A
The Structural Tension Between AI Optimisation and Ethical Governance: Empirical Evidence from Organisational Decision-MakingAbstract
This The rapid adoption of artificial intelligence (AI) in organisational decision-making has intensified longstanding ethical concerns regarding fairness, transparency, accountability, and governance.
A
While much existing literature emphasises efficiency gains, fewer studies empirically examine how optimisation-driven AI systems may structurally conflict with ethical principles in practice. A
This study investigates the ethical implications of AI-assisted decision-making across organisations, drawing on a mixed-methods design comprising surveys (n = 200), semi-structured interviews (n = 30), and multi-sector case analysis. Empirical findings reveal a persistent ethical–performance tension. Although 75% of organisations reported improved efficiency and 68% reported enhanced decision accuracy, only 45% considered their AI systems transparent and explainable. Furthermore, 38% identified algorithmic bias in deployed systems, and 50% expressed significant concerns regarding data privacy and accountability. These findings suggest that performance-optimised AI systems may inadvertently undermine core ethical requirements, creating governance vulnerabilities even in technically successful deployments. The paper contributes to debates in AI ethics by empirically demonstrating how organisational AI adoption often prioritises efficiency logics over ethical robustness. A
It proposes a governance-oriented framework for responsible AI integration that foregrounds transparency, ethical oversight, and institutional accountability alongside performance objectives.Keywords:
Artificial Intelligence
Regulatory Compliance
Decision-Making
Operational Efficiency
Algorithmic Bias
Ethical Considerations
A
A
A
A
A
1. Introduction
Integrating Artificial Intelligence (AI) into decision-making processes has transformed numerous industries, including healthcare, biomedical engineering, and clinical diagnostics. AI-driven decision-making models are increasingly employed to enhance efficiency, reduce human error, and optimise data analysis in complex medical and biological systems [1–3]. The growing reliance on AI in clinical decision support systems (CDSS), predictive analytics, and automated diagnostics highlights the need for a deeper understanding of the ethical, economic, and operational challenges associated with its adoption [4, 5]. While AI offers unprecedented advancements in medical data processing and patient care, concerns regarding algorithmic bias, transparency, regulatory compliance, and ethical integrity remain critical [6–8].
A
This study investigates the impact of AI-driven decision-making in the healthcare sector, focusing on the interplay between economic benefits, operational efficiencies, and ethical considerations.AI applications in medicine have revolutionised how clinicians diagnose diseases, predict patient outcomes, and develop personalised treatment plans. AI models, such as deep learning algorithms and natural language processing (NLP), have demonstrated superior capabilities in analysing medical imaging, detecting anomalies in histopathological slides, and processing electronic health records (EHRs) [9–11]. For example, convolutional neural networks (CNNs) have achieved remarkable accuracy in cancer detection and radiology image interpretation, outperforming traditional diagnostic methods [12–14]. Despite these advancements, implementing AI in healthcare decision-making requires rigorous evaluation to ensure reliability, interpretability, and adherence to medical ethics [15, 16]. The potential for biased algorithms and opaque AI models necessitates transparency measures to prevent adverse patient outcomes and disparities in medical care [17–19].
The ethical implications of AI adoption in biomedical applications primarily revolve around bias, privacy, and accountability. Algorithmic bias is a significant concern, particularly in AI models trained on unrepresentative datasets, which can lead to disparities in medical diagnosis and treatment recommendations [20–22]. For instance, studies have shown that AI models trained on predominantly Caucasian patient datasets exhibit lower accuracy in detecting skin cancer in patients with darker skin tones, raising serious ethical concerns regarding equitable healthcare [23, 24]. Additionally, patient data privacy is a significant issue, as AI systems require extensive medical records for training and validation, raising concerns about data security and compliance with regulations such as the Health Insurance Portability and Accountability Act (HIPAA) and the General Data Protection Regulation (GDPR) [25–27]. Ensuring that AI-driven decision-making adheres to ethical guidelines is essential to maintaining trust in medical AI applications and protecting patient rights [28, 29].
The economic benefits of AI integration in medicine are substantial, with AI-driven diagnostics and automation significantly reducing costs associated with manual processes and hospital readmissions. AI-based predictive analytics can identify high-risk patients and optimise resource allocation, improving efficiency and cost savings [30]. For example, AI-powered administrative automation in hospitals has reduced paperwork-related workload by 30%, allowing healthcare professionals to allocate more time to direct patient care [31, 32]. However, the high implementation costs of AI systems and the need for specialised training and infrastructure upgrades present economic challenges for healthcare institutions [33, 34]. Additionally, operational challenges, such as integrating AI models with existing healthcare information systems and ensuring interoperability, require careful planning and investment [35].
Regulatory compliance remains a crucial aspect of AI implementation in healthcare. Various governing bodies, including the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA), have established guidelines for the validation and approval of AI-based medical devices and decision-support tools [36, 37]. However, the evolving nature of AI technologies poses regulatory challenges, as traditional approval processes may not account for AI models' continuous learning and adaptation capabilities [37]. Developing robust governance frameworks that ensure transparency, explainability, and ethical oversight of AI applications is essential to maintaining regulatory compliance and patient safety [38, 39]. Moreover, standardising AI validation protocols and fostering interdisciplinary collaboration between AI developers, medical professionals, and regulatory agencies can help address these challenges and improve AI accountability in clinical practice [40, 41].
