Revista de Ciencias Sociales (RCS)
Vol. XXX, Núm. 3,
julio-septiembre 2024. pp. 22-36
FCES - LUZ ● ISSN: 1315-9518 ● ISSN-E: 2477-9431
Como citar: Corral, D., García,
M., y Carvajal, M. (2024). Hybrid artificial intelligence: Application in the
banking sector. Revista De Ciencias Sociales, XXX(3), 22-36
Hybrid artificial intelligence: Application in the
banking sector
Corral de La Mata, Daniel*
García de Blanes Sebastián, María**
Carvajal Camperos, Marisol***
Abstract
The integration of
smart technologies, from data mining to Artificial Intelligence, has
revolutionized the way financial institutions manage and use information. This
progress has driven the development of hybrid artificial intelligence
solutions, which fuse neural networks, fuzzy logic, genetic algorithms and
intelligent agents, improving efficiency and accuracy in finance. Key
application areas include implementing machine learning for personalized
financial services, using artificial intelligence to improve credit risk
assessments, and automating operations for greater efficiency. This study aims
to analyze the implementation of hybrid artificial intelligence in the banking
sector. The findings suggest that machine learning significantly personalizes
services, increasing customer satisfaction and retention. Artificial
intelligence has refined credit risk assessment, reducing errors and improving
accuracy, while AI-enabled automation has streamlined operations. In addition,
artificial intelligence helps analyze trends and innovate products. The
combination of traditional and data-driven artificial intelligence techniques
was identified as offering significant competitive advantages for financial
institutions.
Keywords:
Artificial intelligence; banking sector; automation; efficiency; financial
operations.
Inteligencia
artificial híbrida: Aplicación en el sector bancario
Resumen
La
integración de tecnologías inteligentes, desde la minería de datos hasta la Inteligencia
Artificial, ha revolucionado la forma en que las instituciones financieras
gestionan y utilizan la información. Este progreso ha impulsado el desarrollo
de soluciones de inteligencia artificial híbridas, que fusionan redes
neuronales, lógica difusa, algoritmos genéticos y agentes inteligentes,
mejorando la eficiencia y la precisión en las finanzas. Las áreas clave de
aplicación incluyen la implementación del aprendizaje automático para servicios
financieros personalizados, el uso de la inteligencia artificial para mejorar
las evaluaciones de riesgo crediticio y la automatización de las operaciones
para una mayor eficiencia. Este estudio tiene como objetivo analizar la
implementación de la inteligencia artificial híbrida en el sector bancario. Los
hallazgos sugieren que el aprendizaje automático personaliza significativamente
los servicios, lo que aumenta la satisfacción y la retención de los clientes.
La inteligencia artificial ha refinado la evaluación del riesgo crediticio,
reduciendo los errores y mejorando la precisión, mientras que la automatización
habilitada por ella ha agilizado las operaciones. Además, la inteligencia
artificial ayuda a analizar tendencias y a innovar productos. Se identificó que
la combinación de técnicas de inteligencia artificial tradicional y basada en
datos ofrece ventajas competitivas importantes para las instituciones
financieras.
Palabras
clave: Inteligencia artificial; sector bancario; automatización;
eficiencia; operaciones financieras.
Introduction
The
evolution of Information and Communication Technologies (ICT) has been
transforming the banking sector over the decades. From the rise of databases in
the 1980s and 1990s to the rise of Big Data, Artificial Intelligence (AI) and
the Internet of Things in the 2010s, the banking sector has been part of a
substantial transformation in the way it manages information and makes critical
decisions (Fernández, 2019).
Artificial
Intelligence (AI) has emerged as a transformative technology in the banking
sector, driving digitalization and innovation in a competitive environment
marked by the growing presence of fintechs and digital banks. Its application
ranges from improving customer service through chatbots and virtual assistants
to optimizing fraud detection and credit risk assessment (Moposita &
Jordán, 2022; Rodríguez, 2023). In addition, AI enables the automation of
operational processes, increasing efficiency and reducing human errors (Arbeláez-Campillo, Villasmil & Rojas-Bahamón, 2021;
Maita-Cruz et al., 2022). However, its adoption presents significant
challenges, including the need for large investments and ethical considerations
regarding data privacy. Despite these challenges, AI promises to transform
financial services, offering more personalized and secure experiences to
customers.
In
this context, ICT and intelligent techniques, such as data mining, neural
networks, expert systems, fuzzy logic, genetic algorithms, and intelligent
agents, have been integrated to form hybrid AI systems in the banking sector,
combining multiple artificial intelligence approaches to optimize efficiency
and accuracy in decision making and process automation.
This
research paper aims to analyze the implementation of hybrid artificial
intelligence in the banking sector. This paper uses a theoretical-bibliographic
research method, focused on the review and analysis of existing literature on
the implementation of hybrid artificial intelligence in the banking sector.
Relevant academic and technical sources were selected and analyzed to explore
the applications of machine learning for personalized financial services,
improving credit risk assessments using AI, and automating operations for
greater efficiency. This approach allows us to synthesize previous knowledge
and critically assess the advances and challenges in the adoption of hybrid AI
systems in banking.
This
paper aims to analyze the implementation of hybrid artificial intelligence in
the banking sector, addressing three key areas: Machine learning for
personalized financial services, improving credit risk assessments through AI,
and automating operations for greater efficiency. Through a
theoretical-bibliographic review, the advances, benefits, and challenges
associated with these applications will be explored, providing a comprehensive
view of the impact and potential of hybrid AI systems in modern banking.
1. Methodology
This
study is based on a theoretical-bibliographic research method, focused on the
exhaustive review and critical analysis of the existing literature on the
implementation of hybrid artificial intelligence in the banking sector. To this
end, relevant academic and technical sources have been identified and examined
that address various applications of AI, such as machine learning to
personalize financial services, improving credit risk assessments, and
automating operations to increase efficiency. This approach allows us to
synthesize the accumulated knowledge in these areas and critically assess both
the progress made and the challenges that still persist in the adoption of
hybrid AI systems in the banking industry.
