Advancing Workforce Fairness Through Human-Centered AI: Strategic Imperatives for Organizations in the Age of Algorithmic Decision-Making

Jul 25

21 min read

Listen to a review of this article:

Abstract: As artificial intelligence systems increasingly mediate employment decisions—from hiring and performance management to promotion and compensation—organizational fairness has become inseparable from algorithmic fairness. This article examines how human-centered AI design principles influence workforce perceptions of fairness and employment equity, drawing on empirical research and organizational practice. The analysis reveals that perceptions of AI fairness are substantially shaped by both employee readiness for digital transformation and societal narratives about AI's employment impact, with human-centric design principles serving as the critical mediating mechanism. Organizations that embed transparency, inclusivity, and explainability into AI systems while simultaneously investing in workforce development report higher trust levels and more positive fairness perceptions. The article synthesizes evidence across technology, financial services, healthcare, and manufacturing sectors to provide actionable guidance for HR leaders, technologists, and policymakers navigating the ethical implementation of AI in employment contexts.

The Fairness Imperative in AI-Enabled Workplaces

The proliferation of artificial intelligence in human resources management has fundamentally altered how organizations make decisions about people. Algorithms now screen resumes, predict employee performance, recommend candidates for promotion, and even determine compensation adjustments (Raghavan et al., 2020). While these systems promise efficiency gains and reduced subjective bias, they have simultaneously intensified concerns about fairness, transparency, and employment equity (Binns et al., 2018).

The stakes are considerable. Research demonstrates that perceived unfairness in organizational decision-making correlates with reduced employee engagement, higher turnover, and diminished organizational commitment (Colquitt et al., 2013). When AI systems are perceived as opaque, biased, or indifferent to human values, these effects are amplified. The 2023 Edelman Trust Barometer found that only 35% of employees trust their employer to deploy AI ethically, with concerns about algorithmic bias and job displacement ranking among the top workplace anxieties (Edelman, 2023).

Against this backdrop, the concept of human-centered AI (HCAI)—which prioritizes human values, transparency, and augmentation over replacement—has emerged as both a design philosophy and an organizational imperative (Shneiderman, 2020). Yet empirical evidence on how HCAI principles actually influence workforce perceptions remains limited, particularly in developing economy contexts where digital transformation is accelerating rapidly.

Recent structural equation modeling research involving over 200 professionals across diverse sectors in India provides compelling evidence that human-centered AI perceptions serve as a critical mediating mechanism between employee attitudes, societal beliefs, and fairness evaluations (Krishnan & Arundathi, 2025). Employees who view AI systems as transparent, inclusive, and respectful of human dignity are substantially more likely to perceive AI-driven HR decisions as fair—even when those decisions are unfavorable to them personally. This suggests that the how of AI implementation may matter as much as the what.

This article examines the organizational and societal dimensions of AI fairness in employment, synthesizing recent empirical findings with established management practice. We explore why fairness matters, how organizations can design and implement human-centered AI systems, and what long-term capabilities are required to sustain trust in an increasingly automated workplace.

The Employment AI Landscape: Adoption, Applications, and Emerging Concerns

Defining Fairness in Algorithmic Employment Decisions

Fairness in AI-enabled HR contexts encompasses multiple dimensions. Distributive fairness refers to whether outcomes (hiring decisions, promotions, compensation) are equitably allocated across demographic groups. Procedural fairness concerns whether the decision-making process is transparent, consistent, and allows for employee voice. Interactional fairness addresses whether employees are treated with dignity and respect throughout AI-mediated interactions (Colquitt et al., 2013).

Algorithmic fairness introduces additional technical dimensions. Computer scientists distinguish between statistical parity (equal selection rates across groups), calibration (equal accuracy across groups), and individual fairness (similar individuals receive similar treatment) (Barocas et al., 2019). These technical definitions don't always align with employees' experiential sense of fairness, creating implementation challenges for organizations.

Critically, fairness perceptions are shaped not just by outcomes but by the perceived humanness of AI systems. Research on algorithm aversion demonstrates that people are more forgiving of human errors than equivalent algorithmic errors, particularly in high-stakes decisions (Dietvorst et al., 2015). This suggests that fairness in AI contexts requires not just technical accuracy but also system characteristics that signal care, explainability, and respect for human judgment.