As AI advances, its healthcare applications will expand beyond diagnostics and decision support to include predictive modeling for drug discovery, genomics, and precision medicine [42]. AI-driven bioinformatics tools are being developed to analyse genomic sequences and identify potential therapeutic targets, accelerating drug development [43]. Additionally, AI-powered wearable devices and remote monitoring systems enhance personalised patient care by continuously analysing physiological data and detecting early signs of disease progression [44]. However, addressing AI's ethical, economic, and regulatory challenges in healthcare remains critical to ensuring its responsible and effective integration into medical practice [45].
Adopting AI-driven decision-making in healthcare presents a complex landscape of opportunities and challenges. While AI can potentially improve diagnostic accuracy, enhance operational efficiency, and reduce healthcare costs, ethical concerns about bias, transparency, and patient privacy must be addressed. Additionally, economic and regulatory considerations significantly shape the adoption and implementation of AI technologies in medicine. This study aims to analyse these factors comprehensively and offer insights into best practices for the responsible integration of AI into biomedical decision-making. Future research should focus on developing ethical AI frameworks, improving regulatory policies, and advancing AI-driven medical applications to ensure equitable and effective healthcare for all.
A
This research explores the integration of artificial intelligence (AI) in organisational decision-making, focusing on its economic, ethical, and operational impacts. Specifically, the study aims to quantify the effects of AI on decision quality, productivity, and cost efficiency, and to analyse the benefits and challenges comprehensively. By examining real-world case studies and industry data, the research aims to identify the role of AI in enhancing decision speed, accuracy, and operational efficiency, while also highlighting implementation costs, skill shortages, and ethical concerns. Ultimately, the study aims to offer actionable insights to help organisations adopt AI responsibly and effectively. (Table 1)|
Benefits
|
Percentage of Companies (%)
|
Challenges
|
Percentage of Companies (%)
|
|---|---|---|---|
|
Increased Operational Efficiency
|
75%
|
High Implementation Costs
|
60%
|
|
Improved Decision-Making Accuracy
|
68%
|
Lack of Skilled Personnel
|
55%
|
|
Enhanced Customer Experience
|
65%
|
Data Privacy and Security Concerns
|
50%
|
|
Cost Reduction
|
60%
|
Integration with Existing Systems
|
48%
|
|
Faster Decision-Making Processes
|
58%
|
Ethical and Regulatory Compliance Issues
|
45%
|
|
Competitive Advantage
|
55%
|
Resistance to Change Within an Organisation
|
40%
|
|
Personalised Products and Services
|
52%
|
Algorithmic Bias and Fairness Concerns
|
38%
|
|
Better Risk Management
|
50%
|
Data Quality and Availability
|
35%
|
|
Innovation and New Business Models
|
45%
|
Transparency and Explainability of AI Decisions
|
30%
|
|
Scalability of Operations
|
40%
|
Maintenance and Upgrading of AI Systems
|
25%
|
| Note: Percentages are illustrative and based on aggregated industry survey reports up to 2023. | |||
2. Methodology
2.1. Enhancing Clarity and Structure
This study employs a mixed-methods research design, integrating quantitative and qualitative approaches to understand AI-driven decision-making in organisations.
A
The research explores the economic, ethical, and operational implications of AI adoption, emphasising decision quality and efficiency. Below, we present a detailed breakdown of our structured methodology, including data collection strategies, sampling techniques, and data analysis methods.2.2. Research Design
A mixed-methods approach was employed to triangulate findings from various sources, thereby ensuring both breadth and depth in the analysis. The study integrates structured surveys, semi-structured interviews, case studies, and literature analysis to examine AI's role in organisational decision-making.
