Artificial intelligence (AI) is possibly the most significant development across industries and in our daily lives since the advent of the internet. Despite recent hype around AI’s applications, it has in fact already been in use for some time, assisting companies to streamline operations, enhance customer experience, and make informed strategic decisions.
"AI including robotic process automation and machine learning is rapidly transforming entire industries" - Paul Selibas
“AI including robotic process automation and machine learning is rapidly transforming entire industries,” says Ukheshe’s President: Project Engineering, Paul Selibas. “The AI fintech sector alone is projected to grow to almost $32 billion by 2027, but how and where we use AI will determine if it has a meaningful impact in certain industries and use cases.”
AI in financial services
AI refers to the application of advanced computational algorithms and machine learning techniques to automate specific tasks. AI has the capacity to gather, clean, analyse and archive massive amounts of data which would take a human – or a team of humans – a whole lot longer and with a greater margin for error.
In the context of the financial services sector, AI's application is specifically valuable in the following areas:
The more advanced our financial systems become, the more sophisticated the hacks and hackers also become and fintech’s need for fraud detection is contributing to the growth of AI. AI can detect anomalous customer behaviour and suspicious activity and prevent fraudulent transactions, saving consumers and institutions millions of Rands in losses. Using real-time data, AI can prevent fraud from occurring before, rather than after, the fact.
Fraud detection extends to interventions in international financial compliance and money laundering activities, aiding in the battle to combat international financial fraud. The challenge for AI will be to remain one step ahead of criminals who are able to make use of the same AI systems.
Ukheshe has introduced a Scan to Pay Fraud engine which employs machine learning to analyse transactional data that we have collected over time. The algorithm scores a transaction in real time on a scale of the likelihood of fraud. In this way, Ukheshe can decide whether to block the transaction, protecting merchants from potential financial harm.
Risk Assessment and Management
AI can assist financial institutions in managing operational risks, saving millions of Rands in costs associated with defaults on loans and other credit facilities. By assessing creditworthiness using credit history, financial statements and other customer-related data, finance houses can set lending limits and determine insurance premiums. Using predictive analytics, AI identifies potential risks, reduces lending exposure, and improves customer targeting for a range of financial products.
User Experience, Chatbots, ChatGPT
Improved customer service levels and personalised user experiences can all be automated using AI. Chatbots are already commonly used in the financial services sector and are reported to be saving banks up to 30% on customer service costs. Natural language processing (NLP) allows chatbots to understand questions and requests for information and in some cases to process transactions, dealing with customer issues quickly and efficiently. Chat GPT is one such NLP model and can even inject humour into customer engagements. Chat GPT can also be deployed to monitor external data sources such as social media, gathering information on customer sentiment around service levels. Using AI to aggregate data from chatbots and Chat GPT interactions, financial services institutions are also able to make greater personalised recommendations for products and services based on customer history and behaviours.
AI algorithms are assisting money market traders to make predictions on future price movements based on historical data. AI can detect patterns that are undiscernible by humans allowing for faster and more accurate trading execution which in turn can give traders a competitive edge and higher returns.
Robo Advisory Services
By inputting individual financial goals, risk tolerances and prevailing market conditions, AI can assist with automated investment advice and portfolio management. Using machine learning, personal finance management can help customers better understand their finances and answer questions such as ‘Can I afford a new car?’ or ‘Give me three recommendations on where and how to save money’.
AI can help financial institutions comply with complex regulatory requirements at home and across international boundaries. Already anti-money laundering (AML) and Know Your Customer (KYC) software is being utilised by many banks to monitor suspicious transactions and to prevent fraud. AI is able to conduct monitoring and auditing of operations, review policies and procedures and flag areas of potential risk.
AI - Not always for good
As prevalent and sophisticated as AI is in the fintech environment, it is also available to those who would use it to circumvent anti-fraud measures, and for hacking. AI is being deployed by cyber criminals to access financial data and hack into accounts. It is important that financial institutions remain one step ahead and install the most robust cybersecurity measures possible.
AI cannot be used for good if there is not equitable and fair access for all. Data models must be structured to manage diversity and inclusion to prevent unfair lending practices from being perpetuated by automation. Algorithms are, after all, programmed by humans, and if a certain geographical area or postcode, or credit levels are flagged as potentially risky, this may deny a bond applicant from accessing a home loan, or a small business owner from accessing finance. Automated exclusions impede societal and economic growth - a family is denied the chance to purchase a home in a better school catchment area, with improved educational opportunities for the children, and the entrepreneur whose potential to employ many people is prevented from expanding. AI has the potential to perpetuate historical biases without human intervention and adjudication and developers must be sensitive to these nuances.
At Ukheshe, we are carefully evaluating and monitoring the rollout and application of AI models and we are looking at work streams and areas in which it can best be used for the benefit of both the company and our customers.