Laurent -

Artificial intelligence making major inroads into Russian banking

Russian banks are identifying use cases for artificial intelligence, but the gradual adoption of the technology will require access to a deeper skills pool if it is to accelerate

Russian banks are stepping up the use of artificial intelligence (AI) as the technology’s unprecedented evolution looks set to see it boost their competitiveness, but shortages of people with the right skills, as well as infrastructural issues, are hampering the technology’s wider applications.

“AI is undergoing a period of unprecedented evolution. Over the next few years, the technology will progress so far that AI will be employed in financial institutions just as often as humans,” said Sergey Putyatinsky, deputy chairman of Credit Bank of Moscow (CBoM). “Active use of AI technology will be a decisive factor in banks’ competition in mass segments.”

According to a study conducted by the local rating agency Expert RA, in cooperation with the Centre for Financial Technologies, Russian banks most often use AI in credit analysis.

Other areas where AI adoption has been increasing, according to the study, are debt collection and marketing, including creation of individual offers for customers.

Meanwhile, the study revealed that Russian banks mostly put their hopes on AI in such areas as uncovering fraudulent transactions, debt collection and credit scoring, while automating call centres by introducing chat bots, using AI in algorithmic trading, human resources (HR) management and remote customer identification are generally considered less promising.

According to the study’s authors, however, the latter areas may not necessarily be dismissed by lenders as unsuitable for AI adoption, but it is difficult to come up with the returns on investment (ROI) from applying the technology to those areas. Meanwhile, Russian banks often prefer to adopt the technology by small steps.

“We are pragmatic about the adoption of ‘hyped’ technologies,” said Putyatinsky. “We normally start with smaller-scale pilot projects that allow us to evaluate the potential usefulness of the technology and build up in-house competences.

“Wherever possible, we use open-source software. We make calculations for every project to determine if it is financially viable and, based on that, we make decisions on whether to greenlight it.”

According to Putyatinsky, CBoM’s priority areas for AI technology are processing full-text documents, making loan decisions, dealing with over-due debts and financial monitoring.

Another major Russian lender, Rosbank, uses AI for processes involving risk evaluation, loan issuing, optimisation of the branch network, uncovering fraud, communications and interaction with customers.

“We believe that over the next one to two years, AI will also be adopted for the bank’s other processes that are not directly linked to interaction with customers,” said Dmitry Smirnov, head of Rosbank’s data lab.

“Accumulating large amounts of data and the arrival of new data sources will facilitate that. We are actively exploring areas where AI could be potentially adopted. These are processes aimed at improving the organisation’s efficiency and, from the customer’s viewpoint, processes that simplify their interaction with the bank.”

Meanwhile, Promsvyazbank’s main area for AI’s application includes credit decision-making, uncovering fraud and forming offers for customers.

“Currently, we are working on broadening the scope of AI application,” said Daniil Tkach, head of the customer relations department at Promsvyazbank.

“In the short term, automated systems will tell us which products would be the best offer for a customer, what channels will be the most efficient and what communication style will be most amenable to the customer.”

According to Tkach, the main conditions for wider spread of AI include a sufficient degree of automation and manageability, reliable systems for data collection and a sufficient number of reiterations of processes for learning purposes. He said this is applicable for just about any banking processes, such as sales, communications, anti-fraud and operations.

“We could also single out intellectual management systems, in which AI substantially helps superiors to understand the quality of work by their employees and provides tips to all employees for possibly improving their work,” Tkach said.

Still, Russian banks often see AI as a technology that could help automate new areas rather than replace already existing automation solutions.

“We are not trying to revamp existing solutions,” CBoM’s Putyatinsky said. “Instead, we look at areas that have not yet been automated and start automating them from scratch with the use of new technology.”

Obstacles to AI adoption

Meanwhile, the process of adopting AI in the banking sector is not always smooth. There are obstacles in the way of the technology’s wider spread across the industry.

According to the Expert RA study, those obstacles include discrepancies with data in information systems, but once the issue of data consistency is resolved, finding qualified personnel to process data is set to be a major challenge.

Industry insiders have already been complaining about difficulties in finding qualified personnel to operate AI-based solutions.

“The main factors that are impeding the adoption and development of AI are shortages of qualified professionals and problems with the infrastructure of information systems,” said Smirnov.

Putyatinsky agreed, saying: “The acutest issue is training of qualified personnel.”

To help resolve this challenge, CBoM has been running an internship programme called IB Universe for the past 12 months. “This allows students and recent graduates to acquire practical experience in various areas of investment business,” added Putyatinsky.

According to Putyatinsky, educational programmes of that kind will eventually allow banks to train personnel in the working environment, producing a new wave of employees who will already be prepared to deal with new technologies, such as AI and machine learning.

Another issue with application of AI is complexity of the technology’s algorithms, Promsvyazbank’s Tkach said. “Contemporary machine learning algorithms are so complex that humans have problems understanding decisions made by AI,” he added.

Over the next few years, progress with adopting AI systems in Russia’s banking industry is set to largely depend on investments in regional networks, personnel training and banks’ ability to attract and retain customers, according to the Expert RA study.

“The good news is that at this point, a bank doesn’t need to make enormous investment to become one of the [Russian banking industry’s] AI leaders,” said the study’s authors. “But the bad news is that to achieve that, you have to act right now.”

Read more about AI in banking

Read more on IT for financial services

Data Center
Data Management