicators.
For example, an AI backed threat analysis can detect anomalies and breaches in real time and provide automated responses such as blocking certain users before they compromise the cyber security of the organisation.
Market risk analysis — Another way AI is being embraced in Risk Management is in redesigning market risk analysis. For example, AI tools can be used to scan news feeds-in real time to identify emerging risks and anomalies.
AI tools can also process vast amounts of market data such as price movements, shifts in exchange rates, interest rate changes etc.
Analysts now have access to quality insights and patterns that could be consequential to their market positions. Use of AI tools also help to identify emerging opportunities.
Workplace risk reduction — A key risk that many organisations face is physical harm to people and damage to assets of the organisation. Where there is potential for people to get into accidents as part of their job cycles, AI tools can be used to collect and analyse data related to potential hazards.
AI algorithms can assess behavioural patterns exhibited before accidents occur and use this information to generate predictive scenarios which help improve safety procedures and prevent incidents.
Credit Risk Management — Most financial institutions carry a substantial amount of credit risk.
Granting of credit is typically the most important function on the credit risk cycle and the task is usually assigned to the most senior and well experienced analysts.
However, relying on human intelligence alone can also have some limitations. AI has become a game-changer in the realm of credit decisioning.
Some financial institutions are pivoting to AI-driven credit risk models which evaluate credit applications more quickly and can consider a wider array of factors. For example, an AI-driven tool can be used to do Credit scoring of applications.
Model Risk Management — Another use case of AI tools is in the model risk management area. Model Risk is increasingly becoming an important area to many organisations that really on models for wide range of estimations.
Once a model has been designed and deployed it will need to be back tested and validated within a set frequency as required by prudential regulators.
Back testing allows model users to know the efficiency of their models and how well they are performing against their predicted outcomes. AI Tools can also be used for this critical workstream of back-testing and validation.
Challenges with AI
While AI has brought some exciting use cases, there remains a worrying set of challenges which the technology needs to overcome.
Amongst other worries, data privacy has emerged as a key concern. AI tools are data hungry and gobble up large sets of data, which data is usually collected in methods that may infringe on privacy, surveillance, and copyrights rules.
Furthermore, AI has been noted to sometimes display a phenomenon known as algorithmic bias. AI tools that learn from biased data end up perpetuating the same skewedness which was inherent in the data.
An example of a credit granting tool that was trained by biased historical data will tend to replicate the same biases and disadvantage whoever the historical data did not favour.
Other challenges worth mentioning include the chances of unemployment for risk practitioners as AI takes up more responsibilities.
There is also the concern around the lack of understanding of the actual inner mechanics of the AI logic which is often very complex. Lastly AI can itself be used by criminal elements to evade detection through using social engineering, cyber-attacks, and data manipulation.
In summary
Across many fields, AI has become synonymous with improving efficiency and productivity while reducing costs. However, some critics sound caution when it comes to the pace of adoption of these AI tools.
They argue that the hype has gone ahead of the substance and that time is still needed to perfect these AI tools before wide-scale adoption. AI like many technologies before it offers both opportunities and risks and a weighed adoption strategy is advised.
Elgin Chetsanga is a head of risk and compliance at a local financial institution. He writes in his personal capacity. Elgin can be reached on [email protected]
Bongani Ndlovu