Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501

Leveraging Artificial Intelligence for Business

Analytics: A Data-Science based Decision
Support System Framework
Dileesh Chandra Bikkasani 1
1University of Bridgeport

Publication Date: 2025/03/06

Abstract: Artificial Intelligence (AI) transforms business intelligence (BI) by enhancing decision-making speed, accuracy,
and depth in today’s data-driven landscape. Traditional Decision Support Systems (DSS), once foundational to BI, struggle
to handle modern data's complexity, scale, and diversity, often resulting in limited decision-making agility. Integrating AI
into DSS has become essential to bridge this gap, enabling these systems to process vast datasets in real-time and make
predictive, data-informed recommendations. This study presents an AI-powered DSS framework designed to address the
limitations of conventional DSS by incorporating machine learning, natural language processing, and adaptive feedback
mechanisms. Through real-world simulations and industry-specific use cases, the framework demonstrates marked
improvements in decision quality, response times, and user satisfaction compared to traditional systems. Findings suggest
that AI-driven DSS can substantially enhance BI processes, equipping organizations with a proactive, scalable approach to
decision support. By addressing key technical and ethical challenges, this research offers valuable insights for businesses
aiming to leverage AI to stay competitive in increasingly complex environments, positioning AI-powered DSS as critical to
the future of BI.

Keywords: Artificial Intelligence, Business Intelligence, Decision Support Systems, Predictive Analytics, Framework
Development.

How to Cite: Dileesh Chandra Bikkasani. (2025). Leveraging Artificial Intelligence for Business Analytics: A Data-Science based
Decision Support System Framework Case Study in Lusaka. International Journal of Innovative Science and Research
Technology, 10(2), 1507-1515. https://doi.org/10.5281/zenodo.14964501

I. INTRODUCTION sophisticated analytical tools within BI, specifically tools that

are adaptable, predictive, and capable of handling the nuances
In today’s data-driven world [1], the capacity to make of big data [7]. Traditional DSS frameworks, designed
informed, timely decisions is a crucial differentiator for primarily for structured and historical data, often cannot
businesses across sectors. Business Intelligence (BI) and handle these complexities, limiting their effectiveness. For
Decision Support Systems (DSS) [2] have long been instance, they may struggle with integrating unstructured data
foundational tools for turning raw data into actionable (e.g., text, images, or social media content), fail to adapt to
insights, guiding executives, managers, and decision-makers dynamic data environments or provide real-time
toward choices that drive growth and efficiency. recommendations for decision-making in fast-paced
Traditionally, DSS relied on structured data and rule-based industries [8].
models [3], providing structured recommendations within
well-defined business processes [4]. However, as the digital Enter Artificial Intelligence (AI), a transformative
economy expands, businesses are now confronted with an technology with the potential to reimagine DSS by
influx of data at unimaginable scales. This shift has introducing a level of adaptability, automation, and predictive
underscored a growing gap between the capabilities of power previously unattainable. With machine learning (ML),
traditional DSS and the needs of modern, data-intensive natural language processing (NLP), and computer vision, AI
business environments [5]. enables systems to learn from data patterns, automate
repetitive analytical tasks, and generate insights that account
The business data landscape is characterized by the “3 for complex interrelationships in data [9]. AI-powered DSS
Vs”: volume, velocity, and variety [6]. Organizations are can interpret vast datasets more effectively, support faster
generating more data at higher speeds and from increasingly decision-making, and improve the quality of
diverse sources, including social media, sensors, and recommendations by drawing insights from structured and
customer interaction logs. In response to this growing data unstructured data sources [10]. For instance, NLP can analyze
complexity, there has been a push to integrate more text data, providing sentiment analysis on customer feedback,

IJISRT25FEB1475 www.ijisrt.com 1507

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
while ML algorithms can forecast sales trends or detect systems represents a shift towards reactive but also
potential operational risks based on historical patterns [11]. predictive, and proactive decision-making frameworks (see
This evolution from rule-based DSS to AI-enhanced learning Fig.1).