Three
leading academic databases are incorporated: Web of Science, Scopus, and Google
Scholar. This integrated strategy was designed to leverage the distinctive
advantages offered by each database, ensuring a comprehensive and nuanced
understanding of our research topic.
Web of
Science and Scopus were chosen due to their rigorous selection of peer-reviewed
journals and inclusion of high-impact research. These databases offer advanced
bibliometric tools, facilitating detailed citation analysis and trend
identification, allowing for a nuanced review of the literature. Their rich
metadata and analytical functions allow for tracking the evolution of research
topics, identifying seminal works, and assessing the scholarly influence of
various publications.
Complementarily,
Google Scholar, with its broad reach into scholarly materials such as theses,
books, conference papers, and grey literature, enabled a broader review of the
literature. Its open access and algorithmic retrieval of relevant content
ensured that no important research was overlooked, especially that outside
traditional scholarly publishing channels. This approach ensured a
comprehensive survey of existing research and also allowed for critically evaluating
and synthesizing findings from a wide range of sources.
2. Artificial
Intelligence: Evolution, components and characteristics
Information
and Communication Technologies (ICT) encompass a broad spectrum of digital
technologies, tools and systems that enable the manipulation, transmission and
presentation of information in multiple formats. This field ranges from
traditional networks to digital media and mobile technologies, with its
evolution marked by different historical stages. In the 1980s and 1990s,
relational databases and systems such as SQL were consolidated, laying the
foundations for the management of large volumes of data. In the 2000s, the
explosion of digital data drove digitalization, e-commerce and social networks,
facilitating the processing of Big Data through tools such as Hadoop (Bialecki
et al., 2005). The decrease in data storage costs was significant, allowing
companies to handle massive volumes of information.
In the
2010s, the rise of Big Data was reflected in greater investment in technologies
for its use, highlighting the use of tools such as Apache Spark for real-time
analysis (Zaharia et al., 2016). The proliferation of connected devices in the
Internet of Things (IoT) and the development of Artificial Intelligence (AI)
and Machine Learning (ML) increased data generation and the demand for advanced
analytical solutions.
Finally,
in the 2020s, Edge Computing emerged as a key strategy, bringing data
processing closer to its source of origin, especially relevant for IoT and
mobile devices. The integration of AI and ML into Big Data and real-time
analytics solutions, together with the development of edge artificial
intelligence, has been instrumental in meeting the growing demand for computing
services and improving efficiency in data management (Zhou et al., 2019).
Database-
and AI-based technology offers techniques to capture both individual and
collective knowledge, thereby expanding the knowledge base of organizations.
Data mining helps to develop knowledge from large databases, providing new
insights to improve performance and operational strategies. AI techniques such
as expert systems, fuzzy logic, genetic algorithms, and intelligent agents
automate routine tasks and capture tacit knowledge, generating solutions to
complex problems and improving business decision-making.
Data
mining is a technique for identifying patterns and extracting hidden
information from large data repositories such as databases and data warehouses
Bramer (2016). It involves using mathematics, statistics, computer science, and
other methods to process a large amount of information in order to derive
useful conclusions and provide valuable decisions for people (Zhang et al.,
2022). Data mining plays a crucial role in discovering hidden patterns and
relationships within large data sets, thereby providing valuable insights
across various domains and supporting informed decision-making processes
(Bellazzi & Zupan, 2008).
The
application of data mining extends to diverse areas such as e-learning systems,
customer behavior analysis in the banking sector, and customer churn prediction
in telecommunications, demonstrating the wide scope of its applications (Han et
al., 2007; Shaheen et al., 2010; Thongsatapornwatana, 2016; Doğuç, 2022).
On the
other hand, neural networks are computational models inspired by the structure
and function of the human brain, consisting of interconnected nodes that work
together to process complex information and learn patterns (Prieto et al., 2016).
The development of deep learning algorithms and architectures has significantly
accelerated progress in neural network research (LeCun, Bengio & Hinton, 2015).
These networks have found applications in various fields including mechanics,
hydroelectricity, structural engineering, and computer science. The solutions
provided by neural networks appear to be cutting-edge from the versatility in
mechanical data for physics (Koeppe et al., 2022), in the prediction of water
inflows to hydroelectric reservoirs (Berdnikov, 2021), or the diagnosis of
faults in high-speed train bogies (Zhao, Guo & Yan, 2017).
The
architecture of neural networks typically includes different types of layers,
such as convolutional, pooling, and dense layers, which contribute to their ability
to effectively process and analyze data. Moreover, the development of new
neural network algorithms, such as the optimized backpropagation algorithm
based on genetic algorithms, has addressed challenges related to local minima
and convergence rates (Ding, Su & Yu, 2011). Furthermore, artificial neural
networks led to the development of wavelet neural networks, which offer
alternative approaches to traditional nonlinear activation functions (Zhang et
al., 2022), a transformation in network design.
Artificial
Intelligence (AI) involves the simulation of human intelligence in computer
systems and machines, with the aim of addressing complex problems and improving
the efficiency of knowledge management. AI relies on processes such as machine
learning, reasoning ability, and self-correction, allowing machines to acquire
information, apply rules to draw conclusions, and continuously adapt. This is
essential for AI to perform tasks that normally require human intelligence,
such as problem solving and decision making Haenlein & Kaplan (2019). AI
enables the automation of repetitive tasks, increasing efficiency and freeing
up time for strategic activities.