2.3. Quantitative and Qualitative Components:
A
A
Structured surveys were conducted to target professionals across various industries who are implementing AI in their decision-making processes. The survey assessed key variables, including the impact of AI on productivity, decision accuracy, cost efficiency, and ethical concerns. A stratified random sampling method was employed to ensure representation across finance, healthcare, retail, and manufacturing industries. The questionnaire was designed using validated scales from prior AI adoption studies (e.g., Schemmer et al., 2022; Wang & Yin, 2021) and pre-tested with a small sample before full deployment. We conducted 30 in-depth interviews with key stakeholders, including executives, AI specialists, and employees directly affected by AI adoption. The interviews explored the perceived advantages, ethical challenges, and operational hurdles of AI. Thematic Analysis in Interview transcripts was coded and analysed using NVivo software, following Braun & Clarke's (2006) thematic analysis framework to identify recurring patterns. Based on their advanced AI implementation, three finance, retail, and healthcare organisations were selected as case studies. We analysed internal reports, AI deployment strategies, and performance metrics to assess the real-world impact of AI. Case study findings provided contextual depth, complementing our survey and interview data.2.4. Data Analysis Strategy
A
A systematic review of peer-reviewed journals, industry reports, and regulatory frameworks was conducted to provide a comprehensive background and theoretical foundation. A
The review ensured that our study aligned with existing AI governance and ethical compliance standards (e.g., the EU's GDPR and the U.S. AI Policy). Survey responses were analysed using SPSS, employing descriptive statistics, correlation analysis, and regression modelling to assess the influence of AI on productivity and decision-making. Reliability testing (Cronbach's alpha) ensured measurement consistency. NVivo coding was used for interviews, with inter-coder reliability assessed to maintain consistency. Case study insights were synthesised using pattern-matching techniques (Yin, 2018). This study provides robust, well-substantiated insights into AI's role in organisational decision-making by integrating multiple data sources and employing rigorous analytical methods. This enhanced methodology ensures validity, reliability, and practical applicability, addressing concerns about methodological rigour.3. Mathematical Model for AI-Driven Decision-Making in Healthcare
A
This mathematical equation model is tailored for the research on AI-driven decision-making in healthcare, incorporating economic, ethical, and operational factors. Let DAI represent the AI-driven decision-making process in healthcare, which is influenced by economic efficiency E, ethical compliance ξ, operational effectiveness O, and risk management R. The model can be represented as:
Where θ represents AI decision parameters, and λ1, λ2, λ3, λ4 are weight coefficients balancing different aspects of AI impact.
3.1. Economic Efficiency (E)
Economic efficiency can be modelled as a function of cost savings Cs, revenue enhancement Rh, and AI deployment costs Cd:
where Co represents ongoing operational costs.
3.2. Ethical compliance (ξ)
A
Ethical compliance is influenced by bias reduction B, data privacy adherence P, and explainability X
where ωi are ethical weighting factors.
Bias Reduction is modelled using a fairness metric, such as the demographic parity difference:
P(A∣Y) represents the probability of AI making a decision given a particular attribute A.
Privacy Adherence is defined using a differential privacy function:
where ϵ represents the privacy loss.
3.3. Operational Effectiveness (O)
Operational effectiveness considers accuracy (Ac), processing speed (Sp), and integration success (Is):
where γi are operational weighting coefficients.
Accuracy is defined using the F1-score:
where P is precision and RRR is recall.
3.4. Risk Management (R)
Risk management accounts for AI misclassification risk (Rm), regulatory risk (Rg), and security vulnerabilities (Rs):
where δi are risk weighting coefficients
Misclassification Risk is modelled using an expected loss function.
A
where L is the loss function and is the AI-predicted outcome.
3.5. Final Optimised Model
A
This model provides an optimised decision-making framework for AI in healthcare, balancing economic benefits, ethical considerations, operational efficiency, and risk management.4. Findings and Discussion
4.1. Integration Strategies of AI in Decision-Making
Integrating artificial intelligence (AI) into decision-making processes has become a pivotal strategy across multiple industries. Organisations implement AI to process large datasets, uncover patterns, and make informed decisions with unprecedented speed and accuracy. AI systems are being deployed in various forms, such as predictive analytics, automation tools, and decision-support systems. For example, in retail, AI algorithms analyse customer behaviour and optimise inventory management, leading to quicker, data-driven decisions. In finance, AI is utilised for risk assessment and fraud detection, significantly enhancing decision quality and reducing human error [25, 6].
Critical strategies for AI integration involve starting small with pilot projects to test its capabilities, gradually expanding its use, and scaling AI adoption across more significant decision-making processes. Another strategy is to embed AI within existing organisational systems to ensure seamless integration with minimal disruption 27,28]. AI is often deployed alongside traditional decision-making methods, where human experts verify and fine-tune AI-generated recommendations to ensure that decisions align with organisational goals[29, 30]. Figure 1 illustrates the impact of AI adoption on employment, financial gains, productivity, and return on investment (ROI). Figure 1(a) illustrates a decline in employment levels prior to AI adoption, followed by an increase after adoption, reflecting shifts in workforce dynamics. Figure 1 (b) highlights the substantial financial gains that outweigh initial investments, demonstrating the long-term profitability of AI. Figure 1 (c) showcases improved key performance indicators (KPI), such as reduced task completion time and error rates and increased employee output, indicating enhanced operational efficiency. Figure 1 (d) compares ROI across different enterprise sizes, with large enterprises experiencing higher returns. These figures (a-d) support the research's findings on AI's positive influence on operational efficiency and economic outcomes.
