Artificial intelligence in banks

Main article: Artificial Intelligence

The main directions of AI execution in finance

For 2024, Spydell Finance identified several areas of application of artificial intelligence technologies in the field of finance.

Forecasting and Risk Management

Risk management is based on the most formalized segment of the financial industry, which is best able to be "locked" into the framework of AI, which is able to predict risks associated with lending, investments and insurance, taking over the functions of risk managers.

Advanced Data Analysis

Finance, like economics, is a continuous stream of data that can be ordered through identifiers, weights and "beacons." AI is able to analyze huge amounts of financial data in real time, including transactions, economic and market trends, and consumer and corporate behavior.

Trend forecasting

A significant part of the data and processes in the financial industry is regularly repeated through various combinations, and therefore pattern analysis is possible, where AI is very strong (statistics and probabilities), which makes it possible to predict the most likely trends faster, more accurately and more efficiently.

Automated investing

This approach has been used since the beginning of the 2000s within the framework of algorithmic systems and trading robots, but now it can reach a completely different level due to a combination of tools, where probability and pattern analysis, risk management and forecasting are simultaneously combined.

Automate tasks

Over 80% of business operations in finance and insurance are pure routine according to action protocols. If there are protocols, then there is space for AI, which can automate many routine and time-consuming processes, such as processing loan applications, managing client accounts and analyzing insurance claims, which increases efficiency and reduces costs.

An AI-based financial adviser can provide high-quality customer service by providing quick and accurate answers to their questions, as well as helping with financial transactions - much faster and more efficient than a person, except for difficult questions.

Automatic writing of investment and market reviews/news

Training AI on the basis of millions of investment reviews over the past 50-60 years will create a highly developed investment analyst based on AI, which will quickly, relevant and high-quality write reviews.

Detect and prevent fraud

Analysis of transactional data to identify suspicious or unusual patterns (fraudsters in 97% of cases act according to similar schemes), which helps in the fight against financial and insurance fraud.

Automatic control of tax requirements

Automatic control over tax requirements and regulatory norms, which will avoid fines and prosecution by the state.

2024

Named the most important financial AI technologies

Artificial intelligence, including generative intelligence (GENI), is being actively introduced in the financial sector. However, the excitement around these technologies can be replaced by a period of disappointment, since some solutions used may not meet expectations. This is stated in a study by Gartner, the results of which were released on August 20, 2024. Analysts highlight several of the most important financial AI technologies.

GIi agents

Financial services companies can use Genia tools to improve customer experience while reducing costs. One example is the introduction of chatbots to create product recommendations and respond to customer requests. Thanks to GeniI, credit institutions can accelerate lending in underserved markets, especially in developing countries. In addition, Genia funds help identify fraud in claims, credit cards or loans. Although Gartner experts predict some disappointment in Genia tools in finance, this does not mean that the technology will be useless.

The main advantages of Genia in the field of finance are ease of access and ease of use. Since many organizations implement their own solutions in this area, the use of such tools largely amounts to training employees on how to use them and under what circumstances this is a reliable solution, says Mark McDonald, senior analyst at Gartner's financial division.

Machine learning

When it comes to numerical-based tasks, financial institutions will have to rely on other artificial intelligence techniques - primarily various machine learning applications. Such funds, as noted in the study, can help financial professionals solve complex problems, including predicting income or finding errors in large amounts of information. However, the use of machine learning will require some new skills.

Composite AI

This concept provides for a combination of several methods of artificial intelligence for a deeper interpretation of information and solving a wide range of problems. The goal is to create AI solutions that require less data and energy to learn. Composite AI provides greater opportunities for those organizations that do not have access to large amounts of historical or labeled data.

As AI is introduced into the financial sector, it becomes clear that no single technology individually is a panacea. Combining artificial intelligence methods provides higher efficiency than using only heuristics or a completely data-driven approach, "emphasizes McDonald.

Agent-based AI modeling

The approach involves the use of several independent agents, each of which is able to perceive the environment and perform certain functions. Agents can be AI models, programs, etc. Integrating these components into a single entity improves overall awareness of the situation, enables deeper optimization of planning, and improves process sustainability.

Neuro-symbolic AI

This type of artificial intelligence combines machine learning methods and symbolic systems, for example, knowledge graphs. The approach makes it possible to create more reliable and trustworthy AI models.^[1]

VTB launches cluster for the development of artificial intelligence technologies

VTB launches its own cluster for the development of artificial intelligence technologies - "Digital Assistant." This was announced on July 5, 2024 by the Deputy President-Chairman of the Board of VTB Vadim Kulik. [1]Подробнее #.2A.D0.92.D0.A2.D0.91_.D0.B7.D0.B0.D0.BF.D1.83.D1.81.D0.BA.D0.B0.D0.B5.D1.82_.D0.BA.D0.BB.D0.B0.D1.81.D1.82.D0.B5.D1.80_.D0.BF.D0.BE_.D1.80.D0.B0.D0.B7.D0.B2.D0.B8.D1.82.D0.B8.D1.8E_.D1.82.D0.B5.D1.85.D0.BD.D0.BE.D0.BB.D0.BE.D0.B3.D0.B8.D0.B9_.D0.B8.D1.81.D0.BA.D1.83.D1.81.D1.81.D1.82.D0.B2.D0.B5.D0.BD.D0.BD.D0.BE.D0.B3.D0.BE_.D0.B8.D0.BD.D1.82.D0.B5.D0.BB.D0.BB.D0.B5.D0.BA.D1.82.D0.B0 здесь.

Named the main risks of the use of AI in the financial sector of Russia

As of the beginning of 2024, the financial sector is the leader in the introduction of artificial intelligence technologies in Russia. The application of AI opens up huge potential for improving the efficiency, accuracy and security of financial transactions, as well as for the development of new products and services. This is stated in the study, the results of which were published in mid-April 2024.

According to the FinTech Association, about 90% of AI methods and tools used by Russian credit companies are based on machine learning methods. Russian banks use such technologies to create computer vision systems, launch speech services, work with text, analyze data (predictive analytics) and intelligent robotization. In 2023, large language models (LLM) became the technological driver for the development of AI in fintech.

The leadership of the financial industry in the implementation of solutions based on artificial intelligence technologies in Russia is indisputable.