Furthermore,
the application of AI in knowledge management is evident in its ability to
process and analyze large amounts of data quickly and accurately. AI techniques
can unravel complex patterns in massive data sets, further expanding the scope
of AI in knowledge management (Nabi, Bansal & Xu, 2021).
Machine
learning (ML) is a subset of Artificial Intelligence (AI) that focuses on the
development of algorithms and models that enable computer systems to learn and
make predictions or decisions based on data (Tyagi & Chahal, 2022). It
involves building systems that can automatically learn and improve from experience
without being explicitly programmed. ML techniques enable computers to identify
patterns, extract meaningful insights, and make decisions, thereby improving
their ability to perform specific tasks more accurately over time (Rahmani et
al., 2021).
ML has
found applications in several domains, including medicine, networking, and
materials science. In medicine, ML has been used for tasks such as predicting
mortality risk and analyzing medical images for diagnostic purposes (Rahmani et
al., 2021). In networking, ML has been leveraged to optimize network
performance and security (Wang et al., 2018).
ML
techniques encompass a wide range of approaches, including supervised learning,
unsupervised learning, and reinforcement learning, each suited to different types
of problems and data. These techniques have the potential to transform
industries and drive innovation by enabling systems to autonomously learn and
adapt to complex challenges.
An
expert system is a computer system that emulates the decision-making capability
of a human expert. It is designed to solve complex problems by reasoning
through bodies of knowledge, represented primarily as “if-then” rules, rather
than through conventional code. Expert systems are built from knowledge
obtained from human experts, traditionally (Alonso et al., 2012). These systems
have been developed and applied in diverse domains, including law, medicine,
agriculture, engineering, and business, to provide intelligent decision support
and problem-solving capabilities (Susskind, 1988; Mansingh, Reichgelt & Osei, 2007).
The
development of expert systems involves the acquisition and representation of
expert knowledge, usually in the form of rules, and the use of inference
engines to reason and make decisions based on this knowledge. Expert systems
have been used for tasks such as fault diagnosis in power generation, voltage
and VAR control in large-scale power systems, and pest and disease management
in agriculture (Le, Negnevitsky & Piekutowski, 1997; Mansingh et al.,
2007). They have also been applied in international marketing, image processing
and electronic surveillance in medical oncology, showing their versatility in
different domains (Çavuşgil & Evirgen, 1997).
The
use of expert systems has been driven by the need to capture and leverage
expert knowledge to solve complex problems, especially in situations where
human experts may not be readily available. These systems have demonstrated the
potential to improve decision-making processes, automate tasks, and provide
valuable insights in a number of fields, making them a valuable tool for
augmenting human expertise and addressing complex challenges.
Fuzzy
logic is a system that can handle both numerical data and linguistic knowledge
simultaneously (Klement & Slany, 1993). It has been widely adopted in
criticality assessment models for failure modes and effects analysis, where it
efficiently helps formulate effective criticality assessments of potential
causes of failure (Mendel, 1995). Furthermore, fuzzy logic has been used in various
fields such as medical sciences to diagnose disease risk (Braglia, Frosolini & Montanari, 2003), artificial intelligence to create music
(Ivančan & Lisjak, 2021), and, thanks to its ability to handle approximate
reasoning modes, it has been used in the analysis of corporate social
responsibility.
The
concept of fuzzy logic is based on the use of intuitionistic fuzzy logic to
interpret perceptions and solve vague problems. It also has its roots in the
concept of fuzzy sets and the amalgamation of multi-valued continuous logic
systems (Zadeh, 1996). Fuzzy logic is a subtype of multi-valued logic and can
be used in combination with other types of controllers such as PI, PID, neural
networks, and genetic algorithms (Tamir, Rishe & Kandel, 2015). Genetic
algorithms (GAs) are optimization methods based on biological evolution and
genetics, widely used in diverse fields such as computer science, artificial
intelligence, and mathematics (Sivanandam & Deepa, 2008).
GAs
are known for their ability to perform global searches in complex and large
search spaces, making them effective in scenarios where conventional
optimization methods fail to produce the desired results Gen & Cheng
(1999). They are stochastic search algorithms that use genetically inspired
operators to transform potential solutions into offspring populations, allowing
the identification of optimal or near-optimal solutions (Koshka & Novotny,
2020).
Furthermore,
GAs have been recognized for their robustness and simplicity, making them
suitable for solving optimization problems, particularly in cases where the
search space is large and complicated, and conventional optimization methods
are not effective. They have also been integrated with other optimization
techniques such as fuzzy C-means clustering to address specific problems such
as vehicle routing for autonomous driving (Zhu, 2022). Furthermore, GAs have
been combined with other artificial intelligence methods, including neural
networks and fuzzy logic controllers, to create unified software platforms for
solving complex problems.
Intelligent
agents, within the context of artificial intelligence, refer to autonomous entities
that interact with their environment through observations and actions, aiming
to achieve specific goals with the help of rewards. These agents have been
widely used in various domains, including sales training with the use of
artificial intelligence (AI) coaches to improve the skills of sales agents (Luo
et al., 2021). Furthermore, intelligent agents have been employed in the
product design, process design, and production stages, demonstrating their
applicability in solving complex problems (Wang et al., 2020).
Furthermore,
intelligent agents have been associated with the concept of rogue agents, which
are capable of opposing assigned goals or plans, as well as the attitudes or
behaviors of other agents (Coman & Aha, 2018). Collaboration of intelligent
agents through intelligent interfaces has been proposed, highlighting the
potential of combining intelligent agents at the software level through
standard interfaces for communication (Bryndin, 2019). Furthermore, progress in
technology and processing power has enabled the development of sophisticated AI
agents, further emphasizing their importance in AI systems (De Vreede, Raghavan
& De Vreede, 2021).
