Demystifying AI-enabled Information Systems for Smarter Business Decisions

Introduction

In today’s competitive business landscape, organizations can no longer rely on intuitions to make important business decisions. The impact of every decision carries the weight of potential success or failure, with a single misstep potentially jeopardizing an organization’s very existence. Poor decision-making is estimated to cost firms on average at least 3% of profits, which for a $5 billion company amounts to a loss of around $150 million each year (Purdy, 2023). Yet, some companies navigate uncertainty with remarkable precision and achieve sustained growth. This is because they are able to leverage data and draw insights before taking any business decisions. However, with abundance growth in data, information processing by humans is overwhelming. This is where AI-powered information systems step in.

AI-enabled decision-making involves integrating AI technologies and algorithms within the organization’s current environment to enhance decision-making accuracy, speed, and efficiency across various domains (Petrova, 2023). AI systems can churn through large volumes of data, identifying patterns and correlations that are normally difficult for humans and that could consume potential amounts of time for individuals. The use of AI offers objective, data-driven decisions, removing any human-centric biased interpretations.

Today, AI’s capabilities have extended beyond speed and accuracy. Organizations can analyse intricate relationships, identify subtle influences, and anticipate future trends that might escape human attention. It can process unstructured data like text and images, opening doors to previously untapped sources of information. This unlocks a world of possibilities, from optimizing marketing campaigns to predicting equipment failures before they occur, all based on a deeper understanding of the data landscape. While human expertise remains crucial, AI acts as a powerful amplifier, empowering businesses to make smarter, faster, and more informed decisions, propelling them into a future fueled by data-driven intelligence.

Different Phases of AI Evolution

In the early days, AI-enabled information systems were rule-based (Mikullovci, 2023), limited by pre-defined logic that struggled to adapt to changing situations or environments. They were good at tasks with clear-cut rules, like evaluating loan applications based on pre-defined criteria like credit score or income. However, those systems cannot be used beyond their programmed rules.

This gradually led to the emergence of augmented intelligence (Dordevic, 2023) that employs machine learning capabilities, empowering systems to learn from data and improve their capabilities over time. The next leap came with the rise of deep learning, mimicking the human brain’s structure with interconnected layers of artificial neurons. This unlocks extraordinary abilities like image and speech recognition, allowing systems to extract even deeper insights from information. Meta’s photo tagging and understanding to voice commands in smartphones use deep learning AI technologies. Further development came in the form of natural language processing (NLP), enabling information systems to understand and communicate in human language, paving the way for chatbots and virtual assistants.

Today, AI-powered systems integrate multiple techniques, combining rule-based logic with machine learning and deep learning for comprehensive information processing. This ‘fusion approach’ fosters highly effective systems, tackling real-world challenges with ever-increasing accuracy and sophistication. The rise of generative AI (Gen AI), which is based on large language models (LLMs) has fostered innovation and ushered in new ways of generative insights in a more comprehensive and time-efficient manner. Refer to Figure I for a visual representation of the evolution of artificial intelligence.

Figure 1.

Source: Tableau (n.d.)

Benefits of AI-enabled Decision-making

AI systems offer a multitude of benefits for business decision-making.

Personalized decision-making experiences: AI-tailored recommendations based on personal preferences and past buying behaviour enhance the user experience and decision-making. For instance, AI algorithms underpinning online retail platforms analyse buying patterns of thousands of customers and suggest recommended items based on customers’ shopping cart selection. Similarly, social media platforms use AI technologies to curate content and personalize feeds tailored to individual users based on their interactions on the platform. These personalized customer experiences enhance user engagement and contribute to revenue growth for enterprises.

Increased efficiency through automation: Organizations can use AI technologies to automate day-to-day routine tasks, ultimately improving their efficiency and productivity. Menial tasks like data entry, report generation, and customer service inquiries can be automated, freeing up employees’ time for more meaningful and impactful work. For instance, Hinduja Group Solutions (HGS) implemented AI-based automation to achieve a cost-effective customer experience (Desai & Cio, 2023). The company was able to meet customer expectations by automating routine calls with chatbots and self-service portals and using AI sentiment analysis to improve customer interactions. With AI-enabled automation, the company could operate with 40% of its service agents, with a potential cost savings of 50–55%.