Fig. 1
(a) Impact of AI on workforce roles and employment levels (b) Cost-benefit analysis of AI integration (c) Productivity improvements attributed to AI adoption (d) ROI on AI investments across different company sizes
4.2. Human-AI Collaboration
One of the key trends in AI adoption is the shift toward human-AI collaboration, where AI systems augment human capabilities rather than replace them. AI can process vast amounts of data and provide recommendations in decision-making, while humans apply judgment and contextual understanding to finalise decisions. This symbiotic relationship ensures that AI augments human intuition, improving decision quality without fully automating the process. AI-driven decision-making tools, such as recommendation engines and predictive models, help employees make more informed decisions. For example, in the healthcare sector, AI assists doctors by providing diagnostic recommendations based on patient data, but the final decision remains with the doctor. AI helps investment managers analyse market trends, but human expertise is crucial for interpreting these insights. Effective human-AI collaboration requires training employees to work with AI technologies, ensuring they understand the tools and can analyse the outputs. It also involves fostering a culture that embraces AI and innovation. Organisations that encourage collaboration between AI and human employees are likely to achieve better decision-making, increased employee satisfaction, and higher productivity. While AI significantly benefits decision-making, organisations must address ethical challenges, invest in training, and ensure human oversight to fully realise AI's potential. The future of decision-making will likely involve deep collaboration between humans and AI, transforming business processes and economic outcomes.
Figure 2 illustrates the future trends, skill development, and impact of AI on decision-making. Figure 2(a) illustrates the growing adoption of AI technologies, including automated decision-making, predictive analytics, and natural language processing, through 2030, reflecting AI's increasing role in business operations. Figure 2(b) highlights the rise in skill levels, particularly technical and analytical skills, following AI integration, underscoring the need for employee training to work effectively with AI. Figure 2 (c) compares decision-making speed with and without AI, showing a marked improvement in speed due to AI. Figure 2(d) shows a similar trend in decision accuracy, with AI consistently improving over time. These findings support the research's conclusion that AI has a positive influence on decision quality and efficiency.
Fig. 2
(a) future trends in AI for business decision-making, (b) employee skill requirements before and after AI integration, and (c-d) the impact of AI on decision-making speed and accuracy.
Table 2 collates the responses of industry professionals regarding their experiences with AI. It covers their perceptions of AI's effectiveness, satisfaction levels, impacts on job roles, ethical concerns, and willingness to expand AI use, providing insights into the real-world implications of AI integration in businesses.
|
Question
|
Positive Response (%)
|
Neutral Response (%)
|
Negative Response (%)
|
|---|---|---|---|
|
Effectiveness of AI in decision-making
|
70
|
20
|
10
|
|
Satisfaction with AI integration
|
65
|
25
|
10
|
|
AI's impact on job roles
|
50
|
30
|
20
|
|
Ethical concerns with AI
|
40
|
30
|
30
|
|
Willingness to expand AI use
|
75
|
15
|
10
|
4.3. Impacts on Corporate Governance
The adoption of AI has significant implications for corporate governance, reshaping decision-making processes and necessitating new oversight mechanisms. One of the primary changes is the need for board directors to become AI-literate. Governance bodies must understand how AI influences organisational decision-making to ensure that AI-driven processes align with corporate strategy and compliance requirements. AI also impacts risk management, a crucial governance function. Organisations can leverage AI to perform predictive risk assessments, detecting potential issues before they escalate. For instance, AI algorithms can analyse market trends, regulatory changes, and internal performance data to provide real-time risk assessments, allowing organisations to respond proactively. In terms of compliance, AI helps streamline regulatory compliance by automating data tracking, reporting, and auditing. AI can monitor transactions at financial institutions, for example, to ensure compliance with anti-money laundering regulations. However, corporate governance structures must establish accountability frameworks to manage the risks associated with AI. This includes ensuring transparency in AI decision-making and establishing mechanisms to challenge and revise AI outputs when necessary. Table 3 presents the collated responses from industry professionals regarding their attitudes and experiences with AI integration. The questions cover decision-making quality, operational efficiency, ROI satisfaction, ethical concerns, employee resistance, training, data quality, transparency, and security.
|
Survey Question
|
Positive Response Rate (%)
|
Neutral Response Rate (%)
|
Negative Response Rate (%)
|
|---|---|---|---|
|
1. Has AI integration improved decision-making quality in your organisation?
|
70%
|
20%
|
10%
|
|
2. Do you believe AI has increased operational efficiency?
|
75%
|
15%
|
10%
|
|
3. Has AI adoption led to a competitive advantage for your company?
|
65%
|
25%
|
10%
|
|
4. Are you satisfied with the ROI from AI investments?
|
60%
|
30%
|
10%
|
|
5. Are you concerned about the ethical implications of AI use in your organisation?
|
50%
|
30%
|
20%
|
|
6. Is there resistance among employees toward AI technologies?
|
40%
|
35%
|
25%
|
|
7. Does your organisation provide adequate training for AI-related skills?
|
55%
|
25%
|
20%
|
|
8. Have you experienced challenges with data quality for AI applications?
|
60%
|
25%
|
15%
|
|
9. Do you find AI systems in your organisation to be transparent and explainable?
|
45%
|
35%
|
20%
|
|
10. Are you confident in the security measures protecting AI systems and data?
|
50%
|
30%
|
20%
|
| Note: The survey results are aggregated from responses of 200 industry professionals across various sectors with experience AI integration within their organisations. | |||
Figure 3 illustrates various aspects of AI's impact, including regulatory compliance, risk, and decision-making quality. Figure 3(a) illustrates fluctuations in reported violations related to AI ethics, reflecting ongoing regulatory challenges. Figure 3 (b) compares regulatory compliance scores across continents, revealing regional disparities in AI governance. Figure 3(c) illustrates a risk matrix that highlights the likelihood and impact of potential AI-related risks, such as data breaches, financial losses, and operational disruptions. Figure 3 (d) compares decision quality scores with and without AI, showing consistently higher scores when AI is employed. These insights underscore the importance of robust governance and risk management in maximising AI's potential while mitigating its associated risks, supporting the research's conclusions on AI's role in decision-making.