Fig 1 AI-Based DSS

However, despite these benefits, integrating AI into advanced algorithms, real-time processing, and pattern
DSS presents a range of challenges [12]. From a technical recognition.
standpoint, successfully implementing AI in decision support What are the critical components of an AI-powered DSS?
requires significant infrastructure upgrades, such as high- This question seeks to define a framework outlining the
performance computing resources and scalable data storage necessary components of an AI-enhanced DSS, including
solutions capable of handling large datasets [13]. data ingestion, processing, predictive analytics, and user
Operationally, businesses must grapple with data quality, data interfaces.
privacy, and the requirements for skilled personnel to manage
and interpret complex AI models. Ethically, using AI in What challenges and limitations must be addressed for
decision-making raises concerns about transparency, effective AI integration in DSS? This question examines the
accountability, and bias [14]. AI models are often seen as barriers to AI adoption, including technical, operational, and
“black boxes,” meaning their decision processes are not ethical challenges, and considers how organizations can
easily interpretable, leading to mistrust among users, navigate these obstacles.
especially when decisions significantly impact stakeholders.
Moreover, AI’s reliance on historical data can sometimes By addressing these questions, this paper contributes to
enhance existing biases, underscoring the need for careful a comprehensive understanding of how AI can transform
model governance and regulatory compliance [15]. traditional DSS into advanced decision support frameworks,
aligning them with the increasingly complex demands of
In this context, the objective of this paper is to explore modern business environments. Ultimately, this research
how AI can enhance DSS functionality to improve the speed, offers a pathway for organizations to leverage AI as a
accuracy, and adaptability of decision-making in business strategic asset in BI, enabling them to make smarter, faster,
analytics. By addressing the limitations of traditional DSS, and more reliable decisions in an era of rapid technological
this paper proposes a conceptual framework that defines the change.
essential components of an AI-powered DSS and illustrates
how such a system can augment business intelligence II. TRADITIONAL DECISION SUPPORT
processes. The framework will emphasize how AI-driven SYSTEMS (DSS)
analytics can empower organizations to navigate the
complexities of big data better, anticipate future scenarios, Decision Support Systems (DSS) [17] have evolved
and ultimately make better, data-informed decisions [16]. significantly since their inception in the 1960s, driven by the
Through this study, the paper aims to provide insights into the need for better decision-making tools in business contexts.
structural and technical requirements for effective AI Initially developed to assist managers in making informed
integration and the managerial considerations needed to decisions based on quantitative data, traditional DSS have
balance AI’s capabilities with its risks. since expanded their scope to accommodate more complex
analyses. The conventional structure of DSS typically
 The Central Research Questions Guiding this Study comprises three core components: data management, model
Include: management, and user interface [18]. The data management
How can AI improve decision-making accuracy, speed, layer is responsible for data storage, retrieval, and processing,
and quality in business analytics? This question explores AI’s ensuring that relevant information is available for analysis
ability to enhance the core capabilities of DSS by leveraging [19]. Model management involves the application of

IJISRT25FEB1475 www.ijisrt.com 1508

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
mathematical and statistical models to interpret data and the successful application of AI in business intelligence (BI)
generate insights. The user interface facilitates interaction have emerged, showcasing advancements that lead to
between decision-makers and the system, enabling users to improved operational efficiency, enhanced customer
visualize data and model outputs effectively [20]. Despite experiences, and data-driven strategic planning.
their effectiveness, traditional DSS often struggle to handle
the ever increasing size and complexity of modern data [21].  AI-Powered DSS
AI-powered Decision Support Systems builds on the
 Artificial Intelligence in Business Analytics theoretical foundations of traditional DSS by incorporating
In recent years, the integration of Artificial Intelligence AI techniques to enhance decision-making capabilities. The
(AI) into business analytics has transformed how key differences between traditional and AI-driven systems lie
organizations leverage data for decision-making. Various AI in their ability to process large volumes of unstructured data,
techniques, including machine learning, deep learning, and adapt to changing circumstances, and provide real-time
natural language processing (NLP), have been adopted to insights. While traditional DSS typically rely on historical
enhance analytical capabilities. For instance, machine data and predetermined models, AI-driven systems leverage
learning algorithms enable systems to identify patterns in machine learning and other AI techniques to learn from data
large datasets and make predictions based on historical data, patterns and generate recommendations dynamically [12].
while deep learning techniques facilitate the analysis of Various AI-driven DSS models and frameworks have been
unstructured data such as images and text. NLP allows proposed in the literature, illustrating how they have evolved
businesses to extract valuable insights from textual data, to use specific decision-making frameworks using augmented
providing a more comprehensive view of customer analytics, as described in Fig 2.
sentiments and market trends [22]. Case studies highlighting

Fig 2 Evolution of Analytics

 Challenges in Implementing AI in DSS AI-driven DSS, necessitating careful planning and