At the same time, there are also critical risks of applying AI-based solutions in the financial and banking sector. These are, in particular, distortions of automation - when a solution that initially contains errors is automated. There is also the possibility of using poor quality data in AI training, which negatively affects the quality of services. Concerns raise risks of intrusion into the privacy of clients of financial institutions when collecting and using data and ethical risks associated with national, religious, regional components (for example, incorrect access to a client). Plus, experts point to possible problems in the field of cybersecurity.

For Russia, a specific risk is sanctions pressure, which makes it difficult to purchase powerful AI equipment, including accelerators based on GPUs. In addition, unequal competition and monopolization of the market are observed: this is due to the significant opportunities for technical development among large players and their absence in small and regional industry participants.^[2]

How the National Bureau of Credit Histories uses AI

The National Bureau of Credit Histories (NBCH) applied artificial intelligence in PD scoring creation - estimates of probability of non-payment of the credit (default). Oleg Skvortsov, chairman of the board of the Association of Russian Banks (ARB), spoke about this project at the end of February 2024. Read more here.

Why banks are spending billions of dollars on generative AI

Research by Juniper Research, whose results were published on January 23, 2024, suggests that banks are rapidly increasing spending on generative artificial intelligence (GenAI). The implementation of such tools improves the quality of customer service and improves the level of security.

According to analysts, in 2024, banks' global costs for GenAI systems will amount to about $6 billion. In 2030, this value is expected to reach $85 billion. Thus, growth over the period under review will exceed 1300%. Leading banks will roll out GenAI services to offer a more personalized customer experience, according to Juniper Research. This will allow financial institutions to provide more attractive services at reduced prices.

Financial firms can use GenAI-based chatbots to create product recommendations and answer customer questions. Thanks to AI, credit institutions can accelerate lending in underserved markets, which is especially important for developing countries. In addition, GenAI tools will help more effectively identify various fraudulent schemes thanks to the ability to quickly analyze huge amounts of data.

The study found GenAI systems would radically change banks' habitual operations by providing personalized spending information and tracking market trends. Financial institutions, according to analysts, will increasingly move to a strategy focused on artificial intelligence, since such business models are necessary for effective competition in a highly dynamic banking environment. Investing in GenAI will allow banks to gain a competitive advantage as their costs fall and customer service levels rise.^[3]

2023

A debtor location system has been launched in China. It was created with the participation of Russian scientists

Scientists at the Ural Federal University (UrFU), together with Chinese developers, have created a program that can calculate the location of borrowers who miss or do not plan to return loan payments. This technology was reported at a Russian university in December 2023. Read more here.

10 AI trends changing the banking sector named

Artificial intelligence has had an impact on many industries, including the banking sector. Thanks to neural networks and machine learning algorithms, financial organizations are able to improve the quality of services provided, optimize operations and provide more personalized services for customers. On October 21, 2023, Finextra named ten AI trends changing the banking industry.

1. Improved Customer Service

AI-powered chatbots and virtual assistants are becoming commonplace in the banking sector. These intelligent tools can interact with customers around the clock to respond to requests, assist with routine operations, and offer useful information. AI not only increases the efficiency of financial institutions, but also reduces operating costs.

2. Personal recommendations

AI algorithms analyze customer data to assess individual preferences and financial behavior. Using such information, banks can form personal recommendations, which helps in the promotion of products and services.

3. Detect and prevent fraud

AI plays an important role in improving security in the financial industry. Machine learning models can detect anomalies when performing transactions and warn of potentially fraudulent activities. Such systems analyze huge amounts of data in real time, quickly identifying suspicious operations.

4. Credit Assessment and Risk Analysis

Traditional credit scoring models are often limited to a narrow set of parameters. AI-based systems use a broader set of data sources, including social media activity and online behavior. This comprehensive approach allows banks to make more accurate lending decisions while reducing risks.

5. Back Office Automation

AI helps in a number of tasks that were previously carried out manually and took a lot of time. Automation not only reduces the likelihood of errors, but also allows organizations to improve operational efficiency.

6. Predictive Analytics

AI-based models are capable of analyzing huge amounts of customer data. Such tools can be used for marketing purposes, in the development of new products and to attract customers.

7. Know Your Customer

AI simplifies the process of registering new customers by automating document verification. AI-based systems, such as facial recognition and, biometric authentications improve convenience and security.

8. Compliance

AI helps banks comply with established requirements through the automation of regulatory reporting and monitoring. Smart systems can analyze vast datasets for any anomalies or violations, warning of possible problems.

9. Improving Data Security

AI technologies such as natural language processing and machine learning algorithms improve security in banking operations. These tools can identify sensitive information in unstructured data and protect it from unauthorized access. AI also allows you to detect and respond to cyber threats in real time.

10. Investments and Wealth Management

AI-based consulting robots make investment and asset management services more affordable. Automated platforms not only reduce customer costs, but also make financial services more accessible to the general population.^[4]

Russian banks invest about $1 billion a year in artificial intelligence

The largest Russian banks invest a total of about $1 billion a year in the development of solutions based on artificial intelligence, and the profit from investments reaches $3 billion a year. Such figures in the Fintech association in early October 2023.

According to her, the largest financial organizations in Russia have already invested more than $10 billion in the development of AI over the past 10 years. The largest banks mean the five largest credit institutions in Russia in terms of assets at the end of 2022. The study notes that medium and small Russian financial companies invest only 100-300 million rubles a year in the implementation of a portfolio of projects with AI. Thus, the largest banks of the Russian Federation invest 500 times more annually in the development of AI than any other companies on the market.

It also follows from the study that based on the total effects on the revenue of companies for which AI has made a significant effect, the AI market in 2022 is estimated at 647 billion rubles or $7.1 million. The market growth was 17.3% compared to 2021.

The key barrier in the introduction of artificial intelligence, indicates the Fintech association, is the shortage of specialized specialists. 83% of Russian companies noted that they are experiencing a shortage of personnel here.

In July 2023, the head of Sberbank German Gref announced that the bank annually invests about $1 billion in artificial intelligence. At the same time, the return on investments pays off three times, he said. As explained in Sberbank, artificial intelligence in the bank helps to create new products, improve customer experience and develop a customer-centric approach.