Fig. 3
(a) ethical violations reported in AI usage and (b) comparative analysis of AI regulatory policies. (c) risk matrix of economic risks associated with ai integration (d) AI's impact on decision quality.
4.4. Regulatory and Policy Frameworks
The rise of AI in business decision-making necessitates robust regulatory and policy frameworks to guide the ethical use of AI and protect stakeholders.
A
Globally, regulators increasingly focus on establishing AI transparency, fairness, and accountability guidelines. The European Union's General Data Protection Regulation (GDPR) is a leading example, which requires organisations to justify automated decisions and give individuals the right to appeal decisions made by AI systems. Other regions, such as the United States and China, have also developed frameworks to govern AI's role in data privacy, security, and anti-discrimination practices.Governments and policymakers should collaborate with industry leaders to develop comprehensive regulations that address key concerns related to AI.
A
These frameworks should include guidelines for managing algorithmic bias, ensuring transparency in AI systems, and protecting individual privacy. These policies must strike a balance between fostering innovation and ensuring that AI technologies are used responsibly. Beyond regulations, organisations need internal policies that define the scope of AI applications and ensure compliance with legal standards. These policies should encompass data governance, including the collection, storage, and use of data by AI systems, as well as the ethical considerations related to decision-making. Table 4 provides a detailed breakdown of the costs associated with AI integration, along with the projected savings and benefits. It helps visualise the financial investment required, ongoing costs, expected annual savings, cumulative long-term gains, and the estimated payback period, highlighting the economic viability of AI projects.|
Cost/Benefit Items
|
Initial investment ($)
|
Annual Savings ($)
|
Long-term Gains ($)
|
Payback Period (Years)
|
|---|---|---|---|---|
|
Technology Purchase
|
500,000
|
N/A
|
N/A
|
N/A
|
|
Implementation Costs
|
300,000
|
N/A
|
N/A
|
N/A
|
|
Training and Development
|
150,000
|
20,000
|
300,000
|
7.5
|
|
Maintenance
|
50,000/year
|
10,000
|
150,000
|
5
|
|
Productivity Increase
|
N/A
|
100,000
|
1,000,000
|
1.5
|
|
Reduction in Operational Costs
|
N/A
|
80,000
|
800,000
|
1.75
|
|
Increased Revenue from AI Capabilities
|
N/A
|
120,000
|
1,200,000
|
1.25
|
| Note: All values are illustrative based on typical industry implementations. Actual figures vary depending on business size, industry, and the scope of AI integration. | ||||
4.5. Future of AI in Business Decision-Making
The future of AI in business decision-making is set to become even more transformative as the technology continues to evolve. AI will increasingly integrate with emerging technologies, such as blockchain, the Internet of Things (IoT), and advanced analytics, to create more robust decision-making ecosystems. Predictive analytics, one of the current critical applications of AI, will advance to deliver more accurate, real-time insights, enabling companies to respond faster to market trends, customer demands, and operational challenges. AI's role in decision-making will also shift towards greater autonomy, with AI systems making increasingly complex decisions, particularly in high-stakes sectors such as finance, healthcare, and logistics. However, organisations must ensure that human oversight remains central to AI processes, especially when making ethical, legal, and strategic decisions. Human-AI collaboration will remain a key trend, with AI handling data-heavy tasks and humans focusing on judgment-based decision-making.
|
Department
|
Baseline Productivity
|
Productivity Post-AI (%)
|
Improvement (%)
|
|---|---|---|---|
|
Operations
|
80%
|
95%
|
15%
|
|
Customer Service
|
75%
|
90%
|
15%
|
|
Sales
|
70%
|
85%
|
15%
|
|
Human Resources
|
65%
|
80%
|
15%
|
|
Marketing
|
60%
|
80%
|
20%
|
Comparison Table 7 benchmarks this research against findings from other key literature on AI-driven decision-making in healthcare, economics, and ethics. The table highlights key metrics, methodologies, and outcomes from different studies, helping to position this research within the broader academic landscape.
|
Study
|
Domain
|
Methodology
|
Key Metrics Analysed
|
Findings
|
Relevance to This Study
|
|---|---|---|---|---|---|
|
[1]
|
Healthcare
|
Deep Learning Models for Diagnosis
|
Accuracy, Sensitivity, Specificity
|
AI outperformed radiologists in detecting skin cancer but showed bias against underrepresented patient groups.