Despite the potential benefits of integrating AI into investment to overcome these barriers.
DSS, several challenges hinder successful implementation.
Data quality issues, such as inaccuracies and inconsistencies, In conclusion, the literature underscores the
can undermine the effectiveness of AI algorithms. transformative potential of AI in enhancing traditional DSS
Additionally, scalability poses a significant challenge, as and business analytics. While there are significant
organizations must ensure that their infra can handle the advancements and success stories associated with AI-driven
increasing volume of data generated in real-time. Integration DSS, challenges in data quality, scalability, and ethical
issues also arise when integrating AI technologies with considerations must be addressed to fully get the benefits of
existing systems and data sources. Ethical concerns, these systems in modern decision-making environments.
including biases in AI algorithms and the need for
transparency in decision-making processes, have been raised,  Conceptual Framework for AI-Powered Decision Support
emphasizing the importance of responsible AI usage [14]. Systems (DSS)
Furthermore, resource and infrastructure limitations can As AI technology reshapes traditional decision-making
restrict organizations' ability to adopt and maintain advanced tools, understanding the core components of an AI-powered

IJISRT25FEB1475 www.ijisrt.com 1509

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
DSS is essential for leveraging its full potential in business practices) and tacit knowledge (e.g., insights from expert
analytics. This section outlines the conceptual framework for users), which the AI models can use to make more
an AI-enhanced DSS, focusing on the major components, informed recommendations. By incorporating industry-
their functions, and the overall workflow. By identifying specific insights, the DSS can generate more accurate and
these elements, the framework provides a blueprint for contextually relevant outputs, catering to the unique
developing robust, adaptive systems capable of supporting demands of different sectors, such as healthcare, finance,
complex, data-driven decisions. or retail.
 Feedback Mechanism: A feedback mechanism is essential
 Core Components of AI-Powered DSS for continuous learning and adaptation, especially in
dynamic business environments where conditions change
 Data Layer: The data layer [23] forms the foundation of frequently. Reinforcement learning, a type of machine
an AI-powered DSS by integrating various data types and learning that learns from feedback, enables the DSS to
managing the influx of big data. Unlike traditional adjust its models based on new data or user inputs. This
systems that rely on structured data, this layer mechanism ensures that the system remains responsive to
accommodates structured, semi-structured, and shifts in business needs and data trends, effectively self-
unstructured data, such as customer feedback, social optimizing over time to deliver increasingly relevant
media posts, and multimedia. Data preprocessing, insights.
including cleansing, normalization, and feature extraction,
is essential here to prepare the data for analysis and to  Workflow of AI-Powered DSS
ensure it meets quality standards [24]. Furthermore, data The workflow of an AI-powered DSS follows a
integration capabilities allow the system to bring in structured process to ensure data flows seamlessly through
external sources for enriched analysis, enhancing the each component, ultimately enabling informed decision-
relevance and depth of insights. making.
 AI Models: At the heart of AI-powered DSS lies a set of
advanced AI models, such as machine learning, deep  Data Collection: The process begins with data collection
learning, and predictive models, each playing a unique from diverse sources, including databases, real-time
role in decision support [25]. Machine learning algorithms sensors, customer logs, and social media platforms,
identify patterns in historical data, while deep learning among others. Effective data collection is crucial for
models are adept at processing unstructured data, such as capturing the 3Vs of big data (volume, velocity, variety)
images and text. Predictive models, on the other hand, [26] necessary for comprehensive analysis.
leverage past trends to forecast outcomes, offering  Data Processing: In this step, raw data undergoes
decision-makers forward-looking insights. These models preprocessing and transformation to ensure it is suitable
can be tailored to specific decision-making needs, such as for analysis. Techniques such as data cleaning,
customer segmentation, risk assessment, or predictive normalization, and feature extraction help in reducing
maintenance, ensuring that the DSS delivers insights that noise and structuring data, making it usable by AI models.
are actionable and relevant.  Model Training: The AI models are then trained on the
 User Interface (UI): The user interface enables interaction processed data to identify patterns, detect anomalies, or
between the end-user and the AI-powered DSS, ensuring make predictions. Depending on the specific use case,
that complex insights are easily accessible. A well- training may involve supervised, unsupervised, or
designed UI should support data visualization, allowing reinforcement learning methods.
users to understand data patterns intuitively through  Decision-Making: Once trained, the AI models generate
charts, dashboards, and graphs. Moreover, natural insights or recommendations based on real-time data or
language querying capabilities, powered by natural simulated scenarios. This output is delivered to users
language processing (NLP), enhance usability by enabling through the UI, where visualizations and analytics
users to ask questions or request specific analyses in plain summaries assist in interpreting results and guiding
language. This democratizes access to insights, decisions.
empowering individuals across various roles within the  Feedback Loop: The workflow concludes with a feedback
organization to leverage the system’s capabilities. loop that allows the DSS to improve over time. Feedback
 Knowledge Base: Integrating domain-specific knowledge from user interactions and outcome evaluations informs
is crucial to improve the accuracy and relevance of the model adjustments, making the DSS increasingly accurate
DSS. A knowledge base encompasses both explicit and adaptable.
knowledge (e.g., guidelines, regulations, and best

