Page 42 - CMA Journal (Jul-Aug 2026)

F oc us S ec t ion

These records may contain previous managerial Ethical purchasing also involves procedural justice, The Management Accountant’s Role AI algorithms are neither inherently strategically wise,
heuristics, informal rules of thumb, subjective supplier which requires suppliers to be assessed according to nor do they have moral agency: they are mathematical
scores, legacy buyer preferences and institutional transparent, consistent and contestable standards. The vulnerabilities discussed represent a tangible mirrors re ecting historical data, assumptions, and
approaches to risk (Clarkson, 2026). Data science-powered procurement algorithms may danger to the natural selection in sourcing and implies managerial biases embedded within them. If not
challenge this principle if suppliers have limited the need to extend the boundaries of the actions that properly addressed, automated procurement systems
For example, if previous procurement managers a chartered management accountant should be able
preferred local suppliers but were consistently opportunities to understand, question or challenge to execute (Mhlongo, 2025). Finance executives can no also have the potential to perpetuate historical biases,
reluctant to engage with smaller vendors, historical the basis of an automated decision (Pal, 2026). When longer be passive audiences of software-generated exclude diverse supplier communities, violate vendors’
suppliers are excluded based on inaccurate or biased
Introduction The validation of procurement records may establish an apparent information, they may have little opportunity to reports, or themselves be in a position of treating AI economic rights, and expose organizations to legal
an alarming prospect. Hence
and ESG risk,
subjective historical data statistical relationship between supplier characteristics explain the circumstances or correct the underlying sourcing models as if they are frozen, static tools.
One of the most important tasks in large-scale through mathematical and lower perceived delivery risk. Similarly, if previous Instead, management accountants need to make management accountants also need to evolve their • Rainy, T.A. and Chowdhury, A.R., 2022. The role of arti cial intelligence in
organizations that rely heavily on stock and inventory processes may initially buyers consistently undervalued smaller innovative data. themselves the active regulators of the enterprise’s professional responsibilities. Finance leaders should vendor performance evaluation within digital retail supply chains: A
is identifying suitable suppliers and vendors (Sarkar, appear objective and suppliers or penalized alternative vendors for reasons Although the bias may originate from human algorithmic and decision-making structure. This not only consider the immediate line-item cost review of strategic decision-making models. American Journal of Scholarly
Research and Innovation, 1(01), pp.220-248.
Solanki and Gaikwad, 2025). Supplier assessment has harmless, but it has unrelated to their actual performance, these patterns decisions, the use of AI can make the resulting decision reconceptualization involves the modi cation of reductions, but also the second-order system e ects of • Pal, K., 2026. Ethical and Regulatory Bias in Arti cial Intelligence-Based
traditionally been time-consuming and signi cant ethical may subsequently be interpreted by AI systems as appear neutral. This creates a risk that suppliers may be traditional management accounting control schemas – although routine, low value indirect purchases (e.g. such automated choices. Through robust algorithmic Applications for Supply Chain Activities. In Navigating Accountability and
resource-intensive, while international sourcing implications for valid historical indicators. disadvantaged without an adequate mechanism to such as COSO Internal Control and Enterprise Risk regular o ce supplies or small-scale maintenance governance, data integrity audit enforcement, Human Rights in AI-Powered Supply Chains (pp. 133-164). IGI Global
Scienti c Publishing.
thresholds,
and
further increases costs and associated risks. AI-enabled procurement decision- As a result, an AI model used for supplier selection interrogate, dispute or amend the data and Management (ERM) to address control in an products) can also be procured through automation human-in-the-loop control modeling design, • Mhlongo, A.F., 2025. The impact of arti cial intelligence on the role of
end-to-end
multi-criteria
procurement is increasingly being viewed as a means making. A major does not necessarily remove human bias. Instead, it assumptions underlying an automated AI-powered strategic sourcing. with no risk involved, the contracts for high-value, management accountants can guide AI to be an management accountants. In Arti cial Intelligence in Accounting, Auditing
of addressing these challenges by reducing the time objective of adopting AI may encode, automate and scale that bias under the recommendation. Audits for Data Integrity and strategically sensitive or critical elements need to have e ective lever of both organization-level e ciency and and Finance: A Guide for Implementation and Use (pp. 63-82). Cham:
Springer Nature Switzerland.