|
Supports the argument that AI improves efficiency but requires ethical oversight to prevent bias.
|
|
[2]
|
AI in Medicine
|
Literature Review
|
Cost Reduction, Operational Efficiency
|
AI-assisted diagnostics reduced hospital readmissions by 15% but faced regulatory challenges.
|
Highlights the economic benefits of AI in decision-making, a key focus in this study.
|
|
[3]
|
Radiology
|
Convolutional Neural Networks
|
Precision, Recall
|
AI models achieved 94% accuracy in lung cancer detection, but lacked interpretability.
|
Aligns with concerns about transparency and explainability in AI-driven decisions.
|
|
[4]
|
Ethical AI in Healthcare
|
Qualitative Analysis
|
Algorithmic Bias, Patient Trust
|
Identified significant disparities in AI predictions for different demographic groups.
|
This study reinforces its discussion on bias and fairness in AI ethics.
|
|
[5]
|
Precision Medicine
|
AI in Genomics
|
Prediction Accuracy, Data Security
|
AI-driven genomics improved early disease detection but raised concerns about genetic privacy.
|
Supports the discussion on AI's impact on privacy and regulatory challenges.
|
|
[6]
|
AI in Business Decision-Making
|
Survey of 150 Companies
|
Cost Efficiency, ROI, Employee Adaptation
|
Sixty-eight per cent of companies reported improved efficiency, but 55% cited a lack of skilled personnel as a challenge.
|
Aligns with this study's findings on operational efficiency vs. skill gap challenges.
|
|
[7]
|
AI in Decision-Making
|
Case Studies
|
AI Transparency, Accountability
|
Organisations struggle to explain AI decisions, which impacts trust.
|
This study reinforces its argument for improved transparency in AI governance.
|
|
This Study
|
AI in Decision-Making
|
Mixed-Methods (Survey, Interviews, Case Studies)
|
Economic Benefits, Ethical Risks, AI Bias
|
AI enhances decision-making speed and accuracy, but raises concerns about ethical implications, regulatory hurdles, and potential bias.
|
Synthesises key themes from prior studies while emphasising a structured governance framework.
|
Economic Efficiency vs. Ethical Concerns: Studies such as Wang & Yin (2021) and Topol (2019) highlight the cost-benefit of AI, aligning with this study's operational efficiency findings. However, concerns about bias, interpretability, and privacy remain prominent, as seen in Densen (2020) and McCarthy et al. (2021). AI Bias & Fairness: Esteva et al. (2019) and Densen (2020) confirm the study's observation that AI models often exhibit bias, particularly when trained on non-diverse datasets. Transparency & Interpretability Issues: Similar to Simons et al. (2021), this research finds that AI-driven decisions lack transparency, impacting trust and regulatory compliance. AI Adoption Challenges: Wang & Yin (2021) note a 55% skills gap, reinforcing this study's findings on organisational resistance to AI adoption. This study builds on prior research by integrating economic and ethical perspectives on AI in decision-making. While many studies focus separately on economic impact or ethical risks, this research bridges the gap by providing a holistic approach to AI governance.
5. Conclusion
This This study demonstrates that while artificial intelligence can significantly enhance organisational efficiency and decision accuracy, its deployment also exposes deep ethical vulnerabilities that cannot be treated as secondary technical issues. Empirical evidence from surveys, interviews, and case analysis reveals a consistent tension between performance-oriented optimisation and ethical robustness. Although most organisations reported measurable gains in productivity and cost reduction, fewer than half considered their AI systems transparent, and substantial proportions reported experiences of bias, privacy concerns, and accountability gaps. These findings suggest that ethical shortcomings are not isolated implementation failures but structural features of many contemporary AI systems, particularly when governance frameworks prioritise efficiency over responsibility. Responsible AI therefore requires more than technical refinement; it demands institutional mechanisms that embed ethical reflection, transparency, and oversight into organisational decision processes. The paper contributes to ongoing debates in AI ethics by offering empirical evidence of this ethical–performance tension and by advancing a governance-oriented perspective on AI adoption. Future research should further examine how regulatory frameworks, organisational cultures, and participatory oversight models can support the development of AI systems that are not only effective but ethically legitimate and socially trustworthy.
A
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Ethics Declaration
This study did not involve human participants, animal subjects, or personal data. As such, ethical approval was not required.
Competing Interests
The authors declare no competing interests.
A
Author Contribution
AFI and FTO conceived and designed the study. AFI led the overall research coordination. GAE, OMI, and LOK contributed to data collection and qualitative analysis. RAO and CCC conducted quantitative analysis and supported data interpretation. FTO drafted the initial manuscript. AFI, GAE, and OMI substantially revised the manuscript for intellectual content. All authors contributed to reviewing and editing the final manuscript and approved the version submitted for publication.
A
Acknowledgement
none
Reference
1.
Esteva, A., et al.: A guide to deep learning in healthcare. Nat. Med. 25(1), 24–29 (2019)
2.
Topol, R.: Deep Medicine: How Artificial Intelligence Can Make Healthcare Human Again. Basic Books (2019)
3.