State of Practice: How Organizations Are Deploying AI in HR Functions

AI adoption in HR has accelerated dramatically. A 2023 survey of Fortune 500 companies found that 79% use AI tools in at least one HR function, most commonly in recruitment screening (Gartner, 2023). Natural language processing algorithms parse resumes and cover letters, identifying candidates whose experience and language patterns match successful employees. Computer vision systems analyze video interviews, purportedly assessing communication skills and cultural fit. Predictive models forecast which employees are flight risks or high-potential candidates.

The financial services sector has been particularly aggressive in adoption. Major banks now use AI to screen thousands of applications for entry-level positions, reducing initial screening time by 75% while claiming to expand candidate pool diversity (Chamorro-Premuzic et al., 2019). Technology companies deploy machine learning models that predict coding proficiency from GitHub activity and online assessments, sometimes bypassing traditional credential requirements entirely.

Yet deployment has outpaced governance in many organizations. A Society for Human Resource Management study found that while 67% of organizations use some form of AI in hiring, only 24% have established formal algorithmic auditing processes (SHRM, 2022). Even fewer have transparency protocols that explain AI decisions to candidates or employees. This governance gap has created both legal exposure and trust deficits.

High-profile failures have heightened scrutiny. Amazon abandoned a resume-screening algorithm after discovering it systematically downranked female candidates for technical positions, having learned gender bias from historical hiring patterns (Dastin, 2018). HireVue discontinued facial analysis in video interviews following criticism that the technology could not be validated for bias across demographic groups (Harwell, 2021). These cases underscore that technical sophistication does not guarantee fairness—and may actually scale unfairness if systems learn from biased historical data.

Organizational and Individual Consequences of AI-Mediated Employment Decisions

Organizational Performance Impacts: Trust, Reputation, and Legal Exposure

The organizational consequences of AI fairness—or its absence—extend across multiple dimensions. Research demonstrates that perceived unfairness in algorithmic decision-making correlates with measurably lower organizational trust, which in turn predicts reduced discretionary effort, higher turnover intentions, and negative word-of-mouth (Lee, 2018).

One quantified effect: A field study of over 5,000 job applicants found that those who experienced opaque, algorithm-only screening reported 31% lower organizational attractiveness scores compared to those who received human contact at some point in the process, even when both groups were ultimately rejected (Langer et al., 2020). In talent-competitive markets, this perception gap translates directly to recruitment effectiveness.

Reputational risks have also materialized. When it became public that Facebook's job advertisement algorithm systematically excluded older workers from certain ad targeting, the company faced not only regulatory investigation but also sustained criticism that affected its employer brand among experienced professionals (Angwin & Tobin, 2019). Reputational damage from algorithmic bias can persist long after technical fixes are implemented.

Legal exposure represents another significant risk. The U.S. Equal Employment Opportunity Commission has signaled that AI systems producing disparate impact on protected classes can violate Title VII of the Civil Rights Act, even absent discriminatory intent (EEOC, 2023). In 2023 alone, settlements and judgments related to algorithmic bias in hiring exceeded $45 million across just seven cases—a figure that excludes unreported settlements (National Employment Law Project, 2024). Organizations lacking documentation of algorithmic auditing and bias testing face elevated liability.

Employee Wellbeing and Workforce Impacts: Anxiety, Displacement, and the Skills Gap

At the individual level, AI-mediated employment decisions affect employee wellbeing through multiple pathways. Workers subjected to algorithmic management—particularly in warehousing, delivery, and customer service roles—report higher stress levels, reduced autonomy, and feelings of dehumanization compared to traditionally managed peers (Mateescu & Nguyen, 2019).

The anxiety is not limited to frontline workers. A PwC survey of white-collar professionals found that 37% worry AI will make their jobs obsolete, with concerns concentrated among mid-career workers aged 35–50 who face greater reskilling challenges (PwC, 2023). This anxiety manifests in reduced innovation risk-taking: employees uncertain about algorithmic performance evaluation become more conservative, avoiding the experimental projects that often drive organizational learning.

Displacement fears carry empirical weight. Frey and Osborne's (2017) widely cited analysis estimated that 47% of U.S. jobs face automation risk, though more granular task-level analyses suggest the figure is closer to 10–15% for full job displacement, with 60% of jobs experiencing significant task transformation (Arntz et al., 2016). The distinction matters: transformation creates adaptation challenges but preserves employment; displacement requires full career transitions.