and resources required for procurement activities. procurement tools is to appearance of mathematical objectivity. The For management accountants, these ethical a Human-in-the-Loop (HITL) design. HITL systems ethical stewardship (Salerno, 2026). • Coles, A.B., 2025. Optimizing the AI Data Supply Chain: A Framework for
process large volumes of impersonal nature of an algorithm can therefore Provenance Strategic Sourcing and Risk Management. Chicago Booth Accounting
Like other technology-driven solutions, AI-enabled relevant data within a considerations create important nancial and strategic enable the AI to function solely as a decision-support References Research Center Research Paper No. Forthcoming.
procurement promises cost savings and improved limited timeframe reinforce historical assumptions and decision-making trade-o s. Contemporary management accounting Prior to training a machine learning model or putting intelligence product, highlighting data, agging risks, • Shahbazbeigi, A., 2024. Application of Arti cial Intelligence and Machine
e ciency. It can streamline vendor selection, (Akhtar, 2025). From a nancial perspective, this can be patterns (Theodorakopoulos et al., 2025). frameworks, including the principles associated with that model into ERP procurement modules, the and ranking options, while human procurement • Sarkar, S., Solanki, S.K. and Gaikwad, M.H., 2025. E ective Vendor Learning in Accounting and Auditing. Transactions on Data Analysis in
Management. Risk Management, 5(3).
Social Science, 6(2), pp.92-108.
automate routine tasks and enable procurement highly attractive because it may reduce procurement This creates an important dilemma for organizations integrated reporting and global management management accounting process needs to conduct deliberative bodies continue have nal decision • Shaik, A.S., 2026. AI Procurement as Competitive Advantage: Build, Buy, • Kumar, S., Datta, S., Singh, V., Datta, D., Singh, S.K. and Sharma, R., 2024.
specialists to devote greater attention to strategic costs while saving specialists considerable time and adopting AI for strategic decision-making: automated accounting, recognize the importance of protecting rigorous data quality and provenance audits. This making power and operational responsibility. Partner, and the New Make-or-Buy Decision. Buy, Partner, and the New Applications, challenges, and future directions of human-in-the-loop
organizational objectives. The economies of scale resources. historical pattern recognition should not be confused and enhancing value across multiple forms of capital, requires mining historical purchase order databases, Make-or-Buy Decision (April 01, 2026). learning. IEEe Access, 12, pp.75735-75760.
o ered by AI-enabled procurement can be substantial with genuine strategic reasoning. AI can identify including social, relationship and manufactured supplier performance scores, and contract award Multi-Criteria Utility Function Design • Ho, A., 2025. Arti cial intelligence and its application in procurement. • Abdulla, A., 2022. Combining Arti cial Intelligence and Multi Criteria
Decision Making Approaches for Supplier Selection (Doctoral dissertation,
• Lo Ribeiro, J., Plangger, K. and Montecchi, M., 2026. Decoding Responsible
compared with traditional procurement approaches AI can also rapidly identify suppliers based on correlations in historical data, but human professionals capital. records for historical biases, missing data points, or Traditional procurement models are often constrained Procurement: Conceptualizing Bias Co evolution in AI aided University of Hudders eld United Kingdom).
(Shaik, 2026). However, beyond the e ciency gains prede ned criteria and provide recommendations remain responsible for determining whether those A procurement strategy focused solely on short-term region-based procurement anomalies (Coles, 2025). only by the desire to minimize cost. Management Organizational Decision making. British Journal of Management, 37(1), • Ayibam, J.N., 2025. Arti cial intelligence in public procurement: Legal
Cleaning and rebalancing training data can ensure that
p.e70027.
associated with technological implementation, several regarding supplier reliability, nancial strength and correlations are relevant, ethical and appropriate for cost reduction through aggressive automation may the algorithm does not carry forward awed historical Accountants’ need to work with cross functional teams • Akhtar, L., 2025. Procurement 4.0: Investigating AI adoption for intelligent frameworks, ethical challenges, and policy solutions for transparent and
e cient governance. Alkebulan: A Journal of West and East African Studies,
important considerations require the attention of value. However, the speed and apparent objectivity of future decisions. therefore undermine long-term organizational value. biases. to translate these mathematical loss functions into and resilient supply chains. 5(2), pp.54-69.
management accountants. these recommendations should not eliminate the Large-scale global procurement strategies that use Excessive dependence on algorithmically selected multi-criteria decision-making frameworks. This might • ARTS, O.M.O.L., 2025. SHADOWS, SIMULATIONS, AND HALLUCINATIONS • Salerno, B., 2026. AI Architecture for Intelligent Supply Chains (Doctoral
need for professional judgement and independent algorithmic supplier screening may also create mega-suppliers can reduce supply-chain diversity and Algorithmic Auditing and Model mean integrating weighted parameters around vendor (Doctoral dissertation, STANFORD UNIVERSITY). dissertation, Politecnico di Torino).