Demand forecasting and improved strategic planning: AI enables data-driven, real-time decision-making for supply chain managers who deal with challenges such as supplier unpredictability, missing historical data, and various supply chain disruptions. AI tools use historical sales data, market trends, external factors like weather and social media buzz, and even customer sentiments to forecast demands at all product levels more accurately, minimizing stockouts and overstocking.

Walmart, a retail giant, uses AI/ML-driven inventory management systems to deliver seamless holiday shopping experience for its customers. These AI systems analyse historical sales data, online searches, page views, and future data such as weather patterns, macroeconomic trends, and local demographics. This integration allows the system to anticipate demand, which helps it to strategically place holiday items across distribution centres, fulfilment centres, and stores, optimizing the entire shopping experience (Walmart, 2023).

Improved risk mitigation through predictive analytics: AI algorithms allow businesses to take preventive measures instead of reacting to crises after they occur. This finds varied use cases like fraud detection, workforce safety in manufacturing units, etc. J. P. Morgan uses AI-powered LLMs for payment validation screening. This lowers the level of fraudulent activities, with account validation rejections cut down by 15–20% (Morgan, 2023). Similarly, Aditya Birla Sun Life Insurance Ltd., the life insurance arm of Aditya Birla, implemented AI platforms that merged conventional underwriting principles with advanced algorithms. This integration empowered insurers to make well-informed decisions and identify fraudulent activities through the analysis of extensive datasets related to applicants’ profiles, thereby contributing to effective risk mitigation (Aditya Birla Group, 2023).

Dealing with Concerns in AI-enabled Decision-making

As businesses increasingly turn to AI for decision-making, a spectrum of concerns has emerged, prompting a critical examination of the technology’s ethical, social, and operational implications. While AI promises unparalleled efficiency and insights, certain use cases have raised eyebrows, underscoring the need for a thoughtful approach to implementation.

Algorithmic bias: AI systems learn from data, which can unknowingly encode societal prejudices. Even with unbiased data, the way algorithms process information can inadvertently amplify existing biases or create new ones. For example, an algorithm trained on biased hiring data might disproportionately favour certain demographics. A notable case involved Amazon discontinuing the use of AI algorithms when it was revealed that the algorithm exhibited a bias towards applicants having certain parameters in resumes that were more prevalent in men than women (Carnegie Mellon University, 2018).

Lack of transparency: One of the concerns in using AI models is the lack of transparency in their decision-making mechanisms. These deep learning models can sometimes be so complex that even the AI developers find it difficult to comprehend the underlying logic behind specific choices made by these AI models. For businesses, this lack of transparency is a big concern. Decision-makers want to know how AI reaches conclusions because it affects their choices. When the inner workings of AI are unclear, it is hard for them to trust the technology. This lack of trust can make businesses hesitant to fully embrace and rely on AI advancements.

Security and privacy concerns: AI technology not only enables the processing of vast amounts of data but also uses data to learn, develop adaptive models, and make actionable predictions. This presents significant concerns regarding both data privacy and security. Companies utilizing open-source LLMs for decision-making may unintentionally divulge their organization’s confidential information, potentially enabling hackers and malicious entities to compromise the company’s security protocols.

Hallucinations: AI systems, such as the latest version of ChatGPT, have been observed generating responses to queries that lack factual basis, leading to what is commonly referred to as ‘hallucinations’. These misleading outputs can pose significant risks for businesses that heavily depend on AI for decision-making. In critical applications like autonomous vehicles or power grids, AI-induced hallucinations may result in unintended actions, potentially causing safety hazards and financial harm.