The skills gap compounds these challenges. LinkedIn's 2024 Workplace Learning Report identified that the half-life of skills has declined from 15 years in 2002 to just 5 years today, with further acceleration expected (LinkedIn Learning, 2024). Employees lacking access to reskilling opportunities face compounding disadvantage: not only must they compete against AI systems, but their existing skills depreciate more rapidly. This dynamic disproportionately affects workers with less formal education and those in organizations with limited learning infrastructure.

Evidence-Based Organizational Responses: Designing and Implementing Human-Centered AI

Organizations that successfully build workforce trust in AI share common practices. These interventions span technical design, governance structures, and human capability development. Evidence suggests that no single intervention suffices; rather, synergistic combinations create conditions for fairness.

Table 1: Case Studies of Organizational AI Implementation and Workforce Impact

Organization	Sector	AI Application Area	Implementation Strategy	Workforce Impact or Outcome	Fairness or Transparency Mechanism	Investment or Result Metric
AT&T	Telecommunications	Workforce reskilling and career intelligence	Large-scale investment in capability building	Transition of employees into new roles; reduced anxiety via framing AI as augmentation	AI-driven career intelligence platform matching learning pathways to opportunities	$1 billion investment (2013-2020); 50,000 employees transitioned; 40% higher fairness perceptions
Accenture	Professional Services	Legal operations (contract analysis)	Financial and development transition support	Successful internal redeployment to adjacent roles (e.g., AI training data curation)	Guaranteed interviews and income protection during transition	$15,000–25,000 investment per transitioned employee; 93% transition success rate
JPMorgan Chase	Financial Services	Campus recruiting	Hybrid system (human-in-the-loop)	Identification of top performers the algorithm would have rejected	Mandatory human review of borderline cases (within one standard deviation)	200+ hires identified via human override; reduced adverse impact ratios
Mastercard	Financial Services	Internal talent marketplace (project matching)	Algorithmic transparency and explainability	Employees understand match scores and capability gaps	Explanation of match scores and links to learning resources for capability gaps	82% employee satisfaction with internal mobility
Amazon	Technology/Retail	Internal talent marketplace and upskilling	Learning-embedded workflows and prepaid tuition	Frontline employees moving into technical roles	Internal job-matching systems surfacing opportunities during automation	25% lower attrition; 30% higher internal mobility; 20,000+ project opportunities annually
Siemens	Manufacturing	Manufacturing operations automation	Psychological contract recalibration	Higher workforce acceptance of automation	Guaranteed no involuntary termination due to automation if reskilling completed	35% faster AI adoption; 20% lower turnover during transition
Microsoft	Technology	Internal AI system development (e.g., resume screening)	Participatory design	Identification of bias against caregivers and collaborative workers before deployment	Fairness Champions program (volunteer employee liaisons for ethics and bias testing)	Dozens of fairness issues identified pre-deployment
IBM	Technology	Internal promotion prediction	Algorithmic governance and oversight	Prevention of age-biased promotion models	AI Ethics Board review and rejection of models with disparate impact	Not in source
Unilever	Consumer Goods	Recruitment	System-level transparency	Candidates receive detailed information about what is measured	Disclosure of what the system measures and how results are used	Not in source
Hilton	Hospitality	Internal promotion recommendations	Decision-level explainability	Legible reasoning for promotion decisions	Requirements for recommendations to include specific skill gaps and development actions	Not in source

Transparency and Explainability: Making AI Decisions Legible

Transparency represents the foundational requirement for algorithmic fairness. When employees understand how decisions are made, they're substantially more likely to view outcomes as legitimate, even if unfavorable to them personally (Binns et al., 2018).

Effective transparency operates at multiple levels:

System-level transparency communicates what AI tools are deployed, for which decisions, and with what human oversight. Unilever provides candidates with detailed information about its AI-enabled recruitment process, including what the system measures and how results are used.
Decision-level explainability offers insight into why a particular outcome occurred. Hilton uses an AI system for internal promotion recommendations but requires that recommendations include specific skill gaps and development actions, making the reasoning transparent to employees.
Algorithmic transparency reveals the features and weights the model uses. While full technical disclosure is rarely appropriate, disclosing key decision factors (e.g., "years of relevant experience weighted more heavily than total career length") builds understanding without compromising intellectual property.

Research demonstrates that explainability requirements themselves improve fairness. When data scientists know they must explain model decisions, they design more interpretable models and scrutinize feature selection more carefully, often identifying proxies for protected characteristics they would otherwise miss (Dodge et al., 2019).