New technologies help bank employees make a wide range of decisions at all levels of management. And artificial intelligence is already making some decisions better than humans - for example, how to optimally build a collection route and how much money to put in each specific ATM, the bank notes.[5]

The Central Bank highlighted the main risks of the introduction of AI in the banking sector

The central bank identified three main risks of artificial intelligence in the financial market: competition, access to data and an ethical issue. This was announced on October 2, 2023 by the press service of the State Duma deputy RFAnton Nemkin.

The active introduction of artificial intelligence technologies in the banking sector can lead to a threat of reduced competition, compromise of data that are used to train AI, as well as entail ethical risks, said Deputy Chairman of the Bank of Russia Alexei Guznov.

Probably, the result will be received only by those who have the opportunity to invest. And this forms certain distortions, - said the expert.

In addition, there is a risk of insufficient information security and, as a result, the possibility of leakage of user personal data.

This is not only our problem, it is comprehended philosophically - the problem of combining human intelligence and artificial intelligence. Since AI learns the natural language faster, it can form illusions much more skillfully, select the appropriate context, "explained Alexey Guznov.

It's good when we can understand what negative phenomena some technology has and immediately stop them. Another thing is when it is difficult to determine the possible negative consequences at the start. This is largely characteristic of the field of digitalization and is due to the fact that the development of technologies, including AI, is very dynamic - we do not always keep up with them, - explained Anton Nemkin.

As of October 2023, both in Russia and around the world, there is a determination of possible threats posed by AI, as well as the development of measures to eliminate them, the deputy noted. {{quote 'This is a fairly large-scale and painstaking work that requires constant balance. It's important to understand that the AI industry is in its infancy. Therefore, its normative "consolidation" could potentially lead to a slowdown in development. Let me remind you that some countries have already encountered such a practice, in which, as a result of excessive regulation of AI technologies, investment potential has significantly fallen. Therefore, this issue must be approached carefully and sometimes slowed down, - explained Anton Nemkin. }}

Therefore, it is necessary to determine at least a framework understanding of how we will overcome them. For example, in the context of personal data, it is necessary to present clear standards, as well as monitor their implementation. Unfortunately, even without AI technologies in our country, leaks are a colossal problem. I think that the development of AI should take place in the reliance on cybersecurity: without the second, there cannot be the first. The issue of monopolization is also very difficult and there is room for a big discussion, including in the context of stimulating small and medium-sized companies to develop AI from the state, the deputy said.

Let me remind you that in 2022, despite the sanctions pressure, the artificial intelligence market grew by 18%. This is a very positive trend, so this segment should remain under the close attention of all participants in digitalization, - concluded Anton Nemkin.

German Gref spoke about the prospects for using artificial intelligence in banks

Russian banks will actively introduce applications and services based on artificial intelligence as part of a structural transformation aimed at providing new services and improving the quality of customer service. This was announced on July 6, 2023 by the chairman of the board of Sberbank German Gref.

According to him, as the Vedomosti newspaper notes, in the current geopolitical situation and in the context of macroeconomic challenges, financial institutions need to "learn to live in a new way." To do this, in particular, it is necessary to expand capabilities using modern AI systems.

In Sberbank, the share of processes in which AI algorithms are used reached 75% by the beginning of July 2023. Artificial intelligence is involved in making key decisions that a person was responsible for earlier in the bank. Neural networks, among other things, decide which product to offer to a specific client, how to build the optimal route for collection, what to answer the user in a chat bot or call center, how many employees should be in a certain bank branch on a particular day of the week, etc., Gref said.

AI systems can also be used in the field of corporate lending. When servicing legal entities and individual entrepreneurs, "smart" algorithms help structure a credit transaction, analyze risks and check the client's business reputation, as well as make a final decision. Moreover, the quality of the provision of services based on AI in this case turns out to be higher compared to the traditional approach. Sberbank intends to learn how to issue complex structured loans using AI and bring them in volume to 80%. The bank also said that each ruble invested in artificial intelligence brings 6.7 rubles in profit. By the end of 2023, this value may increase to 8 rubles.^[6]

Gentle collector robots and artificial intelligence. How VTB uses digital tools to work with debtors

Since 2018, VTB Bank has been using robotic collectors to work with overdue debts of individuals, gradually expanding this practice. In 2021, this decision accounted for 40% of all voice communications with a client in arrears^[7] At the same time, VTB has no judicial problems in connection with the use of collector robots yet, Vadim Kulik, Deputy President and Chairman of the VTB Board, said at a press briefing on June 6, 2023.

The use of this technology does not pass for everyone without problems. So, at the end of 2022, TAdviser wrote in detail about litigation, one way or another related to the use of collector robots in Sberbank, which has been actively using them since 2017, as well as the existing legislative gap regarding collector robots (see article below). In the case of Sberbank, we are talking in court cases, for example, about the legality of the very fact of the use of robotic collectors, and about the permissible number of interactions between the bank and the debtor with their help.

TAdviser checked: the search for legal information systems does not indeed issue litigation related to the use of collector robots to collect overdue debts at VTB.

Vadim Kulik at a press briefing, answering a TAdviser question, explained what, in his opinion, this is connected with. On average, VTB's customer cut differs from Sberbank's: for various reasons, VTB historically has an average customer check higher, and the share of the "middle +" customer segment is higher. Therefore, the way and manner of collecting debt from VTB Bank is softer, says a top manager of VTB. VTB Bank cannot afford to build an aggressive collection model.

In 2020 and 2021 Russian Forbes issued ratings of 15 best banks for millionaires^[8][9]where VTB took first place both times, ahead of Sberbank. In 2020, in particular, VTB disclosed data on the presence of more than 20 thousand clients with financial assets from $1 million. This turned out to be more than any other participant in the rating.

Vadim Kulik said that in the field of collection direction at VTB there is a large range of different models, not only robots, speech and content synthesis. It all starts with a prediction model, whether the client will be overdue or not. There is also a model of inclination to a particular type of contact - which contact with a client will be more effective.

Then the models come in when a person went into delay or, on the contrary, did not come out and it is better to remind him. Then various robots are connected that either inform or encourage the client to act.

Digital technologies are used when interacting with the debtor not only robots, but also bank employees. To avoid negative situations that may arise during debt settlement, VTB additionally assesses the quality of employees' work, using VTB speech analytics^[10] of^[11].

At the same time, one of the most effective debt settlement tools in VTB is called restructuring programs. Their performance is estimated at 80% - most customers cope with debts and continue to cooperate with the bank.