In the
field of Predictive Analytics and Decision Making, banking institutions have
adopted technologies such as Machine Learning (ML) and Big Data to improve
credit assessment and decision making (Angelini, Di Tollo & Roli, 2008;
Mhlanga, 2021). Credit risk assessment, which refers to the likelihood of a
borrower defaulting on financial obligations such as repaying a loan, has
evolved with the advent of AI. Previously, credit risk management was based on
statistical models and human judgment. However, with AI, financial institutions
have more powerful tools to assess and manage risks more accurately:
a. Predictive
analytics: AI uses advanced models such as support vector machines and neural
networks to predict a borrower's behavior. These models are trained on large
data sets that include borrower financial information, payment history,
transactions, among others, allowing them to accurately predict the likelihood
of default.
b. Unstructured
data processing: Unlike traditional models that primarily use structured data
(such as income, credit history), AI can analyze unstructured data, such as
social media texts, phone records, or online behaviors, to gain additional
insights into a borrower’s financial strength and trustworthiness.
c. Decision
automation: Once set up, AI models make decisions in real time. For example,
when applying for a credit card online, AI can assess the application and
approve or reject it in a matter of seconds.
d. Adaptability:
AI models have the ability to learn continuously. As more data is fed in, the
models adjust and improve, allowing them to adapt to new market conditions or
emerging patterns in borrower behavior.
e. Advanced
segmentation: AI makes it possible to segment borrowers into more specific
groups based on similar characteristics. This makes it easier to tailor
specific credit products to niche markets and define more precise risk
management strategies.
f. Bias
reduction: While AI models can perpetuate biases present in the data they are
trained on, with proper design and oversight, AI has the potential to reduce
human biases in credit decision-making, leading to more objective decisions.
g. Improved
portfolio management: Financial institutions can use AI to monitor the health
of their loan portfolios in real-time, identify emerging trends, and
proactively adjust strategies.
h. Integration
of diverse data sources: AI enables the integration of diverse data sources,
such as credit bureaus, banks, non-bank financial institutions, social media,
among others, to obtain a more complete view of an individual's credit profile.
In the
area of fraud detection, AI and ML-based systems identify anomalous patterns
in financial transactions, enabling real-time fraud detection and prevention
(Aslam et al., 2022). Financial fraud is a persistent and ever-evolving problem
that represents significant losses for institutions and their customers. With
the integration of AI and ML, the ability to detect and prevent fraud has taken
a qualitative leap, allowing institutions to anticipate and act with
unprecedented speed (Ryman-Tubb, Krause & Garn, 2018).
As for
considerations, it is important to take into account data privacy, since the
efficiency of fraud detection by AI depends on the analysis of large amounts of
data, always in compliance with privacy regulations. Likewise, given the increasing
technological dependence, it is essential to have contingency plans in case of
system failures or cyberattacks. Despite the efficiency of these systems, human
intervention and review remain crucial, especially in complex or ambiguous
cases.
3. Hybrid systems: Practical examples
Having
gone through the above, it is important to emphasize the concept of hybrid
artificial intelligence systems (see Figure I). These systems combine various
AI techniques and approaches, such as neural networks, expert systems, fuzzy
logic, genetic algorithms, and intelligent agents, to create more powerful and
versatile solutions than systems based on a single AI technology (Medsker,
2012). These hybrid systems offer a number of advantages over individual AI
solutions, such as a greater ability to handle ambiguous information, adapt to
changing situations, and provide more accurate and effective solutions in a
variety of contexts (Khosla & Dillon, 2012).
Source: Own elaboration, 2024.
Figure I: Hybrid AI systems
Within
the panorama of leading companies in the development of hybrid Artificial
Intelligence (AI) systems, which cover a wide range of services and
technologies, companies such as IBM, Google DeepMind, OpenAI, Microsoft,
NVIDIA, Palantir Technologies and SAS stand out.
IBM,
through its Watson platform, provides solutions that combine neural networks,
expert systems and other AI techniques for various applications in industries
such as health, finance and commerce (Ventura-Fernández, Vidalón-Soldevilla
& Ventura-Fernández, 2021). For its part, Google DeepMind, a subsidiary of
Alphabet, has created systems capable of learning autonomously and applying
that knowledge to complex problems (Powles & Hodson, 2017). OpenAI,
recognized for its advances in language models such as GPT-3, is also dedicated
to the research and development of hybrid AI systems, covering everything from
deep learning to planning (Julianto et al., 2023).
Microsoft,
through its Azure AI platform, offers a wide range of tools and services that
integrate multiple AI techniques, from neural networks for natural language
processing to expert systems for business decision-making (Tiutiunnyk &
Rybachok, 2021). NVIDIA, famous for its graphics cards, is also involved in AI
development with its AI and Deep Learning platform used by researchers and
companies for the creation and implementation of hybrid AI models (Gilman &
Walls, 2021). Finally, Palantir Technologies, for its part, focuses on the
development of software that integrates AI elements for the analysis of large
volumes of data, using everything from fuzzy logic to machine learning,
especially geared towards intelligence and analytics applications for
government and corporate clients (Lanzing, 2023).
Recently,
two companies have created specific applications for the banking and financial
services sector that deserve recognition. EPAM Systems is a global software
engineering, information technology and digital design consulting company,
established in 1993 and has grown to become a leading provider of technology
services to companies in the banking and financial sector. It counts among its
clients five of the main investment banks, as well as retail and commercial
banks, payment providers, wealth management institutions, among others.
One of
its notable solutions is Pling, a voice-activated tool built on Microsoft's
Azure platform. Pling allows users to carry out banking operations using voice
commands, making the process easier and eliminating the need to access specific
banking applications.
On the
other hand, Softtek offers various artificial intelligence solutions and
advanced technologies aimed especially at banks and financial institutions.