IJISRT25FEB1475 www.ijisrt.com 1510

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501

Fig 3 Life cycle of AI-powered DSS

 Integration of AI Models in Decision Support Systems  Unsupervised Learning Models

The integration of AI models into Decision Support In contrast, unsupervised learning involves training
Systems (DSS) is a critical aspect that can significantly models on unlabeled data, enabling them to identify patterns
enhance their capabilities in business intelligence (BI). As and groupings without explicit guidance. Clustering
organizations increasingly confront complex decision- algorithms, such as K-means and hierarchical clustering, are
making scenarios driven by vast amounts of data, traditional commonly used in DSS to segment data into meaningful
DSS are often insufficient. They primarily rely on rule-based categories. For instance, a marketing team might use
logic and historical data, which limit their ability to adapt to unsupervised learning to analyze customer data, identifying
real-time changes and handle the dynamic nature of modern distinct segments based on purchasing behavior,
business environments [27]. This section explores the various demographics, and engagement levels [29]. This
AI models that can be employed in DSS, including supervised segmentation can inform targeted marketing strategies,
and unsupervised learning, and highlights their implications allowing businesses to tailor their offerings and
for improving decision-making processes. communications to specific customer needs.

 Supervised Learning Models Unsupervised learning can also aid in anomaly

Supervised learning is a type of machine learning where detection [30], vital for identifying unusual patterns that may
models are trained on labeled datasets [28], enabling them to indicate fraud or operational inefficiencies. Organizations can
learn patterns and make predictions based on new, unseen proactively monitor transactions or processes by integrating
data. In the context of DSS, supervised learning algorithms, anomaly detection models into their DSS, enhancing risk
such as regression analysis, decision trees, and support vector management and operational resilience.
machines, can be employed to generate predictive insights
that inform decision-makers. For instance, a retail  Natural Language Processing (NLP)
organization may use supervised learning to forecast sales Another essential AI technique that can be integrated
based on historical data, seasonal trends, and consumer into DSS is Natural Language Processing (NLP). NLP
behavior patterns. By integrating these predictive models into enables systems to understand and interpret human language,
their DSS, businesses can enhance their planning and allowing organizations to derive insights from unstructured
resource allocation, ultimately driving efficiency and growth data sources such as customer feedback, social media posts,
[18]. and internal reports [31]. By incorporating NLP into DSS,
businesses can automate the analysis of textual data,
Additionally, supervised learning can support providing sentiment analysis, trend identification, and topic
prescriptive analytics, which goes a step further by modeling.
recommending specific actions based on predicted outcomes.
For example, a financial institution could utilize supervised For example, a company could use NLP to analyze
learning to assess loan applications, evaluating factors such customer reviews and social media mentions, identifying key
as credit scores, income levels, and debt-to-income ratios to themes and sentiments reflecting customer satisfaction and
determine the likelihood of default. The DSS can then suggest improvement areas. This information can be integrated into
optimal lending strategies, enabling the institution to balance the DSS, enabling decision-makers to respond swiftly to
risk and profitability effectively. customer needs and enhance product or service offerings.