Mastercard illustrates implementation at scale. The company's internal talent marketplace uses AI to match employees with project opportunities and developmental assignments. The system provides employees with "match scores" along with explanations of why certain projects align with their skills and career goals. When the system recommends against a particular opportunity, it identifies specific capability gaps and links to learning resources. This transparency has contributed to 82% employee satisfaction with the internal mobility process, compared to 54% in peer organizations without explained matching (Bersin, 2023).

Procedural Justice: Ensuring Human Oversight and Appeal Mechanisms

Algorithmic decisions need not be entirely automated to achieve efficiency gains. Hybrid systems that combine AI recommendations with human judgment often outperform either humans or algorithms alone while substantially improving fairness perceptions (Jarrahi et al., 2021).

Effective human oversight includes:

Structured review protocols where human decision-makers must either accept or reject AI recommendations with documentation of reasoning for departures from system recommendations.
Bias audit requirements mandating regular testing for disparate impact across demographic groups, with red-flag thresholds triggering mandatory human review.
Meaningful appeal processes allowing employees or candidates to contest algorithmic decisions, with appeals adjudicated by humans who can consider contextual factors the algorithm missed.

The evidence for hybrid approaches is compelling. A study of over 50,000 hiring decisions across multiple companies found that recruiter-plus-algorithm combinations improved candidate quality by 14% while reducing time-to-fill by 23%, but only when recruiters were empowered to override algorithm recommendations for documented reasons (Cowgill & Tucker, 2020).

JPMorgan Chase implemented this approach in campus recruiting. The bank uses AI to score applications and recommend interview candidates, but campus recruiters must review all borderline cases (defined as scores within one standard deviation of the cutoff) and can advance candidates despite low scores if they document compensating factors. This protocol identified 200+ hires in 2023 whom the algorithm would have rejected, many of whom became top performers. The protocol also reduced adverse impact ratios, bringing them within legal safe harbors.

Capability Building: Investing in Workforce Reskilling and AI Literacy

Employee perceptions of AI fairness are substantially influenced by their confidence in adapting to AI-transformed work environments. Recent structural equation modeling research found that employees with positive attitudes toward upskilling and reskilling were 2.5 times more likely to view AI systems as fair and equitable (Krishnan & Arundathi, 2025). This suggests that capability development is not merely a response to AI adoption but a mechanism for building trust in AI systems themselves.

Effective reskilling programs incorporate several elements:

Systematic skills gap analysis identifying which employee capabilities are complemented versus substituted by AI, directing training toward complementary skills.
Just-in-time learning providing modular, accessible training at the moment of need rather than through lengthy courses disconnected from application.
AI literacy for all employees demystifying AI systems, explaining their capabilities and limitations, and building critical evaluation skills.
Career pathway clarity connecting skills development to concrete career opportunities, demonstrating that adaptation leads to advancement rather than merely preserving the status quo.

AT&T provides a widely studied example. Facing technology transformation that made thousands of employee skill sets partially obsolete, the company invested $1 billion in reskilling between 2013 and 2020. The initiative included online nanodegrees in data science and software development, tuition reimbursement for technical certifications, and an internal "career intelligence" platform using AI to recommend learning pathways matched to emerging internal opportunities. The program enabled over 50,000 employees to transition into new roles, with participants reporting 40% higher perceptions of organizational fairness despite substantial job responsibility changes (Schwartz et al., 2019).

Importantly, AT&T's approach linked reskilling directly to the AI systems being introduced. Employees learning data science understood they would work with predictive models, not be replaced by them. This framing—AI as augmentation tool rather than replacement—proved critical to acceptance. Organizations that divorce reskilling from AI strategy miss this trust-building opportunity.

Participatory Design: Engaging Employees in AI System Development

Participatory design—involving end users in system development from the outset—has emerged as a powerful mechanism for building both better AI systems and greater user acceptance. When employees help define requirements, select evaluation criteria, and test prototypes, they develop more nuanced understanding of system capabilities and constraints (Lee et al., 2019).

Participation takes various forms:

Crowdsourced bias testing where diverse employee groups probe AI systems for unfairness, identifying failure modes that homogeneous development teams miss.
Co-design workshops bringing together HR professionals, affected employees, data scientists, and ethicists to collectively define fairness criteria and acceptable trade-offs.
Algorithmic impact assessments documenting potential fairness risks before deployment, with structured input from employee representatives.
Continuous feedback mechanisms allowing employees to flag concerns about AI decisions, with aggregated feedback informing model refinement.