These solutions cover IT infrastructure and support technologies, automation
and digitization of banking processes, banking customer experiences, digital
banking and payments platforms, remote workforce enablement, Ellenton (Data
Masking) solutions, Fintech solutions, cost optimization IT and operational
intelligence. Softtek focuses on improving the digital consumer experience,
optimizing business processes and providing innovative solutions for the
financial industry.
4. Hybrid
artificial intelligence in the banking sector
The
practical applications of hybrid Artificial Intelligence (AI) systems in the
banking industry are discussed below, focusing on six fields of action observed
in developed markets. Regarding Personalization and Customer Experience
Improvement, the use of Machine Learning (ML) by financial institutions focuses
on several key areas. First, financial recommendations are used to analyze
customer behavior and transactions, offering personalized products or services
such as loans, investments or credit cards. The abundance of data available on
customer behavior and transactions provides the opportunity to offer tailored
experiences for each customer.
In
addition, fast and efficient analysis is carried out on large amounts of data
generated by customers, including transactional data such as purchase
histories, purchase amounts, locations and frequency of spending, as well as
behavioral data such as frequency of access to banking applications,
interaction with online services, responses to previous offers or communication
preferences. This allows for offering personalized products and services, such
as loans that identify interested or needy customers, investment opportunities
tailored to the customer's risk profile and financial goals, and credit cards
with specific benefits that align with the customer's spending habits.
These
practices not only benefit financial institutions by increasing customer
satisfaction and retention, but also improve marketing efficiency by
specifically targeting customer needs and preferences, and improve
decision-making by better understanding what the customer needs in terms of
product development and market strategies.
Optimizing
operations and efficiency through process automation is a key focus in the
field of AI (Artificial Intelligence). This automation refers to the use of
advanced systems to carry out tasks that previously required human
intervention, especially those that are repetitive, predictable and
high-volume. The main objective is to increase efficiency, reduce errors and
free up employees to focus on higher value-added activities, mainly of a
commercial nature.
In the
realm of financial operations, AI has proven to be particularly effective in
areas such as account reconciliation and transaction categorization. For
example, in the account reconciliation process, which involves reconciling an
organization’s internal and external records, AI can quickly process large data
sets and reconcile accounts in real-time or at specific intervals, identifying
discrepancies and alerting the relevant teams. Furthermore, AI systems can
learn from past transactions and, using ML (Machine Learning) techniques,
automatically categorize incoming transactions into appropriate categories,
such as “travel expenses,” “office supplies,” or “salaries.”
Another
area where AI is transforming operations is in investment management, through
so-called robo-advisors, digital platforms that offer automated investment
management services. These systems use algorithms and ML techniques to advise
users on where to invest their money based on their risk profile, financial
goals, and other personal factors (Belanche, Casaló & Flavián, 2019). Robo-advisors
have democratized access to investment management, offering affordable
solutions to a wide range of clients (Shanmuganathan, 2020), and allowing
financial services firms to operate more efficiently and scale their operations
without needing to proportionally increase their staff.
Before
making any investment recommendations, robo-advisors typically ask users to
complete a questionnaire to determine their risk tolerance, expected time
horizon for investment, financial goals, and also whether all of this is in
line with their current financial situation. Using this information, the
algorithm defines an investment profile for the client and automatically
selects a diversified investment portfolio based on this profile, which can
include stocks, bonds, funds, and other financial instruments.
Additionally,
some robot-advisors offer tax optimization strategies, where loss-making
investments are intentionally sold to offset gains, thereby reducing the
investor’s tax bill. In addition to offering investment management services,
many robo-advisors provide educational resources, such as articles, tutorials,
and simulators, to help clients better understand the world of investing.
The
term Big Data is derived from the enormous volumes of data that, due to their
size, velocity or variety, cannot be efficiently processed with traditional
applications. This data, when properly analyzed, offers valuable insights that
can be used to make strategic decisions across multiple industries, including
the financial industry.
Trend
analysis involves observing and analyzing historical and current data to
identify patterns, changes and possible future directions in various areas,
such as markets, consumer behaviors and macroeconomic factors. In the financial
sector, some specific applications include identifying market opportunities,
managing risks by anticipating possible crises or recessions, evaluating
products and services to tailor offerings, and anticipating market movements such
as interest rates, stock prices, commodities and other financial instruments.
Customer
segmentation involves dividing a company's customer base into smaller,
homogeneous groups based on similar characteristics, such as age, income,
purchasing behaviors or financial needs. This allows for offering personalized
and effective products and services, as well as more precise and relevant
marketing communications. In the financial sector, this translates into product
customization, effective marketing and advertising strategies, improved
customer service, and segmentation-based price optimization.
The
financial industry is constantly evolving, regularly facing new needs and
challenges for individual customers, businesses, and financial institutions
themselves. To stay competitive and relevant, these institutions must
continually innovate their product and service offerings. Big data analytics
provides valuable insights that guide and accelerate the innovation process.
New
product development involves researching, creating, and launching new offerings
to meet identified needs in the market. Some of the areas where this process is
applied in financial market research include identifying underlying needs
through the analysis of large volumes of customer data, pilot testing to
fine-tune and refine products, personalizing offerings based on big data
insights, demand forecasting, and continuous improvement based on monitoring
and tracking product performance in real time.
Financial
institutions use big data to innovate their products and services. Examples
include flexible savings accounts based on customer transaction analysis, and
thematic investment platforms that respond to trends and needs observed in
investor activity and interests.
The
financial sector globally is heavily regulated to maintain economic stability,
safeguard consumers, and deter illicit activities. These regulations are
intricate and subject to frequent changes, posing compliance challenges for
financial entities. The emergence of advanced technologies like Artificial
Intelligence (AI) and Big Data has spurred innovation in Regulatory Technology
(RegTech), facilitating more efficient and cost-effective compliance solutions.