IJISRT25FEB1475 www.ijisrt.com 1511

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
 Feedback Mechanisms and Continuous Learning learning and deep learning. Additionally, Apache Airflow
Establishing effective feedback mechanisms is critical was integrated for task scheduling and workflow automation,
to integrating AI models into DSS. These mechanisms allow allowing seamless data ingestion and preprocessing
the system to learn from past decisions and improve its workflows.
predictive capabilities continuously. Reinforcement learning,
a subset of machine learning, can be particularly beneficial in A variety of AI models were employed to maximize the
this context [32]. It enables the DSS to learn optimal decision- DSS’s decision-making capabilities. Supervised learning
making strategies through trial and error, adjusting its models, such as regression and classification, were applied to
approach based on the outcomes of previous decisions. predict outcomes based on historical data, useful for scenarios
like customer churn prediction in retail or credit risk
For instance, an AI-powered DSS for supply chain assessment in finance. Unsupervised learning models,
management could use reinforcement learning to optimize including clustering algorithms, were used for segmenting
inventory levels. By evaluating the results of past inventory large datasets without labeled outcomes, which is particularly
decisions, the system can refine its strategies to minimize effective for identifying customer segments or detecting
stockouts and excess inventory, ultimately improving anomalies in supply chain data. Reinforcement learning was
operational efficiency and cost-effectiveness. also incorporated to enable the DSS to improve over time
based on feedback, providing adaptability in dynamic
III. METHODOLOGY business environments. The combination of these models
ensures a well-rounded DSS framework capable of handling
The methodology section outlines the research design, diverse data-driven tasks across different industries.
system development approach, evaluation metrics, and a
specific case study to illustrate the practical impact of the AI-  Evaluation Metrics
powered DSS. Each element is critical for understanding the To measure the effectiveness of the AI-powered DSS,
structure and implementation of this framework and how it evaluation metrics were established that reflect both the
can be adapted to different business contexts for decision- technical performance of the models and user-centric
making enhancement. outcomes. The primary metrics include:

 Research Design  Accuracy: This metric assesses the correctness of the AI

This study employs a mixed-methods research design, model’s predictions and classifications. For example, in a
incorporating both qualitative and quantitative data to provide financial DSS, accuracy measures the system’s ability to
a comprehensive evaluation of the AI-powered DSS correctly classify high-risk versus low-risk credit applicants.
framework. The qualitative aspect includes a review of  Decision Quality: This metric evaluates the relevance and
existing literature and case studies to understand the practical applicability of the insights generated by the DSS. It involves
needs and challenges of decision support across different assessing if the system's recommendations align with
industries [33]. The quantitative component focuses on business goals and are actionable.
statistical analysis and model performance metrics to assess  Response Time: Especially critical for real-time decision-
the effectiveness of the AI-powered DSS in delivering making, response time measures how quickly the DSS
accurate, efficient, and actionable insights. processes data and delivers insights. Short response times are
essential for industries like retail and logistics, where timely
Data sources for this study include a blend of real-world decisions impact operational efficiency.
business datasets, industry-specific case studies, and  User Satisfaction: To gauge the end-user experience,
simulated data environments to test the adaptability and satisfaction surveys and usability tests are conducted. User
performance of the system. Real-world datasets from sectors satisfaction reflects the intuitiveness of the DSS and its ability
like finance, retail, and supply chain are used to train and to provide insights that add real value to decision-makers.
validate AI models, ensuring that the DSS is robust enough Together, these metrics offer a balanced view of the system’s
to handle diverse data types and operational scenarios. technical robustness, operational efficiency, and user
Simulated data environments allow for controlled testing of acceptance, providing a holistic assessment of its
system responses under various conditions, such as high data effectiveness.
velocity or unstructured data input, which mirror the demands
of contemporary business operations.  Case Study/Use Case
An illustrative case study within the finance industry
 System Development demonstrates the application of the AI-powered DSS and its
The system development process involved selecting potential impact on decision-making processes. The finance
appropriate tools, technologies, and AI models to build an sector, with its reliance on data-driven insights for risk
adaptive and scalable AI-powered DSS framework. For the assessment, investment analysis, and regulatory compliance,
tools and technologies, open-source platforms like Apache is well-suited for AI-enhanced decision support [29].
Hadoop and Apache Spark were used for big data processing
and real-time data handling, ensuring that the DSS can In this use case, the AI-powered DSS is deployed within
efficiently manage high-volume datasets. Python was chosen a financial institution to streamline credit risk assessment.
for its flexibility in developing AI models, supported by The system processes large volumes of historical customer
libraries such as TensorFlow and Scikit-Learn for machine data, transaction records, and credit history to predict the