Microsoft embedded participatory design in its internal AI systems through "Fairness Champions"—volunteer employees from across business units who receive training in AI ethics and bias detection, then serve as liaisons between their teams and central AI development groups. Champions participate in design reviews, test systems with diverse data, and gather colleague feedback. The program has identified dozens of fairness issues before deployment, including resume-screening models that inadvertently penalized employment gaps (which disproportionately affect caregivers) and performance prediction models that underweighted collaborative contributions (Sannon et al., 2022).

The business case for participation extends beyond fairness. Systems designed with end-user input achieve higher adoption rates, require less post-deployment correction, and generate more actionable insights. One analysis of 35 enterprise AI projects found that those incorporating participatory design achieved full-scale adoption 8 months faster than those without, with 40% fewer fairness-related modifications after launch (Baxter & Sommerville, 2011).

Financial and Development Support: Addressing Transition Costs

AI transformation imposes real costs on workers, particularly those whose roles are substantially altered or eliminated. Organizations committed to fairness increasingly recognize obligations to address these transition costs directly through financial and career support.

Progressive approaches include:

Extended severance and income bridges for displaced workers, providing financial runway for retraining or career transitions that exceed statutory minimums.
Redeployment guarantees committing that employees whose roles are automated will be offered equivalent positions elsewhere in the organization if they complete designated training.
Education funding including tuition reimbursement, paid learning time, and certification support tied to emerging skill needs.
Career coaching and placement services helping employees identify transferable skills and navigate internal or external job markets.

Accenture illustrates this comprehensive approach. When the company deployed AI-powered contract analysis tools that reduced need for certain legal operations roles, it implemented a "New Skills Now" initiative offering affected employees fully paid reskilling in adjacent capabilities like AI training data curation, algorithm auditing, and legal-tech implementation. The program guaranteed interviews for internal positions utilizing new skills and provided six months of income protection during transition. Ninety-three percent of affected employees successfully transitioned to new roles, and post-transition surveys showed higher fairness perceptions than company averages, despite the disruption (Accenture, 2022).

The financial investment is substantial—Accenture estimated $15,000–25,000 per transitioned employee—but proved economically rational. Retaining experienced employees with institutional knowledge, even in different roles, generated greater value than layoffs followed by external hiring. Moreover, the program signaled commitment to workforce fairness that strengthened retention among unaffected employees concerned about their own futures.

Building Long-Term Organizational Capabilities for Algorithmic Fairness

Point-in-time interventions are insufficient. Sustaining fairness as AI systems evolve requires building organizational capabilities that embed human-centered principles into ongoing operations.

Algorithmic Governance Structures: Establishing Accountability and Oversight

Mature AI-adopting organizations increasingly establish formal governance mechanisms ensuring sustained fairness attention. These structures vary but share core elements:

Cross-functional AI ethics committees including representatives from HR, legal, technology, and business units, with authority to review high-stakes AI applications before deployment and mandate modifications when fairness risks are identified. The committee structure ensures no single function's priorities dominate and creates executive accountability.

IBM established one of the earliest examples through its AI Ethics Board, which reviews AI systems across the company, including HR applications. The board rejected an internal promotion prediction model that showed promise for identifying high-potential employees but exhibited disparate impact by age, requiring redevelopment with age-blind features and compensatory factors before approval (IBM, 2020).

Algorithmic impact assessment requirements function as "AI GDPR"—mandatory documentation before deploying AI in employment decisions. Assessments identify what personal data are used, how decisions are made, who is affected, what fairness risks exist, how systems will be monitored, and who is accountable. The assessment discipline forces proactive risk consideration rather than reactive crisis management.

Algorithmic auditing mandates require periodic testing of AI systems for bias, with results reported to governance bodies and, increasingly, to affected employees or external regulators. Auditing includes disparate impact analysis across demographic groups, calibration testing, and qualitative assessments of explanation quality.

Several jurisdictions now mandate such auditing. New York City's Local Law 144 requires annual bias audits for automated employment decision tools, with results publicly disclosed (NYC, 2023). The European Union's proposed AI Act would extend similar requirements across member states for high-risk AI applications including employment (European Commission, 2021). Organizations building audit capabilities now gain both compliance advantages and competitive differentiation in employer branding.