RegTech, a burgeoning field within fintech, utilizes technology to streamline regulatory
compliance processes, as outlined by Anagnostopoulos (2018); and Bayramoğlu
(2021). This includes real-time monitoring of operations, leveraging AI to
detect anomalies or violations and prompt necessary actions.
Furthermore,
RegTech solutions leverage Big Data analytics to process vast transactional
data, identify patterns, and facilitate automated reporting in adherence to
regulations. These technologies also aid in risk identification, employing AI
and predictive analytics to assess and address potential risks like money
laundering or fraud. RegTech systems also support staff training by utilizing
AI for ongoing education and content updating to align with evolving
regulations.
Examining
specific cases, RegTech tools are instrumental in Anti-Money Laundering (AML)
efforts by tracking suspicious transactions based on patterns or unusual
behavior. For Know Your Customer (KYC) processes, AI automates customer
verification, document validation, and security checks, minimizing biases.
Additionally, with data privacy regulations such as GDPR, RegTech tools ensure
institutions comply with laws regarding the handling and storage of customers'
personal data, enhancing overall data privacy compliance efforts.
Conclusions
The
combination of traditional and data-driven artificial intelligence techniques
was identified as offering significant competitive advantages for financial
institutions. This translates into greater operational efficiency, improved
decision-making, and a greater ability to anticipate and respond to customer
needs.
A
comprehensive strategy is suggested that includes staff training in new
technologies, the establishment of strategic collaborations with specialized
technology companies, and the development of flexible regulatory frameworks that
encourage innovation without compromising data security and privacy.
On the
other hand, the need to further explore the impact of hybrid AI on financial
risk management, the customization of banking products and services, as well as
its potential to boost financial inclusion and reduce the digital divide in
certain populations is highlighted. These areas represent promising fields for
the development of new research that contributes to expanding knowledge and
optimizing the application of artificial intelligence in the banking sector.
Bibliographic references
Alonso, F., Martínez, L., Pérez, A., &
Valente, J. P. (2012). Cooperación entre el conocimiento experto y el
conocimiento descubierto por minería de datos: lecciones aprendidas. Expert
Systems with Applications, 39(8),
7524-7535. https://doi.org/10.1016/j.eswa.2012.01.133
Anagnostopoulos, I. (2018). Fintech
and regtech: Impact on regulators and banks. Journal of Economics and Business, 100, 7-25. https://doi.org/10.1016/j.jeconbus.2018.07.003
Angelini, E., Di Tollo,
G., & Roli, A. (2008). A neural network
approach for credit risk evaluation. The Quarterly
Review of Economics and Finance, 48(4), 733-755. https://doi.org/10.1016/j.qref.2007.04.001
Arbeláez-Campillo, D. F.,
Villasmil, J. J., & Rojas-Bahamón, M. J. (2021). Inteligencia artificial y
condición humana: ¿Entidades contrapuestas o fuerzas complementarias? Revista
de Ciencias Sociales (Ve), XXVII(2), 502-513. https://doi.org/10.31876/rcs.v27i2.35937
Aslam, F., Hunjra, A.
I., Ftiti, Z., Louhichi, W., & Shams, T. (2022). Insurance fraud detection: Evidence from artificial
intelligence and machine learning. Research
in International Business and Finance, 62, 101744. https://doi.org/10.1016/j.ribaf.2022.101744
Bayramoğlu, G. (2021). An overview
of the artificial intelligence applications in fintech and Regtech. In S. Bozkuş (Eds), The Impact of Artificial Intelligence on Governance, Economics and
Finance (Vol. I, pp. 291-298). Springer. https://doi.org/10.1007/978-981-33-6811-8_15
Belanche, D., Casaló, L. V., &
Flavián, C. (2019). Artificial Intelligence in FinTech: understanding
robo-advisors adoption among customers. Industrial Management & Data
Systems, 119(7), 1411-1430. https://doi.org/10.1108/imds-08-2018-0368
Bellazzi, R. & Zupan, B.
(2008). Predictive data mining in clinical medicine: current issues and
guidelines. International Journal of
Medical Informatics, 77(2), 81-97. https://doi.org/10.1016/j.ijmedinf.2006.11.006
Berdnikov, V. (2021). Generation of
prognostic interval estimates of water inflows to hydroelectric reservoirs
using multiparametric neural network. E3S Web of Conferences, 289, 01003. https://doi.org/10.1051/e3sconf/202128901003
Bialecki, A., Cafarella, M.,
Cutting, D., & O´Malley, O. (2005). Hadoop: A framework for running
applications on large clusters built of commodity hardware. http://lucene.apache.org/hadoop
Braglia, M., Frosolini,
M., & Montanari, R. (2003). Fuzzy
criticality assessment model for failure modes and effects analysis. International Journal of Quality &
Reliability Management, 20(4), 503-524. https://doi.org/10.1108/02656710310468687
Bramer, M. (2016). Principles of data mining. Springer. https://doi.org/10.1007/978-1-4471-7307-6
Bryndin, E. (2019). Collaboration of intelligent interoperable
agents via smart interface. International Journal on Data Science and
Technology, 5(4), 66-72. https://doi.org/10.11648/j.ijdst.20190504.11
Çavuşgil, S. T., & Evirgen, C.
(1997). Use of expert systems in international marketing. European Journal of Marketing, 31(1), 73-86. https://doi.org/10.1108/03090569710157043
Coman, A., & Aha, D. W. (2018). AI rebel agents. AI Magazine, 39(3), 16-26.
https://doi.org/10.1609/aimag.v39i3.2762
De Vreede, T.,
Raghavan, M., & De Vreede, G.-J. (2021). Design
foundations for AI assisted decision making: A self determination theory
approach. Proceedings of the Annual 54th
Hawaii International Conference on System Sciences. https://doi.org/10.24251/hicss.2021.019
Ding, S., Su, C., &
Yu, J. (2011). An
optimizing BP neural network algorithm based on genetic algorithm. Artificial
Intelligence Review, 36(2), 153-162. https://doi.org/10.1007/s10462-011-9208-z
Doğuç, Ö. (2022). Data mining applications in banking sector
while preserving customer privacy. Emerging Science Journal, 6(6), 1444-1454. https://doi.org/10.28991/esj-2022-06-06-014
Fernández, A. (2019). Inteligencia
artificial en los servicios financieros. Boletín
Económico / Banco de España, 2, 1-10. https://repositorio.bde.es/handle/123456789/8448
Gen, M., y Cheng, R.
