What Is the Place of Artificial Intelligence in Banking Automation?

The focus of the 21st century's third decade is expected to be on using AI for smart banking.

The article was written by Valentin Sabev and published in CIO Media on 05 August 2020. Read the original in Bulgarian.

The last development of banking computerisation is focused mainly on procedural or situational processing and exchange of bank operations and transactions efficiently and effectively on an algorithmic basis.

In the first decade of this century, banking automation passed under the sign of IT centralisation. The second decade focused on electronic channels for distributing banking products and services.

The focus of the third decade is expected to be on the use of artificial intelligence (AI) for smart banking. Relevant solutions with built-in AI work without the involvement of human users, freeing them to do other work. The goal is to reduce costs and increase productivity without degradation of customer satisfaction.

In addition, AI is expected to impact on more efficient computerisation of preventive, operational or control back-office processes in banking. The predictions are that AI directly increases added value.

Entry of systems with embedded artificial intelligence

The first attempts to apply artificial intelligence to the banking business were based on conventional expert systems. By design, this type of system is used to solve problems in a specialised field, which usually requires a competent conclusion from a person.

Decision-making through them is based on rules (rule-based approach) such as "if-then". For this reason, the strength of expert systems depends on the number of conditions set in them, but the precision is lost in cases "unknown" to them or in the absence of pre-coded situations.

Both approaches are outlined for decision-making. The technique applied in the first approach is data-driven from available data and facts to bottom-up to reach the target state, starting from the states of available data and facts. In contrast, the second approach is characterised by goal-driven top-down decomposition of the goal into sub-goals.

At present, rule-based expert systems have found a place in banking, e.g.

  • The scoring of borrowers.

  • Detecting or preventing card fraud.

  • Compliance with the payments.

  • Identifying a client and providing access to their assets, etc.

Their solutions depend on threshold values calculated from preloaded static rules with constant coefficients. For this reason, these systems are not only adjustable from the accumulated history or adaptable to changing conditions and depend on expert intervention.

Chatbots are also entering, imitating conversations with real customers by generating responses based on built-in templates. However, future banking systems with built-in AI are required to meet the Turing criterion formulated in the distant 1950s. According to this criterion, a machine has intelligent behaviour if it can naturally convince its interlocutor that he is communicating with another person and not with a machine.

Artificial intelligence use cases to improve customer service in the front office

AI in customer communication

Mobile banking provides an increasing convenience for customers to communicate with their bank. But communication remains computer-dependent, forcing people to comply with the relevant technical capabilities.

We are entering a time when the direction is turned to the usual human way of communicating. Switches from text to voice dialogue, whether for audio or video exchange, as well as remote or direct communication. As a result, the power of intelligent banking systems will be felt in providing "live" communication in customer service.

Artificial intelligence tools have begun to be included in the software of some chatbots. This means that along with the built-in software templates, the customer's questions are stored and the answers given to him. Subsequently, the accumulated information is used for subsequent dialogue with the client. In this way, systems evolve and do not need constant changes in the software template they work on.

The development of a smart user interface in natural language will cause a digital transformation of distribution channels. They will undergo their digital development, expressed in "talking" applications or humanoid-like robots in call centres, providing equal service to all customers. However, for greater closeness to the client, it is necessary to incorporate emotions, e.g., to show sympathy in case of an incident, to show adaptive behaviour towards the customer segment of the caller, etc.

The time is not far off when, with the help of 3D printing of human elements, robotic droids will appear, possessing artificial intelligence and strongly resembling humans. As a result, robotic "service personnel" in the branch network will be common, and so on.

Mobile 5G communication will provide the ability to provide a fast and efficient connection between the "peripheral" processors on the robots and the central "brain" server located in the data centre. Intelligent decisions are made by a central server with a built-in artificial intelligence system and a large amount of data, and the "peripheral" processors on the robots should perform them.

AI for client identification

The incorporation of AI and 5G communication in productive servers will catalyse the use of intelligent software solutions for real-time facial recognition or video link analysis for snapshots/selfies, characteristic facial expressions, gestures, lip-reading, voice message or fingerprints.

AI for intelligent personalised service

The implementation of data warehousing (DWH) and customer relationship management (CRM) has led to the accumulation of big data from different sources. The large volume of own data and artificial intelligence form the engine that drives all creative efforts for automatically personalised banking, differentiated by customers. The effect is in creating more efficient operations, more satisfied customers and as a result - more benefits for both parties.

The next step in personalised service is to be tailored to the needs of the individual customer intelligently. It is a common practice to offer the same "ready-made" products or services for pre-selected channels. However, it does not consider the individual preferences of customers to ensure better individual satisfaction.