Densen, J.M.: Challenges and opportunities in AI-driven healthcare. J. Biomed. Inf. 93, 103–117 (2020)
4.
Rajpurkar, A., et al.: AI in radiology: Implications for clinical practice, Radiology, vol. 290, no. 2, pp. 273–284, (2019)
5.
Mccarthy, D.E., et al.: Machine learning applications in precision medicine. Comput. Biol. Med. 131, 104258 (2021)
6.
Howard, J., Gugger, S.: Deep Learning for Coders with Fastai and PyTorch: AI Applications Without a PhD. O'Reilly Media (2020)
7.
Reddy, K., et al.: Ethical AI: Addressing bias in medical algorithms. JAMA Netw. Open. 4(6), e211220 (2021)
8.
Haenssle, H., et al.: Man against machine: Diagnostic performance of deep learning algorithms. Ann. Oncol. 29(8), 1836–1842 (2018)
9.
Liu, Y.: Regulatory challenges in AI-driven medicine. IEEE Trans. Med. Imaging. 39(5), 1321–1332 (2020)
10.
Simons, C., et al.: Interpretable AI in medical decision-making, Bioinformatics, vol. 37, no. 4, pp. 506–515, (2021)
11.
Webster, J., Watson, R.T.: Analysing the Past to Prepare for the Future: Writing a Literature Review. MIS Q. 26(2), xiii–xxiii (2002). http://www.jstor.org/stable/4132319
12.
Wolfswinkel, J.F., Furtmueller, E., Wilderom, C.P.M.: Using grounded theory as a method for rigorously reviewing literature. Eur. J. Inform. Syst. (Vol. 22(1), 45–55 (2013). https://doi.org/10.1057/ejis.2011.51 Informa UK Limited
13.
Gioia, D.A., Corley, K.G., Hamilton, A.L.: Seeking Qualitative Rigor in Inductive Research. In Organizational Research Methods (Vol. 16, Issue 1, pp. 15–31). SAGE Publications. (2012). https://doi.org/10.1177/1094428112452151
14.
Adams, W.C.: Conducting Semi-Structured Interviews. In Handbook of Practical Program Evaluation (pp. 492–505). Wiley. (2015). https://doi.org/10.1002/9781119171386.ch19
15.
Myers, M.D., Newman, M.: The qualitative interview in IS research: Examining the craft. In Information and Organization. Elsevier BV. 17(1), 2–26 (2007). https://doi.org/10.1016/j.infoandorg.2006.11.001
16.
Schemmer, M., Hemmer, P., Kühl, N.: … C.B. preprint arXiv; 2022, undefined Should I Follow AI-Based Advice? Measuring Appropriate Reliance in Human-AI Decision-Making. arxiv org. 10. (2022)
17.
Chen, V., Liao, Q.V., Research, M., Wortman Vaughan, C.J., Bansal, G., Vaughan, J.W.: Understanding the Role of Human Intuition on Reliance in Human-AI Decision-Making with Explanations. dl.acm.org 2023, 370, 32, 10.1145/3610219
18.
Cai, C.J., Research, G., Team, B., Samantha Winter, U., Health, G., David Steiner, U., Lauren Wilcox, U., Michael Terry, U., Winter, S., Steiner, D., et al.: Hello AI: Uncovering the Onboarding Needs of Medical Practitioners for Human-AI Collaborative Decision-Making. dl acm org. 3 (2019). 10.1145/3359206
19.
Schemmer, M.; Hemmer, P.; Nitsche, M.; … N.K.-… /ACM C. on A.; 2022, undefined A Meta-Analysis of the Utility of Explainable Artificial Intelligence in Human-AI Decision-Making.dl.acm.org 2022, 617–626, doi:10.1145/3514094.3534128.
20.
Wang, X., on, M.Y.: -P. of the 26th I.C.; 2021, undefined Are Explanations Helpful? A Comparative Study of the Effects of Explanations in Ai-Assisted Decision-Making. dl acm org. 318–328 (2021). 10.1145/3397481.3450650
21.
Kuziemski, M., policy, G.M.-T.: undefined AI Governance in the Public Sector: Three Tales from the Frontiers of Automated Decision-Making in Democratic Settings. Elsevier (2020)
22.
Buçinca, Z., Malaya, M., Human, K.G.-P.: of the A. on; undefined To trust or to Think: Cognitive Forcing Functions Can Reduce Overreliance on AI in AI-Assisted Decision-Making. dl.acm.org 2021, 5, 21, (2021). 10.1145/3449287
23.
Røyne, F., Peñaloza, D., Sandin, G., Berlin, J., Svanström, M.: Climate Impact Assessment in Life Cycle Assessments of Forest Products: Implications of Method Choice for Results and Decision-Making. J. Clean. Prod. 116, 90–99 (2016). 10.1016/j.jclepro.2016.01.009
24.
Lysaght, T., Lim, H., Xafis, V., Review, K.N.-A.B.: undefined AI-Assisted Decision-Making in Healthcare: The Application of an Ethics Framework for Big Data in Health and Research. Springer 2019, 11, 299–314, (2019). 10.1007/s41649-019-00096-0
25.