Digitalization also helps in this. The bank has pre-approved offers that proactively minimize cases of delay. A scoring model based on artificial intelligence shows which of the borrowers needs to be helped, taking into account many factors: the payment discipline of customers, their income, loan term, etc. Based on this, the bank forms individual financial proposals - for example, a reduced monthly payment.

Examples of AI use in the financial sphere in 2023

• At the design level: forecasting the demand for banking products, predicting changes in demand, automated risk assessment.
• At the production level: automation and optimization of interaction with existing and potential customers. Automate document processing and credit approvals.
• At the promotion level: providing personalized offers at the right time. Automatic adjustment of interest rates depending on the customer's history.
• At the service level: development of automated systems and self-service interfaces in all communication channels.

2022

Collector robots will not leave the "gray" zone in any way. Sberbank fights in the courts for their use without the consent of debtors

The bodies of the Federal Bailiff Service (FSSP) and arbitration courts have repeatedly regarded the use of collector robots without a written agreement with the debtor as illegal, follows from judicial practice in the file of arbitration cases. However, Sberbank, which has been using collector robots since 2017, has stubbornly continued to challenge their findings in court. The law on collecting debts from individuals still does not spell out the use of this technology.

TAdviser got acquainted with a part - more than a dozen - of litigation in various instances with the participation of Sberbank recently, one way or another related to the use of collector robots to return overdue debts. In them, the bank, as a rule, disputes the protocols of the FSSP bodies with fines against it on violation of the legislation on the protection of the rights and legitimate interests of individuals when returning overdue debts. We are talking in such cases both about the legality of the very fact of the use of robotic collectors, and about the permissible number of interactions between the bank and the debtor with their help.

Activities to return overdue debts of individuals are regulated by federal law of 03.07.2016 No. 230-FZ (230-FZ^[13]. In the current version, it prescribes that when performing actions aimed at returning overdue debts, the creditor or a person acting on his behalf or in his interests may interact with the debtor using:

• face-to-face meetings, telephone conversations (direct interaction);
• telegraph messages, text, voice and other messages transmitted via telecommunication networks, including mobile radiotelephone communication;
• mail at the place of residence or place of residence of the debtor.

Other than the above, methods of interaction with the debtor may be provided for by a written agreement between the debtor and the creditor or a person acting on his behalf or in his interests.

In addition, the 230-FZ establishes that it is not allowed to interact with the debtor at the initiative of the creditor through telegraph messages, text, voice and other messages transmitted over telecommunication networks, including mobile radiotelephone communications:

• on working days from 22 to 8 hours and on weekends and non-working holidays from 20 to 9 hours local time at the place of residence or stay of the debtor;
• total number: a) more than twice a day; b) more than four times a week; c) more than sixteen times a month.

In one of the latest court cases, the decision on which was made in December 2022, for example, it is indicated that in order to return the overdue debt, Sberbank for a period of less than two months made 198 calls of the collector robot to the debtor's mobile number, on some days the subscriber received 9-11 calls^[14].

According to the Arbitration Court of the Krasnoyarsk Territory, it can be concluded that the bank had a psychological impact on the debtor through numerous phone calls, which violated the privacy of the debtor, caused anxiety and caused a sense of anxiety, which runs counter to the requirements of the 230-F3.

And in this case, the court attributed the calls of the collector robot to "other methods of interaction on the return of overdue debts" and considered it a violation 230-FZ making such calls without a written agreement with the debtor.

A similar case, in which the decision was made by another arbitration court - Udmurtia, but also in December 2022, cites the position of Sberbank regarding the use of robotic collectors^[15]. Calls using a robot collector are voice and other messages transmitted over telecommunication networks that are allowed 230-FZ, and not related to other methods of interaction that require a written agreement, the bank believes.

Sberbank also pointed out the lack of grounds that led the FSSP body to the conclusion that calls to the collector robot are just another way of interaction, and cannot be attributed to voice and other messages transmitted over telecommunication networks.

In addition, from some court decisions we can conclude that when calling, including a collector robot, refusing to communicate, resetting the call, silence in Sberbank is not considered an interaction with the debtor. This follows from some of the bank's responses in open sources regarding its use of an informant within the framework of the 230-FZ^[16]

There is also an explanation of the positions of ships regarding the use of collector robots, which most often do not coincide with the vision of Sberbank. Thus, the Arbitration Court of the Krasnoyarsk Territory believes that the voice informant does not refer to any of the methods of interaction with the debtor established by law, since it is not a telephone conversation or voice message, within the meaning transmitted by these concepts of 230-FZ.

Telephone negotiations are understood as interactions through the use of telephone communications, as a result of which a dialogue is established between two individuals: the creditor's representative and the debtor, within which they exchange information on the issue of returning overdue debts. And voice messages are an instant messaging system in which pre-recorded messages are transmitted through a voice environment in advance. Voice messages are an alternative to voice calls or text messages, the court believes^[17].

The position of the same court regarding what can be considered interaction with the debtor is also unambiguous:

No matter what kind of information was transferred or not transferred during negotiations, in this case it has the purpose of returning overdue debts. The fact that the subscriber hung up or did not want to listen to a company employee at all, and therefore the call duration was several seconds, does not indicate that the interaction with the debtor did not take place and the call was "unsuccessful"[18].

Some court decisions state that when a robot collector calls, the subscriber sometimes does not realize that he was not talking with a human employee of Sberbank.

In the course of interaction regarding overdue debt on a subscriber phone number using a collector robot, the debtor or a third party is not suspected, that the dialogue is not with a human employee of Sberbank PJSC, but with a computer program, having signs of artificial intelligence, which at the same time poses as a person to the interlocutor, asks a third person questions to which he receives answers, and also perceives the speech delivered in response, that is, it behaves like a person, - states another arbitration court - Mordovia.

The Arbitration Court of Mordovia concluded that the use of this method of interaction with the use of artificial intelligence does not apply to any of the methods of interaction with the debtor allowed by the 230-FZ without a written agreement^[19]

Sberbank itself could not answer TAdviser's questions about the use of collector robots.

As of December, the file cabinet contains fresh appeals from the bank against previously issued court decisions in cases involving the use of collector robots.