Recommendations: Implementing and Using AI Systems in an Effective Manner

AI has emerged as a key tool in corporate decision-making and is finding growing application in shaping policies and public sector initiatives globally. However, the effective utilization of AI for strategic decision-making remains a capability limited to a select few organizations. Prioritizing the following actions will empower organizations to harness AI technology effectively within their decision-making processes.

Focus on human—AI collaboration: Human-AI collaboration (Ambrozi, 2023) holds immense potential for revolutionizing business decision-making. The synergy between human intelligence and artificial intelligence offers a unique avenue for innovation, productivity, and societal progress. While AI excels in automating routine tasks and processing vast amounts of data to uncover patterns, its true strength emerges when integrated with human cognitive abilities, such as creativity, emotional intelligence, and critical thinking. This collaboration enhances decision-making processes by providing valuable insights and data-driven recommendations, elevating the outcomes beyond what either humans or AI could achieve individually. Embracing AI also opens up opportunities for the workforce to upskill and focus on higher-level activities, fostering a symbiotic relationship that enhances productivity, improves decision-making, and promotes overall well-being.

Choose the right AI model: Without jumping on the latest trend, organizations should select an AI model that aligns with their problem, data, and technical capabilities. Various AI models exhibit expertise in different domains, such as image recognition, natural language processing, or predictive analytics. It is crucial to align the chosen model with the specific nature of the business problem at hand. Additionally, the selected model should demonstrate adaptability to evolving business requirements and the capacity to handle growing data volumes as the organization expands. Evaluating the seamless integration of the chosen AI model with the existing organizational infrastructure is equally vital for successful implementation.

Consider ethical implications: Prioritizing ethical guidelines and principles is essential for organizations to mitigate biases and prevent unintended consequences in AI systems. Regular checks and audits in AI models can enhance fairness and transparency. Organizations should establish robust protocols for handling sensitive information and adhere to regulatory frameworks while implementing AI systems to protect user privacy. As technology evolves, ethical guidelines must be reassessed at regular intervals to address emerging challenges and maintain ethical AI practices.

Case Study: Maersk Optimizes Logistics Using AI

Maersk, a leading Danish transportation, shipping, and logistics company, faced a significant challenge in determining competing performance indicators in their operations. The dilemma revolved around prioritizing between speed (rapid loading and unloading of ships or trucks, increasing throughput but potentially incurring higher short-term costs) and reliability (maintaining a consistent schedule during the loading process, potentially sacrificing some speed but ensuring on-time arrivals). Human judgement struggled to determine the optimal balance between these objectives due to the complex interplay of factors throughout the entire supply chain.

Maersk’s data team leveraged AI to analyse vast amounts of data and model different scenarios. The AI identified a counterintuitive solution: prioritizing reliability (slower loading/unloading) resulted in better overall outcomes. The AI-driven analysis contradicted the port managers’ experience, suggesting that while increasing speed at individual ports might improve local throughput, it created bottlenecks elsewhere, negating overall gains. Maersk built a model to simulate the impact of both approaches (speed and reliability) across their entire value chain. The model confirmed the AI’s findings.

By prioritizing reliability over speed based on AI insights, Maersk achieved measurably better organizational alignment, translating into tangible improvements in outcomes. The AI-driven approach not only optimized performance but also enhanced decision-making by providing a nuanced understanding of the intricate relationships between various factors. Maersk’s case demonstrates how AI can play a pivotal role in resolving complex business challenges and steering organizations towards more informed and effective strategies (Fortune, 2023).

Conclusion: The Way Forward

The integration of AI into decision-making processes is no longer a distant future but a present reality. While challenges and ethical concerns exist, the potential for AI-powered decision-making to propel businesses forward is undeniable. By integrating AI tools, businesses can gain a competitive edge, optimize operations, and unlock new possibilities. However, responsible implementation is crucial. Businesses must actively address ethical concerns, prioritize human oversight, and foster transparency in their AI-driven decision-making processes. Embracing AI with careful planning, ethical considerations, and continuous learning can equip organizations to navigate the ever-evolving business landscape with increased efficiency and more informed insights.