The use of artificial intelligence paves the way for the transition from ready-to-wear through reference to customised banking. The unified and static offer for almost all customers is replaced by a dynamic-adaptive service for each random customer. The goal is for each customer to receive products and services to their individual needs at any given time for the most appropriate channel. As a result, customer satisfaction increases while ensuring optimal profitability.

But robotic solutions are unlikely to be suitable for financial advice to bank customers.

Architectural landscape with AI components

Artificial intelligence for better, faster, and cheaper processes in the back office

Along with the use of artificial intelligence in customer service, robotic process automation (RPA) solutions will be widely applied in back-office activities. This type of solution is based on software robots (bots) or systems with built-in artificial intelligence.

Advances in artificial intelligence technologies make it easier for businesses to obtain the benefits of RPA without spending a large budget on development work. However, for the decisions to be effective, truly clear rules and goals must be established for the smooth running of banking processes.

RPA Use Cases

The power of RPA is in the execution of repetitive mass business processes by automated software robots. Computerised pipelines or conveyors are set up without human participation, removing repetitive and routine tasks from the daily activities of the bank employee. The employee focuses on intellectual work with greater benefits for the bank. The impact is in increasing the operational efficiency, improving the quality and in strict compliance with the normative regulations.

The expected transformation is to move from a conventional solution to robotic solutions for automation of banking processes in the back office, where intelligent computer solutions will initially replace those for trivial or routine tasks.

A new point would be the "smart" robotic data transfer management and their "smooth" loading of data in DWH without direct human involvement. Automatic integration into new data sources, transfer of informal information in a computer-compatible format, as well as monitoring when entering incorrect data are part of the functions of RPA solutions.

The next direction is to improve the used expert systems in the direction of artificial intelligence or impose their future alternatives in the background. Potential domains are:

  • Total control of final processing of transactions.

  • Ensuring compliance with regulations (e.g., AML, CFT, KYC, etc.).

  • Maintenance of business risk within certain limits and early notification of possible deviations.

  • Preparation of forecast analyses and operational reports from various sources of information, types of media and various.

Future "smart" platforms with fully autonomous operation, resulted in avoiding the disappointment of the affected loyal customers, will provide:

  • Fraud prevention and elimination extra efforts: reversals, refunding or voiding in banks.

  • Access to client assets with the relevant operating rights.

  • The steady rhythm and uniform quality in the processing of products and services.

Besides, robotic droids with built-in artificial intelligence could serve cashiers: transferring banknotes, counting them and their packaging in bundles. However, their role could decline with the increasing penetration of digital currencies e.g., CBDC.

Artificial intelligence for more efficient IT services

The increasing level of computerisation requires a real-time assessment of software delivered and implemented. AI systems will undoubtedly be a necessary benefit, without being considered a panacea for better IT services.

One of their expected uses is to trigger the transformation of test operations and quality assurance (QA). AI will soon become part of the ongoing quality process of software engineering in functional, regressive and other types of business testing with depersonalised data. In addition, "smart" test platforms will cover the operations of generating test scenarios and analysing results to determine regular performance and maximum workload.

The second option is to use AI techniques for providing software fixes or upgrading to the next versions. In this way, the requirement for a segregation of duties in IT operations complies. Currently, this activity carries some risk of operator accidental errors or intentional actions.

The next possible option of AI is to perform real-time monitoring and diagnostics of ongoing IT processes for compromised performance or deviations from the expected schedule of completion of operational activities.

In parallel, the weaknesses or bottlenecks that cause disturbances in the bank operations are identified. Along with the diagnostics made and the accumulated historical data, a relevant system with AI should be able to predict or offer solutions, e.g., analysis of registered events in log files.

The multidimensional penetration of AI is also key in the field of IT security: auditing and analysis of registered events in log files, detection of regulatory compliance or white spots in user rights, forecasting of risk events, etc.

Conclusion

In conclusion, the rise of systems with embedded artificial intelligence will cause one of the defining digital transformations in the banking business in the coming years to reduce operating costs with increased quality, being equal for all customers.

At the front office, AI systems are expected to be effective and attractive. At the same time, the AI or RPA solutions used in the back office will be more effective and efficient.

The real benefits will come when the use of artificial intelligence for marketing purposes is replaced by direct use in banking operations. Perhaps one of the methods for its successful entry is to maintain the gradual adaptive implementation (agile), but the teams of business analysts and IT experts need to expand with professionals from other specialities, e.g., psychologists and lawyers. To a certain extent, a new way of thinking will be obtained in interdisciplinary teamwork in solving business, technological and ethical challenges.

About INDUSTRIA

INDUSTRIA is a global technology consulting, development and ventures firm with extensive expertise in the field of enterprise blockchain, confidential computing, process automation and digital experience. Official partner of R3.

Let’s talk about making your idea a reality — contact us!

Follow us on TwitterLinkedIn and Facebook.