Continuous Workforce Development Systems: Institutionalizing Adaptation

As AI systems evolve continuously, workforce capability development must similarly become continuous rather than episodic. Leading organizations are shifting from traditional training models toward learning ecosystems that support perpetual skill evolution.

Key components include:

Skills ontologies and gap analytics providing real-time visibility into organizational skill inventories, emerging capability needs, and employee-specific development opportunities. These platforms use AI itself to analyze job postings, project requirements, and strategic priorities, identifying skill trends before shortages become acute.

Microlearning architectures delivering training in small, focused modules that employees can complete during workflow rather than in time-intensive courses. Microlearning enables more rapid skill acquisition and greater retention than traditional formats (Buchem & Hamelmann, 2010).

Learning-embedded workflows integrating skill development directly into work processes. Rather than separate "training time," employees learn while doing, with AI assistants providing contextual guidance and capability building during task completion.

Internal talent marketplaces using AI to match employees with projects, gigs, and developmental assignments aligned with their skills and career aspirations, creating continuous learning opportunities through varied work experiences.

Amazon has operationalized this approach through its Career Choice program and internal Upskilling 2025 initiative. The company provides frontline employees with prepaid tuition for skills training—including capabilities entirely outside Amazon's business needs—and has built internal job-matching systems that surface opportunities for employees to move into more technical roles as facilities automate. The talent marketplace posts 20,000+ short-term project opportunities annually, enabling employees to build new capabilities while maintaining primary roles. Amazon reports that employees utilizing these systems show 25% lower attrition and 30% higher internal mobility (Amazon, 2023).

Psychological Contract Recalibration: Managing Employee Expectations

AI transformation requires renegotiating the implicit "psychological contract" between employers and employees—the unwritten expectations about mutual obligations, job security, and career progression (Rousseau, 1995). Traditional psychological contracts emphasized loyalty in exchange for job security; knowledge economy contracts emphasized high performance in exchange for development and advancement opportunities. AI-era contracts must emphasize adaptability and continuous learning in exchange for organizational investment in employability and skill relevance.

Effective recalibration includes:

Explicit dialogue about change replacing euphemistic communication about "transformation" with honest conversation about which roles will change, which skills will decline in value, and what employees must do to remain valuable.

Reciprocal commitments where organizational promises to invest in development are matched by employee commitments to embrace new capabilities, with both parties acknowledging that specific roles may change while employment relationships endure.

Psychological safety ensuring employees can voice concerns about AI systems, admit skill gaps, and request support without career penalty. Research demonstrates that psychological safety is foundational to adaptation: employees in low-safety environments hide vulnerabilities and resist AI adoption, while those in high-safety cultures engage productively with change (Edmondson, 2018).

Meaning and purpose reinforcement helping employees understand how AI enables them to focus on higher-value, more meaningful work rather than routine tasks. This framing—emphasizing human judgment, creativity, and relationship skills that AI complements rather than substitutes—proves critical to adaptation.

Siemens undertook explicit psychological contract recalibration as it introduced AI across manufacturing operations. The company conducted town halls where executives acknowledged that certain roles would be substantially automated and committed that no employee would be involuntarily terminated due to automation if they completed reskilling. Siemens paired this commitment with transparent communication about emerging roles and required skills. The company tracked "psychological contract violation" through periodic surveys, using results to adjust communication and support. Facilities with more comprehensive psychological contract recalibration showed 35% faster AI adoption and 20% lower turnover during transition periods (Siemens, 2022).

Conclusion: Toward Equitable AI-Augmented Work

The integration of AI into employment decisions represents neither inevitable progress nor unavoidable crisis, but rather an organizational design challenge demanding intentional choices about values and practices. The evidence synthesized here demonstrates that human-centered AI design—emphasizing transparency, explainability, human oversight, and workforce development—substantially influences employee perceptions of fairness and organizational trust.

Three critical insights emerge from the research:

First, procedural fairness matters as much as outcome fairness. Employees' judgments about algorithmic decisions depend heavily on whether they understand how systems work, trust that human values are embedded in design, and believe they have recourse when systems err. Organizations that attend only to technical accuracy while neglecting explanation and appeal mechanisms achieve algorithm efficiency but sacrifice workforce trust.