(1999). Genetic algorithms
and engineering optimization. https://doi.org/10.1002/9780470172261
Gilman, G., & Walls, R. J. (2021).
Characterizing concurrency mechanisms for NVIDIA GPUs under deep learning
workloads. Performance Evaluation, 151,
102234. https://doi.org/10.1016/j.peva.2021.102234
Haenlein, M., & Kaplan, A.
(2019). A Brief History of
Artificial Intelligence: On the Past, Present, and Future of Artificial
Intelligence. California
Management Review, 61(4), 5-14. https://doi.org/10.1177/0008125619864925
Han, J., Cheng, H., Dong, X., &
Yan, X. (2007). Frequent pattern mining: Current status and future directions. Data Mining and Knowledge Discovery, 15(1),
55-86. https://doi.org/10.1007/s10618-006-0059-1
Ivančan, J., & Lisjak, D.
(2021). New fmea risks ranking approach utilizing four fuzzy logic systems. Machines, 9(11), 292. https://doi.org/10.3390/machines9110292
Julianto, I. T., Kurniadi, D.,
Septiana, Y., & Sutedi, A. (2023). Alternative Text Pre-Processing using
Chat GPT Open AI. https://app.readwonders.com/article/18519-alternative-text-pre-processing-using-chat-gpt-open-ai
Khosla, R., & Dillon, T.
(2012). Engineering intelligent hybrid
multi-agent systems. Springer Science & Business Media.
Klement, E. P., & Slany, W. (Eds.)
(1993, June). Fuzzy logic in artificial intelligence. In Proceedings of the 8th Austrian Artificial Intelligence Conference,
FLAI (Vol. 93). https://doi.org/10.1007/3-540-56920-0
Koeppe, A., Bamer, F.,
Selzer, M., Nestler, B., & Markert, B. (2022). Explainable artificial intelligence for mechanics:
physics-explaining neural networks for constitutive models. Frontiers in Materials, 8, 824958. https://doi.org/10.3389/fmats.2021.824958
Koshka, Y., & Novotny, M. A.
(2020). Comparison of d-wave quantum annealing and classical simulated
annealing for local minima determination. IEEE
Journal on Selected Areas in Information Theory, 1(2), 515-525. https://doi.org/10.1109/jsait.2020.3014192
Lanzing, M. (2023).
Traveling technology and perverted logics: Conceptualizing Palantir’s expansion
into health as sphere transgression. Information,
Communication & Society, 1-17. https://doi.org/10.1080/1369118X.2023.2279557
Le, T. L., Negnevitsky, M., &
Piekutowski, M. (1997). Network equivalents and expert system application for
voltage and VAR control in large-scale power systems. IEEE Transactions on Power Systems, 12(4), 1440-1445. https://doi.org/10.1109/59.627839
LeCun, Y., Bengio, Y., &
Hinton, G. (2015). Deep learning. Nature,
521(7553), 436-444. https://doi.org/10.1038/nature14539
Luo, X., Qin, M. S., Fang,
Z., & Qu, Z. (2021). Artificial Intelligence coaches for sales agents: Caveats and solutions. Journal of
Marketing, 85(2), 14-32. https://doi.org/10.1177/0022242920956676
Maita-Cruz, Y. M.,
Flores-Sotelo, W. S., Maita-Cruz, Y. A., y Cotrina-Aliaga, J. C. (2022). Inteligencia artificial en la gestión pública en tiempos de
Covid-19. Revista de Ciencias Sociales (Ve), XXVIII(E-5), 331-330. https://doi.org/10.31876/rcs.v28i.38167
Mansingh, G., Reichgelt,
H., & Osei, K.-M. (2007). CPEST: An expert
system for the management of pests and diseases in the Jamaican coffee
industry. Expert Systems with
Applications, 32(1), 184-192. https://doi.org/10.1016/j.eswa.2005.11.025
Medsker, L. R. (2012). Hybrid intelligent systems. Springer
Science & Business Media. https://doi.org/10.1007/978-1-4615-2353-6
Mendel, J. M. (1995). Fuzzy logic
systems for engineering: A tutorial. Proceedings
of the IEEE, 83(3), 345-377. https://doi.org/10.1109/5.364485
Mhlanga, D. (2021). Financial
inclusion in emerging economies: The application of machine learning and
artificial intelligence in credit risk assessment. International Journal of Financial Studies, 9(3), 39. https://doi.org/10.3390/ijfs9030039
Moposita, D. A., &
Jordán, J. E. (2022). Chatbot
una herramienta de atención al cliente en tiempos de COVID-19: Un acercamiento
teórico. Uniandes
Episteme, 9(3), 327-350. https://revista.uniandes.edu.ec/ojs/index.php/EPISTEME/article/view/2481
Nabi, W., Bansal, A., & Xu, B.