Sberbank previously said that they began to use collector robots in 2017, and even then noted that "the technology is so worked out that often debtors cannot determine that they are communicating with a robot, and not an employee of the bank."^[20]

In August 2022, RBC published information about the upcoming amendments to the 230-FZ, which would oblige collection companies, banks and microfinance organizations from January 1, 2024, when calling debtors, to warn citizens about who is talking to them - a robot or a living employee^[21].

Also, amendments should introduce a new concept - "automated intellectual agent." It means software for sending voice messages, as well as programs with "generation and speech recognition," which are able to support the conversation.

However, by the end of 2022, there was no new information regarding this draft amendment.

An attempt to remove collector robots from the "gray" legislative zone was also made in 2019. Then the Ministry of Justice came out with a bill on changes in 230-FZ, which, among other things, were supposed to introduce the concept of "automated intellectual agent" ("robot collector")^[22]. By it, the bill meant a tool for sending voice messages transmitted over telecommunication networks, including mobile radiotelephone communications, which uses a speech generation system and supports various arbitrary scenarios of conversations with debtors or other persons, depending on the course of dialogue received from debtors or other persons during the dialogue information. But it did not come to the adoption of this change.

Gartner named 4 main rules for the successful use of AI in the financial sector

On June 22, 2022, Gartner published a study in which it identified 4 main rules for the implementation of artificial intelligence (AI) in the financial sector, which can achieve or exceed the expected effect and ensure the achievement of critical results in the field of finance and business processes.

The use of artificial intelligence in finance departments is still in its infancy, and most companies have only embarked on this in the last two years, said Jacob Joseph-David, director of finance research at Gartner. Most also fail to get the expected returns from such projects quickly.

Given the initial stage of AI development in the financial sphere, CFOs do not have a clear definition and strategy for its successful application. To help CFOs, Gartner has identified four critical rules for the success of AI adoption in the financial sector.

Hire external AI specialists

As a rule, there are three options for attracting specialists with skills and experience in the field of AI: hire new specialists, improve the qualifications of existing specialists, or involve specialists from an existing IT department. Organizations that focus on attracting external employees with AI skills in their talent acquisition strategies are significantly more likely to become leading financial institutions using AI.

IT professionals have indispensable professional skills in working with various technical nuances of AI, which allows the company to overcome difficulties in working with AI applications and reduce the curve of technical training of other employees. Conversely, while improving the skills of financial personnel may be less costly, it risks slowing progress and increasing the likelihood of mistakes. In addition, new IT professionals provide an opportunity to go beyond the usual processes and installations, introducing new ideas to promote the implementation of AI.

Invest in software with built-in AI for quick wins

Acquiring software with embedded AI allows companies to experiment more easily with the use of AI and use it in more areas of the financial sphere; they can simplify the creation of pilots for unique business challenges. In contrast, creating your own AI solutions for all financial processes creates much more work and reduces the opportunity to study new experimental projects or scenarios for their use.

Conduct early trial and diverse experiments

Leading companies in finance are taking an experimental approach to implementing AI early on, rather than making a few big bets. With the number of early pilots increasing, there are more uses for AI and implementation is faster as the organization can focus on the most successful pilots.

Generally, the most successful organizations still explore the same use cases as the less successful ones, with the most common being the three: accounting processes, back office processing, and cash flow forecasting. The only exception is customer payment forecasting, which is used by about half of the leading companies, but by very few of the less successful organizations.

Choosing a leader in the implementation of analytical AI

To realize the benefits of AI, CFOs must select the appropriate person to lead the AI implementation. In particular, it can be the head of financial planning and analysis (FP&A) or the head of financial analytics, who will be engaged in the implementation of AI, and not the persons controlling them.

The heads of FP&A and Financial Analytics are successfully meeting the challenge of implementing AI thanks to their strong analytical training and work with data. They rely less on knowledge of traditional financial processes and more on understanding the complexities of AI in the business environment.^[23]

2020

Artificial intelligence helps companies boost profits 80% faster

Organizations that use artificial intelligence (AI) and other promising technologies in financial and operational activities increase annual profits 80% faster. This conclusion is made in the global study Emerging Technologies: The competitive edge for finance and operations ("Competitive advantage in financial management and operations"), prepared by Enterprise Strategy Group and Oracle. It was attended by 700 heads of financial and operational functions from 13 ^[24]

The survey showed that in mastering promising technologies such as AI, the Internet of Things (IoT), blockchain and digital assistants, a turning point has come: the results of their application exceed expectations and provide significant competitive advantages.

Organizations implementing new technologies for financial management receive much greater benefits than originally expected.

• The number of errors in the work of financial departments decreased by an average of 37%.
• 72% of AI-enabled organizations reported having a clearer view of overall business performance.
• 83% of executives believe that within five years, AI will make financial closures completely automatic.
• Digital assistants increase productivity by 36% and enable financial analysis to be performed 38% faster.

AI, the Internet of Things, blockchain and digital assistants help increase accuracy, speed and understanding of operations and supply chains. And respondents expect additional business benefits once blockchain applications become widely applied.

• Organizations that use AI in supply chain management note that lead times have been reduced by an average of 6.7 working days.
• The use of IoT data in supply chain management reduces ordering errors by 26%.
• AI helps reduce orders by 25%, inventory shortages by 30%, and production downtime by 26%.
• Organizations that use digital assistants to manage supply chains have been able to increase employee productivity by 28% and analyze speed by 26%.
• 87% of organizations that have implemented blockchain have achieved the planned return on investment (or exceeded it); 82% of companies expect significant business benefits during the year.
• 78% of executives believe that being able to verify supply chain monitoring using blockchain will reduce fraudulent transactions by 50% (or more) over five years.
• 68% of respondents cite improving business intelligence as a key advantage of using new technologies in supply chains.

The vast majority of organizations have already mastered promising technologies. Pioneering companies that use at least three similar solutions have the greatest advantages and are more likely to outperform competitors.

• New technologies have become widespread, and 84% of organizations use at least one of them (AI, IoT, blockchain, digital assistants) in productive systems.
• 82% of organizations using three or more promising technologies are ahead of the competition, while only 45% in organizations that do not use any of them.
• 9.5 times more likely that organizations that have implemented several new technologies will achieve financial and operational accuracy that is highest in the market
• Organizations are 2-3 times more likely to purchase pre-configured solutions based on promising technologies than to develop their own (the percentage depends on the specific solution).
• Almost all respondents (91%) consider SaaS applications a key factor in the development of new technologies.