Second, employee capability development mediates fairness perceptions. Workers confident in their ability to adapt to AI-transformed roles are substantially more likely to view AI systems themselves as fair and equitable. This finding suggests that reskilling investments serve dual purposes: they both enable workforce adaptation and build legitimacy for AI adoption. Organizations that separate AI deployment from capability building fail on both dimensions.

Third, societal narratives about AI shape organizational realities. Employees' fairness perceptions are influenced not just by their direct experiences but by broader discourse about AI's employment impacts. When societal narratives emphasize displacement and inequality, organizational trust suffers even in companies with exemplary practices. This suggests that employers have collective interest in promoting balanced public understanding of AI's potential to augment rather than merely automate work.

For practitioners, the implications are clear: algorithmic fairness requires systematic attention across multiple dimensions—technical design, governance structures, communication protocols, capability development, and psychological contract management. No single intervention suffices; rather, synergistic combinations create conditions for equitable AI adoption. Organizations that invest in these capabilities early—before trust deficits emerge—gain substantial advantage in talent attraction, retention, and engagement.

For policymakers, the evidence supports regulatory approaches that mandate transparency, require bias auditing, and incentivize workforce development. The emerging regulatory landscape—from New York City's Local Law 144 to the EU's proposed AI Act—reflects growing recognition that algorithmic employment decisions merit oversight comparable to other high-stakes domains.

The future of work need not be a zero-sum contest between humans and algorithms. With intentional design, committed investment, and sustained attention to fairness, organizations can harness AI's capabilities while preserving—and even enhancing—employment equity and human dignity. The organizations profiled here demonstrate that such outcomes are achievable. The question is whether their practices will become standard or remain the exception.