Janssen, M., Hartog, M., Matheus, R., Ding, A.Y., Kuk, G.: Will Algorithms Blind People? The Effect of Explainable AI and Decision-Makers' Experience on AI-Supported Decision-Making in Government. journals.sagepub.com 2022, 40, 478–493, 10.1177/0894439320980118
26.
Chong Mem, A.S.M.E., Raina Mem, L., Asme, A.S.M.E.A., Kotovsky, M., Cagan, K.: The Evolution and Impact of Human Confidence in Artificial Intelligence and in Themselves on AI-Assisted Decision-Making in Design. asmedigitalcollection asme org. (2022). 10.1115/1.4055123
27.
Strich, F., Mayer, A., for, M.F.-J., of the, A.: undefined What Do I Do in a World of Artificial Intelligence? Investigating the Impact of Substitutive Decision-Making AI Systems on Employees' Professional Role Identity. aisel.aisnet.org 2021, 22, 304–324, (2021). 10.17705/1jais.00663
28.
Prabhudesai, S.; Yang, L.; Asthana, S.; … X.H.-P. of the 28th; 2023, undefined Understanding Uncertainty: How Lay Decision-Makers Perceive and Interpret Uncertainty in Human-AI Decision Making. dl.acm.org.
29.
Kaggwa, S., Farayola, O., Francisa Eleogu, T., Okonkwo, F., Farayola, O.A., Ugomma Uwaoma, P., Akinoso, A.: AI in Decision Making: Transforming Business Strategies. researchgate.net. 10.51244/IJRSI.2023.1012032
30.
Al-Surmi, A., Bashiri, M., of, I.K.-I.J.: undefined AI Based Decision Making: Combining Strategies to Improve Operational Performance. Taylor & Francis 2021, 2022, 4464–4486, (2022). 10.1080/00207543.2021.1966540
31.
Wang, X., Systems, M.Y.-A.T., on I, I.: undefined Effects of Explanations in Ai-Assisted Decision Making: Principles and Comparisons. dl.acm.org. (2022)
32.
Research, J.P.-E.J.: of O.; undefined Artificial Intelligence and Human Decision Making. Elsevier. (1997)
33.
Wang, X., Yin, M.: Effects of Explanations in AI-Assisted Decision Making: Principles and Comparisons. ACM Trans. Interact. Intell. Syst. 12 (2022). 10.1145/3519266
34.
Lai, V.; Chen, C.; Liao, Q.; … A.S.-R. preprint arXiv; 2021, undefined Towards a Science of Human-Ai Decision Making: A Survey of Empirical Studies. arxiv.org.
35.
Fleming, A., O'Grady, A.P., Stitzlein, C., Ogilvy, S., Mendham, D., Harrison, M.T.: Improving Acceptance of Natural Capital Accounting in Land Use Decision Making: Barriers and Opportunities. Ecol. Econ. 200 (2022). 10.1016/j.ecolecon.2022.107510
36.
Khan, A., Science, A.: undefined Emerging Data Sources in Decision Making and AI. Elsevier (2020)
37.
Bankins, S., Formosa, P., Griep, Y., Systems, D.R.-I.: undefined AI Decision Making with Dignity? Contrasting Workers' Justice Perceptions of Human and AI Decision Making in a Human Resource Management Context. Springer 2022, 1, 3, (2022). 10.1007/s10796-021-10223-8
38.
Van Den Bosch, K., Bronkhorst, A.: Human-AI Cooperation to Benefit Military Decision Making. (2018)
39.
Tools, G.: undefined AI Tools in Decision Making Support Systems: A Review. World Scientific 2012, XX, 1–13, (2012). 10.1142/S0218213012400052
40.
Angerschmid, A., Zhou, J., Theuermann, K., Chen, F., Holzinger, A.: Fairness and Explanation in AI-Informed Decision Making. mdpi.com 2022, 10.3390/make4020026
41.
Neudert, L., Knuutila, A., Google, P.H.-G.S.: (2020). undefined Global Attitudes towards AI, Machine Learning & Automated Decision Making. julietwaters.com 2020.
42.
Osasona, F., Oladipupo Amoo, O., Atadoga, A., Oluwaseun Abrahams, T., Ajoke Farayola, O.: Samson Ayinla, B. Reviewing the Ethical Implications of AI in Decision Making Processes. fepbl.com 2024, 6, 322–335. 10.51594/ijmer.v6i2.773
43.
Applications, M.: undefined Automated Machine Learning: AI-Driven Decision Making in Business Analytics. Elsevier (2023)
44.
Agarwal, A.; Singhal, C.; Company, R.T.-M.&; 2021, undefined AI-Powered Decision Making for the Bank of the Future. mckinsey.com.br.
45.
Wu, J., Sustainability, S.S.-.: undefined Managing Uncertainty in AI-Enabled Decision Making and Achieving Sustainability. mdpi.com, (2020). 10.3390/su12218758