The Main Issue With AI Procurement validation. challenges for small and medium enterprises (SMEs), increase concentration risk. It may also expose ESG performance, supplier diversity, carbon intensity, • Clarkson, M., 2026. Arti cial Intelligence And Machine Learning In
Decision Making Ethics and AI procurement minority-owned business enterprises (MBEs), organizations to legal and reputational risks as Validation and procedural justice within the utility computation Procurement Decision-Making. Journal of Procurement & Supply Chain,
10(1), pp.1-13.
AI used in procurement relies heavily on historical AI systems are not immune to technical limitations. women-owned businesses and suppliers from ESG-related reporting and due-diligence requirements In the same way that physical inventory counts and of an algorithm (Abdulla, 2022). This can allow the AI to • Theodorakopoulos, L., Theodoropoulou, A. and Halkiopoulos, C., 2025.
Cognitive bias mitigation in executive decision-making: A data-driven
nancial internal control testing are performed by
developing countries (Muldoon et al., 2025). AI models
organisational data to identify relationships between Users with experience of AI models are familiar with are generally designed to optimize organizational continue to develop. management accountants, algorithmic audits must be seek to maximize whole-enterprise corporate value as approach integrating big data analytics, AI, and explainable systems.
Electronics, 14(19), p.3930.
variables and generate recommendations (Ho, 2025). issues such as hallucinations, inaccurate outputs and objectives based on available data. In such conducted periodically. The reviews analyze model opposed to narrow, short-term reductions in unit • Muldoon, J., Cant, C., Graham, M. and Ustek Spilda, F., 2025. The poverty of
However, the historical data used by AI systems may the misidenti cation of patterns (Arts, 2025). In circumstances, smaller suppliers with limited historical settings to ensure that the software recommendation costs. ethical AI: impact sourcing and AI supply chains. AI & society, 40(2),
itself contain subjective human decisions and biases. procurement, an algorithm may misinterpret changing records or complex digital compliance pro les may be engine continues to be in line with the strategic Conclusion pp.529-543.
Procurement specialists have traditionally made customs information, incorrectly group unrelated disadvantaged during automated supplier evaluation business priorities, ethical principles, and risk tolerance
decisions based on their experience, judgement and corporate entities or generate inaccurate nancial (Rainy and Chowdhury, 2022). levels of the rm (Shahbazbeigi, 2024). Sensitivity The application of enterprise-wide Arti cial
the information available to them at the time. viability scores for a nancially distressed supplier. If analyses should be performed to determine how the Intelligence to business procurement is a paradigm
Consequently, historical procurement data may re ect nance teams treat AI-generated outputs as Emerging suppliers may o er signi cant local algorithm ranks suppliers if the assessment parameters shift in enterprise management. As shown by
preferences, assumptions and decision-making authoritative without appropriate validation, economic bene ts, innovative products or change, for example, if the business size is larger or enterprise leaders such as Siemens, AI solutions
patterns that are not necessarily objective. organizations may assume operational, credit and lower-carbon solutions but may nevertheless be smaller, if the geographic origin is di erent, or if the provide tangible operational bene ts, signi cantly
excluded during pre-quali cation because they do not
relieving the burden of manual research work,
This creates an important challenge. If historical data legal risks that have not been adequately assessed. have the volume or depth of historical data required by length of operating history is longer or shorter. shortening cycle times, and o ering strategic sourcing
contains bias, machine-learning systems trained on A closer examination of AI-based procurement also an algorithm. Over time, such exclusions may become Implementing Human-in-the-Loop insight in the face of global supply chain disruptions
that data may reproduce and reinforce those biases. highlights the in uence of historical data. AI models embedded in the system's evaluation history and (HITL) Controls (Ayibam, 2025). The economic and tactical cases for
The resulting decisions may therefore appear objective built using enterprise systems such as ERP platforms create structural barriers to market entry for SMEs and enabling AI in enterprise procurement are compelling.
because they are generated mathematically, even do not operate independently of the organisational other underrepresented suppliers. This may con ict The AI structures need to be developed with However, e ciency should not be achieved through
though the underlying data re ects earlier human environment. Instead, they analyze historical with organizations’ corporate social responsibility well-de ned policy thresholds to distinguish when compromising moral soundness, systemic fairness, and
judgements (Lo Ribeiro et al., 2026). enterprise data that re ects years of human decisions. (CSR) commitments. turning over control to human strategic long-term governance.
decision-making (Kumar et al., 2024). They explain that

40 ICMA’s Chartered Management Accountant, Jul-Aug 2026