Artificial Intelligence is Already Managing Banks


Written by Dr. Darius Dilijonas, Founder, Head of Product Development

We see only the beginning of wide application of artificial intelligence methods in the field of business operations management. Artificial intelligence includes a wide range of areas of research: intelligent machines (robots), software robots. The essential distinctness of such systems is their ability to reason, collect and extract knowledge, recognize patterns, learn and adapt to new situations or environment.

Human beings are thinking entities, able to learn from information collected from the environment and latter apply the accumulated knowledge in various situations. For decades scientists researched the abilities of biological systems to extract, accumulate and apply knowledge. Artificial neural networks were created by imitating activity of human brain.

Evolutionary computation algorithms were developed by analysing behaviour of biological organisms. They are based on the principles of natural selection – the strongest survive. The research of behaviour of birds and ants led to development of swarm intelligence algorithms. They are successfully applied in the development of software robots, which are able to automate solutions to complex problems: to analyse the current state, predict behaviour and model the optimum management solutions. These systems can instantly model millions of possible solutions taking into account various environmental factors. People are generally incapable of that. Furthermore, they are prone to error. Intelligent software robots can employ computational capacity of the present computer systems to make better decisions than a man. Five elements of the system are necessary for this purpose: virtual environment for business modelling and simulation; intelligent communication technologies; intelligent prediction models; intelligent models of optimisation and analysis.

The software robots use communication technologies to collect real-time information of the business environment. They may be integrated into banking systems in order to collect information on services provided to clients, for example, withdrawn and deposited cash or the flow of clients in branches of banks and the scope of the services they received. A forecast is prepared using the collected data: the need for cash in every ATM or the demand for services in every branch is determined. Intelligent systems are able to form adaptive and flexible prediction models according to historical data. The behaviour of the system is simulated and the factors determining certain fluctuations in scope of services are identified. Such systems are quick to learn that the demand for cash increases before holidays, they recognise pay days. Individual behaviour prediction models are developed for every ATM or a bank branch.

After it becomes clear how the controlled objects behave, activation of intelligent optimization algorithms follows. Each of such algorithms may have a specific function of objective. For example, reduction of cost of cash management by ensuring that a necessary amount of cash is always present in ATMs. Implementation of this objective includes modelling of various solutions. The software robots model solutions which may be limited by static or dynamic rules. Intelligent systems may learn new rules or their combinations, which produce the best results. This is called the ability to reason.

Similar sets of rules are used by people in their everyday lives. Knowledge is accumulated from experience and may be applied later. The software agents have an advantage in this area due to their high efficiency in solving specific problems, evaluating all possible options for a solution. People are usually incapable of that. The quality of solutions depends on a person’s individual abilities to apply the acquired knowledge and learn. Meanwhile, the software robots may find the very best solutions if they have all the necessary information. The system may tell the time and the amount of cash money that needs to be delivered to any ATM, also how preparation and delivery of cash is to be scheduled in order to reduce the costs to the minimum. Practical applications of such systems allow saving up to 40 per cent of costs and improving accessibility of the services in bank systems.

People will be replaced by the management systems of intelligent business operations in the long run, which will enable more efficient management of service systems. Such automation of production will result in the reduction of human labour. The forthcoming decades will see automation of operational decision-making. A large volume of data is currently accumulated in business information systems and this amount increases every day. A man is hardly capable of processing such amounts of data. This data may be employed to make better business management decisions through use of intelligent models.