IJISRT25FEB1475 www.ijisrt.com 1512

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
likelihood of default. Supervised learning algorithms are outcomes with higher precision. Prescriptive capabilities,
trained on historical data, and unsupervised clustering models such as recommending actions based on forecasts,
help segment customers based on risk profiles, allowing for empowered businesses to respond proactively to potential
tailored recommendations on loan offerings or interest rates. risks or opportunities.
Real-time decision-making is facilitated by the system’s
ability to incorporate new customer data on demand, enabling Improvements in data-driven decision-making were
the institution to adjust credit assessments as economic also significant, as AI enabled organizations to generate
conditions change [32]. actionable insights more rapidly, improving responsiveness
in data-heavy environments. This benefit is particularly
The impact of the AI-powered DSS in this scenario is valuable in fast-paced industries like finance and retail, where
significant. It not only enhances the accuracy of risk timely and accurate decisions are essential for competitive
predictions but also reduces response times, allowing for advantage.
faster loan approvals and a more streamlined customer
experience. By automating a portion of the decision-making  Discussion
process, the institution reduces operational costs associated The findings have critical implications for businesses
with manual reviews, while maintaining high standards of considering adopting AI-driven DSS. This system offers
risk management. Additionally, user satisfaction metrics substantial advantages, notably in adaptability, scalability,
show improved acceptance by loan officers, who benefit from and predictive accuracy, supporting organizations in making
the system’s intuitive UI and actionable insights, leading to quicker and more reliable decisions. However, successful
higher levels of trust and reliance on the DSS in everyday implementation requires investments in infrastructure, talent,
operations. and a commitment to aligning AI with organizational goals.

IV. RESULTS AND DISCUSSION The proposed framework has the potential to be
generalized across industries. While finance and retail were
This section highlights the outcomes of implementing the primary use cases, sectors like healthcare, logistics, and
the AI-powered DSS and explores the broader implications manufacturing can similarly benefit from AI-powered DSS
of these results for businesses and industries. by customizing models to address industry-specific needs,
such as patient care optimization, inventory management, or
 Results production forecasting.
The system's performance was evaluated using metrics
such as decision accuracy, speed, and quality, compared to However, it is important to acknowledge the limitations
traditional DSS methods. In terms of decision accuracy, the revealed in this analysis. Although the AI-driven DSS
AI-powered DSS outperformed conventional models by 15- outperformed traditional systems, challenges remain
20%, particularly in scenarios requiring complex pattern regarding data integration and model scalability.
recognition or unstructured data analysis, like sentiment Furthermore, the system’s reliance on historical data could
analysis and risk prediction. The enhanced speed of decision- introduce biases, highlighting the need for transparent
making, achieved through real-time data processing, reduced governance and periodic model evaluation to ensure ethical
average decision latency by up to 40%, enabling faster and accurate outcomes.
responses to market shifts and operational changes. Decision
quality also improved, as AI-driven insights were V. CHALLENGES AND LIMITATIONS
consistently more precise and actionable, facilitating data-
informed decisions that align closely with business While AI-powered DSS presents transformative
objectives. potential, there are several key challenges and limitations
associated with its adoption.
The AI-powered system demonstrated a significant
advantage over traditional DSS. Traditional DSS, often  Technological Limitations
reliant on rule-based or statistical methods, typically showed One major technological limitation is data integration.
limitations in adapting to large datasets or diverse data types, Businesses typically gather data from diverse sources such as
often requiring manual intervention. The AI-driven CRM systems, social media, and IoT sensors—and
framework, by contrast, allowed for automated analysis and integrating these into a cohesive data stream for AI analysis
adjustment based on real-time data, improving adaptability is complex. Data inconsistency, varying formats, and real-
and reducing human dependency. This comparison revealed time data requirements all add to this challenge, potentially
the AI-powered DSS's strength in handling dynamic data and impacting the system’s ability to provide timely, accurate
making nuanced decisions beyond the capabilities of insights.
traditional systems.
Another hurdle is the scalability of AI models for real-
 Key Findings time decision-making. As datasets grow and decision-making
The results underscore AI's impact on predictive and requirements increase, scaling AI models to handle large
prescriptive decision-making within business intelligence volumes of data in real time becomes computationally
(BI). Predictive accuracy, supported by machine learning intensive. Current solutions may not be viable for small—to
algorithms, allowed the DSS to anticipate trends and mid-sized businesses without access to high-performance