(2021). Applications of artificial intelligence and machine learning approaches
in echocardiography. Echocardiography, 38(6),
982-992. https://doi.org/10.1111/echo.15048
Powles, J., & Hodson,
H. (2017). Google DeepMind and healthcare in an age of algorithms. Health and Technology. 7, 351-367. https://doi.org/10.1007/s12553-017-0179-1
Prieto, A., Prieto, B., Martinez,
E., Ros, E., Pelayo, F., Ortega, J., & Rojas, I. (2016). Neural networks: An overview of early research,
current frameworks and new challenges. Neurocomputing,
214, 242-268. https://doi.org/10.1016/j.neucom.2016.06.014
Rahmani, A. M., Yousefpoor, E.,
Yousefpoor, M. S., Mehmood, Z., Haider, A., Hosseinzadeh, M., & Naqvi, R. A.
(2021). Machine learning (ML) in medicine: Review, applications, and
challenges. Mathematics, 9(22), 2970.
https://doi.org/10.3390/math9222970
Rodríguez, Y. (2023). Análisis del uso de la Inteligencia
Artificial en las diversas operaciones financieras [Tesis de pregrado, Unidades
Tecnológicas de Santander]. http://repositorio.uts.edu.co:8080/xmlui/handle/123456789/14465
Ryman-Tubb, N. F., Krause, P.,
& Garn, W. (2018). How Artificial Intelligence and machine learning
research impacts payment card fraud detection: A survey and industry benchmark.
Engineering Applications of Artificial
Intelligence, 76, 130-157. https://doi.org/10.1016/j.engappai.2018.07.008
Shaheen, M., Shahbaz, M., Ur Rehman,
Z., & Guergachi, A. (2010). Aplicaciones de la minería de
datos en la exploración de hidrocarburos. Artificial Intelligence Review, 35(1), 1-18. https://doi.org/10.1007/s10462-010-9180-z
Shanmuganathan, M. (2020).
Behavioural finance in an era of artificial intelligence: Longitudinal case
study of robo-advisors in investment decisions. Journal of Behavioral and Experimental Finance, 27, 100297. https://doi.org/10.1016/j.jbef.2020.100297
Sivanandam, S. N., &
Deepa, S. N. (2008). Introduction to Genetic Algorithms. Springer Berlin Heidelberg.
Susskind, R. (1988). Expert systems in law: A jurisprudential
inquiry. Oxford University Press.
Tamir, D. E., Rishe, N. D., &
Kandel, A. (2015). Complex fuzzy sets and complex fuzzy logic an overview of
theory and applications. In D.
Tamir, N. Rishe & A. Kandel (Eds.), Fifty years of
fuzzy logic and its applications (Vol.
326, pp. 661-681). Springer. https://doi.org/10.1007/978-3-319-19683-1_31
Thongsatapornwatana, U. (2016). A
survey of data mining techniques for analyzing crime patterns. Second Asian Conference
on Defence Technology (ACDT), Chiang
Mai, Thailand, 123-128. https://doi.org/10.1109/acdt.2016.7437655
Tiutiunnyk, P. B., & Rybachok,
N. A. (2021). Creating web application for organizing teamwork online using Microsoft
azure cloud services. Control Systems and
Computers, (2-3), 52-59. https://doi.org/10.15407/csc.2021.02.052
Tyagi, A. K., & Chahal, P.
(2022). Artificial intelligence and machine learning algorithms. In Information Resources Management Association
(Ed.), Research anthology on machine
learning techniques, methods, and applications (pp. 421-446). IGI Global. https://doi.org/10.4018/978-1-6684-6291-1.ch024
Ventura-Fernández, T., Vidalón-Soldevilla,
E., & Ventura-Fernández, F. (2021). Predictibilidad en el diagnóstico
utilizando Watson de IBM. Vive. Revista de Investigación en
Salud, 4(10),
86-96. https://doi.org/10.33996/revistavive.v4i10.78
Wang, M., Cui, Y., Wang,
X., Xiao, S., & Jiang, J. (2018). Aprendizaje automático
para redes: flujo de trabajo, avances y oportunidades. IEEE Network, 32(2), 92-99. https://doi.org/10.1109/mnet.2017.1700200
Wang, Y., Zheng, P., Peng, T.,
Yang, H., & Zou, J. (2020). Fabricación aditiva
inteligente: métodos actuales habilitados por inteligencia artificial y
perspectivas futuras. Science China
Technological Sciences, 63(9), 1600-1611. https://doi.org/10.1007/s11431-020-1581-2
Zadeh, L. A. (1996). Fuzzy logic =
computing with words. IEEE Transactions
on Fuzzy Systems, 4(2), 103-111. https://doi.org/10.1109/91.493904
Zaharia, M., Xin, R. S.,
Wendell, P., Das, T., Armbrust, M., Dave, A., Meng, X., Rosen, J.,
Venkataraman, S., Franklin, M. J., Ghodsi, A., Gonzalez, J. E., Shenker, S. J.,
& Stoica, I. (2016). Apache Spark: A unified engine
for big data processing. Comunicaciones de la ACM, 59(11), 56-65. https://doi.org/10.1145/2934664
Zhang, L., Liu, K., Ilham,
I., & Fan, J. (2022). Application of
data mining technology based on data center. Journal of Physics Conference Series, 2146, 012017. https://doi.org/10.1088/1742-6596/2146/1/012017
Zhao, Y., Guo, Z. H., & Yan, J.
M. (2017). Vibration signal
analysis and fault diagnosis of bogies of the high-speed train based on deep
neural networks. Journal of
Vibroengineering, 19(4), 2456-2474. https://doi.org/10.21595/jve.2017.17238
Zhou, Z., Chen, X., Li, E., Zeng,
L., Luo, K., & Zhang, J. (2019). Edge intelligence: Paving the last mile of artificial intelligence with edge computing.
Proceedings of the IEEE, 107(8),
1738-1762. https://doi.org/10.1109/jproc.2019.2918951
Zhu, J. (2022). Solving capacitated
vehicle routing problem by an improved genetic algorithm with fuzzy c-means
clustering. Scientific Programming,
1-8. https://doi.org/10.1155/2022/8514660