2019

Digital Bank "Tochka" has introduced artificial intelligence that predicts the blocking of the tax account

On December 12, 2019, a digital bank for entrepreneurs "Point" announced the introduction, artificial intelligence which is able to predict account blocking. Federal Tax Service (FTS) More. here

JPMorgan began using an AI copywriter who writes advertising texts better than marketers

In early August 2019, JPMorgan Chase signed a five-year contract with a startup using artificial intelligence for copywriting. The deal follows a successful pilot trial of the new technology. Read more here.

2018

Banks earned $41 billion on artificial intelligence

In 2018, banks earned about $41.1 billion thanks to the use of artificial intelligence. This amount includes both direct revenues from the implementation of such technologies, as well as the amount of reduced costs and the benefit of improving the efficiency of financial institutions (compared with if they left the same processes and infrastructure). This is evidenced by data from analysts at IHS Markit, released on April 10, 2019.

According to experts, by 2030, commercial AI projects will bring banks a total of $300 billion.

The innovative opportunities that artificial intelligence provides to the financial services sector can lead to dramatic transformations, says Don Tait, lead analyst at IHS Markit. - AI is ready to challenge and blur our concepts of computing and the "ordinary" person. This major change will require companies and governments to develop deep vision and a critical understanding of all the consequences of digitalization and emerging technologies.

Experts say artificial intelligence is revolutionizing the banking sector by identifying fraud in financial transactions based on a predetermined set of rules.

North America remains the largest market for using AI solutions in the banking sector: there, companies earned $14.7 billion on such technologies in 2018. By 2030, the economic effect of the introduction of artificial intelligence in the region will jump to $79 billion, analysts predict.

However, by 2024, the Asia-Pacific region will take the lead, where banks will earn and save about $50.6 billion thanks to AI against $11.5 billion in 2018. By 2030, the figure will rise to $98.6 billion largely due to demand in countries such as China (including Hong Kong), Japan, South Korea and Singapore

However, the introduction of artificial intelligence technologies in the banking industry also has negative consequences - job cuts and personnel redistributions due to improved productivity of financial companies due to AI technologies.

Artificial intelligence will impact tens of millions of jobs in the global financial industry, analysts predict. In the United States, for example, 1.3 million people will be affected by 2030, and in the UK - 500 thousand.^[25]

Among the bank employees who may be affected by the spread of AI, IHS Markit named cashiers, employees of customer service departments, interviewers and clerks, financial managers, supervisors and credit specialists.

But in general, artificial intelligence technologies will change the structure of the financial industry, making the banking sector more humane and intelligent, says Don Tate.

The fact that banks are increasingly using artificial intelligence is confirmed in the consulting company Deloitte. According to a study published in April 2019, 29% of companies in the financial industry operating in different countries use robotic process automation - software that automates monotonous routine work. In this sample, 25% of respondents use such technologies for risk management, 21% for risk reporting, 20% for regulatory reporting.

Big data and analysts have also become a priority for banks - 40% of them use such tools along with artificial intelligence.

About 25% and 19% of companies surveyed said they use machine learning and cognitive analytics (including natural language processing), respectively, to reduce costs and improve the accuracy of operations, while 24% said they use business decision modeling tools.

Wall Street banks start using machine learning to analyze currency markets

On June 29, 2018, Bank of America announced the start of using machine learning to analyze currency strategies. The reason for the research in the field of artificial intelligence, which the bank's analysts began in June 2018, was the unstable political situation in Italy - experts feared that it would affect not only the euro, but also the rest of the European currencies, and this threatens the next financial crisis.

In Bank of America's first study, machine learning algorithms are evaluated on performance with fundamental and overview data, such as those related to government spending and consumer expectations. The task of AI is to make a forecast of the relationship between the euro-dollar currency pair. The team used both supervised learning, where the machine must analyze human-labeled data and identify patterns, and uncontrolled learning, where the person no longer controls the process and gives no guidance to the AI.

Due to the nature of the foreign currency market, predicting its future only on the basis of known situations is quite difficult, so we are trying to attract machine learning for alternative valuation strategies, "said currency strategy specialist Alice Leng, who developed an AI-based market study at Bank of America.

The use of machine learning for complex analyses is not an innovation in the financial sphere. But foreign exchange markets still pose a particular challenge for AI algorithms, according to Vasan Dhar, a professor of computer science at New York University and founder of SCT Capital Management - a hedge fund that has relied on machine learning applications for two decades. The complexity and diversity of macroeconomic factors that can influence interrelational relations can significantly complicate analysis in this area, unlike conventional exchange markets that have long applied AI and machine learning.

Despite the active use of AI, most banks have not yet managed to implement it in their work at the global level. In a digital banking report in the fall of 2017, the vast majority of financial institutions noted that they used machine learning to some extent, but, as analysts say, only less than 20% went beyond the simplest methods of working with AI.

Among the three largest US banks, Bank of America was the first to include the development of machine learning models in the publication of foreign exchange research results. The research group of financial holding JP Morgan studied machine learning applications, but has not yet decided to use them. Banking company Wells Fargo says it takes a fundamental economic approach to analyze currency markets because it trusts its expertise in the field. Many do not trust computers that analyze information in ways that a person cannot understand, and argue that they are not ready to accept the predictive conclusions of AI processing data outside of cause-and-effect relationships.

However, changes are already coming - for example, Morgan Stanley, a commercial bank, hired Michael Kearns, a professor of applied computer science at the University of Pennsylvania, who previously worked at a hedge fund, to expand the use of AI, and the Deutsche Bank team included machine learning in the analysis of its data.

Some analysts argue that thanks to the public availability of machine learning tools, Wall Street research will lose its relevance as investors can develop their own AI-based analysis techniques. But Peter Wadkins, an analyst at FX Aite Group, believes that this is not as likely as it seems, because machine learning requires quite large amounts of data and high-tech methods of processing them.^[26]

How collectors use artificial intelligence to knock out debts

By June 2018, Chinese collectors began to actively use new technologies, for example, artificial intelligence, in order to collect debts that are supposed to arise due to the speculative $200 billion credit bubble that has formed in the lending industry between individuals in the country.

From 2013 to 2018, thousands of new companies appeared in China, which acted as intermediaries between private creditors and people in need of cash. However, the scandal has left these companies in the crossfire of regulators, and since mid-2017, when the Chinese government imposed controls on the provision of loans, as well as the licensing of creditors and intermediaries, so many such companies that provided their services as individuals have completely ceased operations.