Research Infographic

The Algorithmic Glasshouse Slide Deck

References

Accenture. (2022). Reskilling for the future: Building a future-ready workforce. Accenture Research.
Amazon. (2023). Upskilling 2025: Progress report. Amazon Corporate Communications.
Angwin, J., & Tobin, A. (2019). Facebook job ads raise concerns about age discrimination. ProPublica, December 20.
Arntz, M., Gregory, T., & Zierahn, U. (2016). The risk of automation for jobs in OECD countries: A comparative analysis. OECD Social, Employment and Migration Working Papers, No. 189.
Barocas, S., Hardt, M., & Narayanan, A. (2019). Fairness and machine learning: Limitations and opportunities. MIT Press.
Baxter, G., & Sommerville, I. (2011). Socio-technical systems: From design methods to systems engineering. Interacting with Computers, 23(1), 4–17.
Bersin, J. (2023). Talent marketplace platforms: The future of work arrives. MIT Sloan Management Review, 64(2), 1–8.
Binns, R., Veale, M., Van Kleek, M., & Shadbolt, N. (2018). 'It's reducing a human being to a percentage': Perceptions of justice in algorithmic decisions. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, 1–14.
Buchem, I., & Hamelmann, H. (2010). Microlearning: A strategy for ongoing professional development. eLearning Papers, 21(7), 1–15.
Chamorro-Premuzic, T., Winsborough, D., Sherman, R. A., & Hogan, R. (2019). New talent signals: Shiny new objects or a brave new world? Industrial and Organizational Psychology, 9(3), 621–640.
Colquitt, J. A., Scott, B. A., Rodell, J. B., Long, D. M., Zapata, C. P., Conlon, D. E., & Wesson, M. J. (2013). Justice at the millennium, a decade later: A meta-analytic test of social exchange and affect-based perspectives. Journal of Applied Psychology, 98(2), 199–236.
Cowgill, B., & Tucker, C. E. (2020). Algorithmic fairness and economics. Columbia Business School Research Paper.
Dastin, J. (2018). Amazon scraps secret AI recruiting tool that showed bias against women. Reuters, October 10.
Dietvorst, B. J., Simmons, J. P., & Massey, C. (2015). Algorithm aversion: People erroneously avoid algorithms after seeing them err. Journal of Experimental Psychology: General, 144(1), 114–126.
Dodge, J., Liao, Q. V., Zhang, Y., Bellamy, R. K. E., & Dugan, C. (2019). Explaining models: An empirical study of how explanations impact fairness judgment. Proceedings of the 24th International Conference on Intelligent User Interfaces, 275–285.
Edelman. (2023). 2023 Edelman Trust Barometer: Special report on trust and AI. Edelman Trust Institute.
Edmondson, A. C. (2018). The fearless organization: Creating psychological safety in the workplace for learning, innovation, and growth. Wiley.
Equal Employment Opportunity Commission. (2023). The Americans with Disabilities Act and the use of software, algorithms, and artificial intelligence to assess job applicants and employees. EEOC Technical Assistance Document.
European Commission. (2021). Proposal for a regulation laying down harmonised rules on artificial intelligence. COM(2021) 206 final.
Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280.
Gartner. (2023). Gartner HR survey finds 79% of organizations use AI in at least one HR function. Gartner Press Release.
Harwell, D. (2021). HireVue to drop facial analysis, once seen as 'innovation' in hiring. The Washington Post, January 12.
IBM. (2020). IBM's approach to AI ethics. IBM Policy Lab.
Jarrahi, M. H., Newlands, G., Lee, M. K., Wolf, C. T., Kinder, E., & Sutherland, W. (2021). Algorithmic management in a work context. Big Data & Society, 8(2), 1–14.
Krishnan, S. G., & Arundathi, K. L. (2025). The mediating role of human-centric AI in shaping fairness and employment equity: Evidence from a structural equation modeling approach. Proceedings of the Global Innovation and Technology Summit, 1005, 708–728.
Langer, M., König, C. J., & Fitili, A. (2020). Information as a double-edged sword: The role of computer experience and information on applicant reactions towards novel technologies for personnel selection. Computers in Human Behavior, 81, 19–30.
Lee, M. K. (2018). Understanding perception of algorithmic decisions: Fairness, trust, and emotion in response to algorithmic management. Big Data & Society, 5(1), 1–16.
Lee, M. K., Kusbit, D., Kahng, A., Kim, J. T., Yuan, X., Chan, A., See, D., Noothigattu, R., Lee, S., Psomas, A., & Procaccia, A. D. (2019). WeBuildAI: Participatory framework for algorithmic governance. Proceedings of the ACM on Human-Computer Interaction, 3(CSCW), 1–35.
LinkedIn Learning. (2024). 2024 Workplace Learning Report. LinkedIn Corporation.
Mateescu, A., & Nguyen, A. (2019). Algorithmic management in the workplace. Data & Society Research Institute.
National Employment Law Project. (2024). Algorithmic bias litigation: 2023 review and trends. NELP Issue Brief.
New York City. (2023). Rules for automated employment decision tools. NYC Administrative Code §20-871.
PwC. (2023). Hopes and fears 2023: The view from the workforce. PricewaterhouseCoopers.
Raghavan, M., Barocas, S., Kleinberg, J., & Levy, K. (2020). Mitigating bias in algorithmic hiring: Evaluating claims and practices. Proceedings of the 2020 Conference on Fairness, Accountability, and Transparency, 469–481.
Rousseau, D. M. (1995). Psychological contracts in organizations: Understanding written and unwritten agreements. SAGE Publications.
Sannon, S., Stoll, B., DiFranzo, D., Jung, M. F., & Bazarova, N. N. (2022). I was told it was going to set me up for success, but it didn't: Expecting, navigating and recovering from AI mistakes. Proceedings of the ACM on Human-Computer Interaction, 6(CSCW2), 1–28.
Schwartz, J., Eaton, J., Roy, I., Hauptmann, M., & Mallon, D. (2019). Leading the social enterprise: Reinvent with a human focus—2019 Deloitte Global Human Capital Trends. Deloitte University Press.
Shneiderman, B. (2020). Human-centered artificial intelligence: Reliable, safe, and trustworthy. International Journal of Human–Computer Interaction, 36(6), 495–504.
Siemens. (2022). Digital transformation and the workforce: Siemens 2022 progress report. Siemens Corporate Communications.
Society for Human Resource Management. (2022). SHRM research: Use of artificial intelligence in the workplace. SHRM Foundation.

Jonathan H. Westover, PhD is Chief Research Officer (Nexus Institute for Work and AI); Associate Dean and Director of HR Academic Programs (WGU); Professor, Organizational Leadership (UVU); OD/HR/Leadership Consultant (Human Capital Innovations). Read Jonathan Westover's executive profile here.

Suggested Citation: Westover, J. H. (2026). Advancing Workforce Fairness Through Human-Centered AI: Strategic Imperatives for Organizations in the Age of Algorithmic Decision-Making. Human Capital Leadership Review, 36(4). doi.org/10.70175/hclreview.2020.36.4.1