IJISRT25FEB1475 www.ijisrt.com 1513

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
computing resources or cloud services, which can result in The proposed framework's core components—
delayed decision times or reduced model accuracy. including the data layer, AI models, interactive user interface,
knowledge base, and feedback mechanism—collectively
 Ethical and Security Concerns create a robust decision support infrastructure. This system
The ethical and security aspects of AI in DSS are empowers businesses not only to interpret vast and varied
increasingly significant. Transparency and explainability are datasets but also to forecast trends, identify potential risks,
ongoing concerns, as AI models, particularly complex ones and optimize resource allocation. By automating data
like deep learning, often operate as “black boxes” that make handling and offering data-driven recommendations, this
it difficult for users to understand the basis for certain framework enhances the effectiveness of decision-making
decisions. This opacity can lead to trust issues, especially if processes, making it adaptable to a wide range of industries.
decisions impact customer outcomes or carry legal
implications.  Contribution to the Field
This research contributes to bridging the gap between
Data privacy and security risks also intensify with AI- existing literature on AI and traditional DSS by presenting a
driven systems, as they rely on large datasets, often structured framework that delineates how AI can be
containing sensitive information. Safeguarding this data incorporated within BI systems to advance decision-making
against breaches and unauthorized access requires robust practices. While existing studies have primarily explored AI
security protocols, but even with these measures, privacy and DSS separately, this paper provides a unified approach
concerns remain. Striking a balance between data utility for that addresses technical, ethical, and operational challenges
AI processing and individual privacy rights is a significant in a practical, business-focused context. The findings
challenge for organizations, especially given evolving data underscore the strategic advantage of AI-driven DSS and
protection regulations. outline the specific components and workflow required to
implement these systems effectively.
 Resource Constraints
The high computational and financial costs of AI In practical terms, this research offers a roadmap for
integration represent another limitation. Implementing AI businesses aiming to integrate AI into their decision-making
models requires substantial computational resources, processes. By identifying core components and evaluating
including high-performance hardware and potentially costly performance metrics like decision accuracy, response time,
cloud solutions for scaling. For many organizations, the and user satisfaction, this paper supplies a comprehensive
financial investment needed for data storage, processing, and guide for companies looking to make informed investments
skilled AI professionals is substantial, which can limit the in AI technology. The use case analysis further illustrates
adoption of AI-powered DSS, especially in smaller how an AI-driven DSS can be applied in industry-specific
businesses. scenarios, underscoring its adaptability across sectors such as
finance, retail, and supply chain management.
While AI-powered DSS frameworks offer notable
advancements in decision-making capabilities, challenges REFERENCES
related to technological limitations, ethical considerations,
and resource constraints must be managed carefully. [1]. Henke, N. and L. Jacques Bughin, The age of
Businesses considering adoption should weigh these factors, analytics: Competing in a data-driven world. 2016.
establish clear governance, and explore cost-effective ways [2]. M Alasiri, M. and A.A. Salameh, The impact of
to mitigate challenges while maximizing the system’s business intelligence (BI) and decision support
benefits. systems (DSS): Exploratory study. International
Journal of Management, 2020. 11(5).
VI. CONCLUSION [3]. Rossille, D., J.-F. Laurent, and A. Burgun, Modelling
a decision-support system for oncology using rule-
This study highlights the transformative potential of based and case-based reasoning methodologies.
integrating AI into decision support systems (DSS) to International journal of medical informatics, 2005.
enhance business intelligence (BI) and decision-making 74(2-4): p. 299-306.
capabilities. Through advanced machine learning models, [4]. Melão, N. and M. Pidd, A conceptual framework for
natural language processing, and real-time data processing, understanding business processes and business
AI improves decision accuracy, speed, and quality, allowing process modelling. Information systems journal, 2000.
organizations to gain deeper insights and respond to emerging 10(2): p. 105-129.
challenges and opportunities more effectively. The AI- [5]. Jovanovic, P., O. Romero, and A. Abelló, A unified
powered DSS framework proposed in this paper addresses the view of data-intensive flows in business intelligence
limitations of traditional DSS, particularly by handling the systems: a survey. Transactions on Large-Scale Data-
complexities of large, dynamic, and unstructured data sets. and Knowledge-Centered Systems XXIX, 2016: p. 66-
The comparative analysis indicates that AI-driven systems 107.
outperform conventional methods, enabling proactive [6]. Kockum, F. and N. Dacre, Project management
decision-making and fostering agility across business volume, velocity, variety: A big data dynamics
operations. approach. Advanced Project Management, 2021.
21(1).