According to the estimates of the analytical online firm Wdzj.com the outstanding debt between individuals as of May 2018 amounted to more than $200 billion, and the growing number of refusals from paying the debt opened the door to a wave of startups based on the latest technologies with which creditors are trying to restore the issued funds, reports the Financial Times.

Lending between individuals is widely used in China, but the government only closely monitors the official banking system, notes Cherry Sheng, chief executive of Shanghai-based debt collection firm Ziyitong and former manager of Citigroup and ANZ Bank. However, thanks to the emergence of advanced technologies, even individuals have the opportunity to repay the debt.

Ziyitong, which has managed to recover about $29 billion in debt since it opened in 2016, recently launched an artificial intelligence-based platform to recover overdue loans. Ziyitong's customers are roughly 600 debt collection agencies and more than 200 lenders, including Alibaba Group and Postal Savings Bank of China, according to Cherry Sheng.

The system analyzes the data about borrowers and their friends available on the Internet, and then contacts the borrower by phone using a dialog robot. Conversations are recorded and analyzed using an algorithm that then determines the wording that is most likely to take effect on the borrower and force the debt to be repaid. The system also contacts his friends and with their help asks the borrower to return the money.

As of May 2018, the AI-based system used by Ziyitong demonstrated a very high recovery rate of 41% for large customers on loans overdue for up to one week, Cherry Sheng said. For comparison, the effectiveness of traditional collection methods for returning debts on similar loans is only 20%. Ziyitong also plans to use the AI system to return loans more than one week overdue.

Yigou, another start-up for debt collection, has launched a mobile phone app that allows collectors to search thousands of individual debt records and select needed cases, making it easier to interact between lenders and collectors. The company can also provide geolocation data to some borrowers to help collectors track their location.

Wen Yong, chief executive of Yigou, noted that the latest technologies have begun to play a significant role in the collection industry. According to him, many companies providing lending services between individuals were forced to organize their own collection cells, since the number of cases of unpaid debts in this sector has grown significantly.

Given that regulators are not abandoning attempts to intercept the flow of cash from shadow banking and asset managers, which ensure that lending funds are filled between individuals, collectors of such companies expect more borrowers to evade loan refunds by the end of 2018. Since individuals do not report on their activities, it is difficult to accurately determine the amount of debt, but collectors assess the situation as disappointing.^[27]

2017

Replacing thousands of employees with robots in Japanese banks

At the end of October 2017, it became known about the plans of leading Japanese banks to automate about 30 thousand jobs, since, according to companies, the traditional business model no longer allows increasing profits.

According to the Japanese business publication Nikkei, Mizuho Financial Group is going to replace about 8 thousand employees with computers by fiscal 2021, and 2026 - to increase this figure to 19 thousand.

Another major financial institution from Japan, Sumitomo Mitsui Financial Group, is preparing for large-scale automation. According to her plans, by fiscal 2020, robots will perform tasks for which 4 thousand people are needed by October 2017.

Bank of Tokyo-Mitsubishi UFJ is also not far behind competitors. The plans of this financial corporation include automation of 9,500 working positions by fiscal year 2023. For many Japanese companies, the fiscal year ends at the end of March.

By using computational algorithms instead of people, Mizuho Financial Group expects to consolidate clerical work, minimizing the number of personnel with duplicate functions.

Also, about 100 routine work tasks will be taken over by a new robotic processing system, which Mizuho Financial Group first used only to enter data when opening investment accounts on its website.

However, large-scale digitalization does not imply only a reduction in the staff of the Mizuho Financial Group. For example, in the fall of 2017, about 200 back-office employees whose functions were replaced by computers were transferred to customer service departments. In addition, Mizuho Financial Group intends to increase the number of financial technology specialists.

Sumitomo Mitsui Financial Group plans to digitize some of the services provided by banking branches. By October 2017, the company had opened nine data centers in Japan that will process new data.^[28]

AI "Robot Vera" for the selection of candidates for vacancies

AI "Robot Vera" is a service for automated selection of candidates for vacancies. The service is based on machine learning technology, is able to "understand" a person's natural speech and process more than 10 thousand calls at the same time, thus speeding up the candidate selection process. According to the founder of Stafory (Stafori) Vladimir Sveshnikov, the entire hiring process thanks to the service is reduced to three hours.

How artificial intelligence is changing banks. 6 trends from Mikhail Khasin, Senior Managing Director of Sberbank

In his speech at TAdviser SummIT 2017, Mikhail Khasin, Senior Managing Director of the Technologies block of Sberbank, told how artificial intelligence (AI) becomes a driver of technological innovations in banks. Read more here.

R-Style Softlab Study

Only every fifth domestic bank uses this technology, but the vast majority of banks consider it promising. Half of the surveyed organizations are ready to transfer payment transactions and information services to. messengers In every third bank they are ready to entrust chat bots with the functions of blocking payment cards, in every fifth - confirmation of transactions. This is the data of the study, R-Style Softlab which took place from February to April 2017, it was attended by heads and specialists of IT and business divisions of 100 banksRussia and the CIS, more than half of which banks are top 100 categories.

The increase in the number of Russian Internet users, the availability smartphones and further development of mobile Internet are forming new habits and patterns of behavior. Users of social networks and mobile applications are increasingly focused on obtaining an instant result and performing a targeted action in a couple of clicks, which largely explains the rapid rise in popularity, and messengers WhatsApp. Viber Telegram Messenger

However, the need for high-quality financial services and personal consultations has not disappeared anywhere: people still call call centers. Despite the development of RBS systems, the number of phone calls, according to representatives of the 30 largest credit institutions, has recently increased significantly.

Chatbot technology allows you to optimize business processes and find a reasonable compromise in solving several diverse tasks at once: simplify user interaction with the bank, increase the level of service and reduce the financial costs of the call center and SMS notification services. The dialogue is simulated in the usual and comfortable chat environment for the client, while he receives a choice of services previously available only on the site or through the RBS system - all this allows him to maintain and increase loyalty.

Unfortunately, at the moment, full text recognition and processing of arbitrary requests from the interlocutor using artificial intelligence technologies cannot be brought to an acceptable level.