IJISRT25FEB1475 www.ijisrt.com 1514

Volume 10, Issue 2, February – 2025 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165 https://doi.org/10.5281/zenodo.14964501
[7]. Ekbia, H., et al., Big data, bigger dilemmas: A critical [22]. Bharadwaj, L., Sentiment analysis in online product
review. Journal of the Association for Information reviews: mining customer opinions for sentiment
Science and Technology, 2015. 66(8): p. 1523-1545. classification. Int J Multidiscip Res, 2023. 5(5).
[8]. Boppiniti, S.T., Real-time data analytics with ai: [23]. Ong, I.L., P.H. Siew, and S.F. Wong, A five-layered
Leveraging stream processing for dynamic decision business intelligence architecture. Communications of
support. International Journal of Management the IBIMA, 2011. 2011: p. 1-11.
Education for Sustainable Development, 2021. 4(4). [24]. Ramachandran, K., et al. AI-powered decision making
[9]. Hull, R. and R. King, Semantic database modeling: in management: a review and future directions. in
Survey, applications, and research issues. ACM 2023 3rd International Conference on Advance
Computing Surveys (CSUR), 1987. 19(3): p. 201-260. Computing and Innovative Technologies in
[10]. Azad, P., et al., The role of structured and unstructured Engineering (ICACITE). 2023. IEEE.
data managing mechanisms in the Internet of things. [25]. Heilig, T. and I. Scheer, Decision Intelligence:
Cluster computing, 2020. 23: p. 1185-1198. Transform Your Team and Organization with AI-
[11]. Cornwell, N., et al., The role of data analytics within Driven Decision-Making. 2023: John Wiley & Sons.
operational risk management: A systematic review [26]. Younas, M., Research challenges of big data. Service
from the financial services and energy sectors. Journal Oriented Computing and Applications, 2019. 13: p.
of the Operational Research Society, 2023. 74(1): p. 105-107.
374-402. [27]. Angelov, P.P., Evolving rule-based models: a tool for
[12]. Soori, M., et al., AI-based decision support systems in design of flexible adaptive systems. Vol. 92. 2002:
Industry 4.0, A review. Journal of Economy and Springer Science & Business Media.
Technology, 2024. [28]. Nasteski, V., An overview of the supervised machine
[13]. 1Schadt, E.E., et al., Computational solutions to large- learning methods. Horizons. b, 2017. 4(51-62): p. 56.
scale data management and analysis. Nature reviews [29]. Usman, M., M. Moinuddin, and R. Khan, Unlocking
genetics, 2010. 11(9): p. 647-657. insights: harnessing the power of business intelligence
[14]. Osasona, F., et al., Reviewing the ethical implications for strategic growth. International Journal of
of AI in decision making processes. International Advanced Engineering Technologies and Innovations,
Journal of Management & Entrepreneurship Research, 2024. 1(4): p. 97-117.
2024. 6(2): p. 322-335. [30]. Saravanan, R. and P. Sujatha. A state of art techniques
[15]. Patel, K., Ethical reflections on data-centric AI: on machine learning algorithms: a perspective of
balancing benefits and risks. International Journal of supervised learning approaches in data classification.
Artificial Intelligence Research and Development, in 2018 Second international conference on intelligent
2024. 2(1): p. 1-17. computing and control systems (ICICCS). 2018. IEEE.
[16]. Valle-Cruz, D., J.P. Muñoz-Chávez, and R. García- [31]. Chowdhary, K. and K. Chowdhary, Natural language
Contreras, Navigating the Fourth Industrial processing. Fundamentals of artificial intelligence,
Revolution: Empowering Socio-Technical 2020: p. 603-649.
Organizations With Data-Driven Business Intelligence [32]. Oliver, J., Continuous improvement: role of
Systems, in Data-Driven Business Intelligence Systems organisational learning mechanisms. International
for Socio-Technical Organizations. 2024, IGI Global. Journal of Quality & Reliability Management, 2009.
p. 1-27. 26(6): p. 546-563.
[17]. Keen, P.G., Decision support systems: a research [33]. Mozumder, M.A.S., et al., Revolutionizing
perspective. Decision Support Systems: Issues and Organizational Decision-Making for Banking Sector:
Challenges, International Institute for Applied A Machine Learning Approach with CNNs in Business
Systems Analysis (IIASA) Proceedings Series, 1980. Intelligence and Management. Journal of Business and
11: p. 23-27. Management Studies, 2024. 6(3): p. 111-118.
[18]. Power, D.J. and R. Sharda, Model-driven decision
support systems: Concepts and research directions.
Decision support systems, 2007. 43(3): p. 1044-1061.
[19]. Abbasi, M.A., et al., Addressing the future data
management challenges in iot: A proposed framework.
International Journal of Advanced Computer Science
and Applications, 2017. 8(5).
[20]. Matheus, R., M. Janssen, and D. Maheshwari, Data
science empowering the public: Data-driven
dashboards for transparent and accountable decision-
making in smart cities. Government Information
Quarterly, 2020. 37(3): p. 101284.
[21]. Klashner, R. and S. Sabet, A DSS design model for
complex problems: Lessons from mission critical
infrastructure. Decision Support Systems, 2007.
43(3): p. 990-1013.

IJISRT25FEB1475 www.ijisrt.com 1515