The spread of promising, according to many companies, so-called interface bots created on Telegram and Facebook platforms does not solve the issue of high-quality imitation of a live conversation and maintaining customer loyalty. "Conversational" bots, primarily their primitive versions created for an entertaining purpose, are quite often criticized due to the limitations of the topics on which they are able to conduct dialogues.

Since the feeling of live contact is important for a person when discussing issues with a bank, the most correct direction is the development of "conversational" bots, provided that they have wide opportunities for language analysis. With thoughtful implementation, they can be called "correct" chat bots that can qualitatively imitate human speech.

Such a decision will seriously reduce the load on the call center, preserve the possibility of a live dialogue and allow in difficult cases to transfer the conversation to a bank specialist, helping him in solving the problem - the function of offering prompts to the operator from the database of template phrases is activated.

2016: AI to recruit on Wall Street

On June 7, 2016, Reuters published an article on how Wall Street banks, in an attempt to cut costs, turn to software developers to help streamline the process of finding suitable employees. The bet is on artificial intelligence (AI).

Such technologies allow you to identify in applicants useful qualities for the employer, including the ability to work in a team, dedication, willpower and other advantages that can not always be found in the resume or during an interview.

Financial giants such as Goldman Sachs Group, Morgan Stanley, Citigroup and UBS Group are considering putting AI into operation. For example, Citigroup is testing a technology developed by Koru Careers to drop out candidates by early June 2016. Software is tested on a small group of employees working in corporate and investment structures.

The program defines the "corporate imprint" of the business (a set of qualities of active employees on which the company's high performance depends) and evaluates the qualities of candidates based on an analysis of a short video in which applicants talk about their strengths and career aspirations. The system takes into account not only the speech of the speaker, but the way the presentation is presented, including "body language " and the pace of conversations. Koru allows testing via the Internet, mobile phone or on a local computer in the office where the job seeker came.

Users of Koru software pay developers to draw up a "corporate fingerprint," as well as for each candidate who is tested. Koru claims that the software offered by the company can reduce the number of unsuccessful hiring by 60%.

Until recently, technology only helped find the best resume, but now they will be able to truly understand the people who applied for work, "said Mark Newman, head of HireVue, which is developing an AI platform for evaluating candidates for video conversations when hiring.

The banks hope that such developments will help get rid of expenses in the event of problematic hiring and improve the situation in the labor market. Artificial intelligence, according to financiers, will allow you to select employees who can cope with a particular job, thanks to the creation of templates built on the analysis of large amounts of data.

Hiring a bad employee can cost the company dearly - leading to a lot of financial spending and losing business opportunities. According to Capital One Financial experts, losses from an unsuccessfully hired employee can be measured by three salaries of a person who would be ideal for this position.

Recruiting and human resources software developers are looking to rid their customers of human errors, such as dropping out strong candidates who at first glance seemed weak, says Matt Doucette, director of recruiting at Monster Worldwide.

A top salesman is not usually someone who plays the audience, but someone who sits modestly in a corner who avoids attention and asks the right questions, "Doucette said.

According to knowledgeable Reuters sources, UBS Bank uses a computer algorithm that allows analyzing resumes to find candidates with the necessary parameters, as well as technology for selecting strong candidates.

Goldman Sachs Group uses its own software to search the resume for the right qualities, such as teamwork, honesty and judgment. The company also uses personality tests to better understand the qualities of the most successful bankers and traders.^[29]

Artificial intelligence is involved not only in the American banking sector, but also in the Russian sector. In early 2016, the Russian company Krawlly and iBank Global introduced a personal financial assistant capable of aggregating data from different banks, categorizing spending and giving personal advice based on big data analysis. Software leveraging AI capabilities helps offer various partner cash investment programs to banking clients.^[30]

Next-generation banking platform

Chatbots and Roboediting

Modern chatbots can:

• Informing about the features of products and services
• Providing contact details
• Payment transactions
• Financial recommendations to the customer

• show rates and exchange currency
• carry out accounting of personal finances
• transfer from card to card
• send applications for trading and Internet acquiring and check the counterparty according to TIN/OGRN (IP)
• answer user questions

Robo-Advisors as a promising example of AI application

Roboedifying has become an alternative to financial advisers on banking issues, specific purchases and other cash transactions online.

Robofacers provide great advantages in the field of online trading. First of all, these are applications in one click and opening an account in real time, monitoring, current news and processing large volumes of transactions at once. The proliferation of brokers on social networks makes investment knowledge more accessible and understandable, and communication with the client is simple and targeted.

Automation allows you to present information in 24/7 mode, while reducing process costs. Roboedvisers are available on the desktop or in the format of mobile applications, carry the functions of a portfolio manager, determining risks and an optimal investment strategy.

Individual Offers and Loyalty Boost

• Recommendations of banking products and purchases (loyalty programs from various retailers), including using knowledge about the client from social networks
• Definition of B2B customer connections with subsequent recommendations of new counterparties
• Simulating financial risks for small businesses (default, cash gap) in real time with recommendations for targeted strategies and products

IoT (Internet of Things)

• Manage and track the use of leasing assets
• "Smart" insurance for retail clients (medicine, car loans)
• Smart Home + Daily Shopping: Ordering groceries, paying utility bills, signing up for TV content

Antifrod. External and insider threats

• Signs of a third-party use of a customer's plastic card
• Signs of the so-called "droppers" based on the nature of receipts and transactions in the Internet bank and ATMs
• Identification of fictitious salary projects (loans, cashing)
• Identify unauthorized expense transactions on customer accounts and customer plastic cards
• Errors in parameterization of bonus programs on plastic cards, which lead to "cheating" and damage
• Cash cashing schemes, including the use of an Internet bank and plastic cards
• Abuse of conversion transactions for both individuals and legal entities
• Unauthorized connection of the Internet Bank to the customer's accounts and issuance of plastic cards without the customer's knowledge
• Unauthorized credit card limit increases

Operational efficiency

• Identify and automatically correct transaction variances
• Natural Language Processing Algorithms for Analysis and Generation of Claims
• Monitoring and forecasting of infrastructure failure (ATMs, IT resources)
• Optimization of cash turnover and balances at cash registers and ATMs. Optimizing Collection Services
• Optimization of staff search and recruitment (resume analysis and initial selection)
• Real-time voice analytics for call centers and offices (consultation quality management)

Artificial intelligence in Sberbank

Main article: Artificial intelligence in Sberbank

Robotics