Featuring scholarly and practitioner insights from HR and people leaders, industry experts, and researchers.

In this HCI Webinar, I talk with Tom Ulbrich about the intersection between for-profit and non-profit organizations and leadership. Tom Ulbrich is an entrepreneur, educator, speaker, author, social sector CEO at Goodwill of Western New York, a member of the Forbes Non-Profit Council and Executive in Residence for Entrepreneurship at the University at Buffalo School of Management. He is an entrepreneurial leader with broad-based management experience in both the for-profit and non-profit sectors. His passion for social innovation is focused on nurturing strong relationships and building consensus across diverse groups of stakeholders in the academic, for-profit, non-profit and government sectors.

In this HCI Webinar, I talk with Derrick Mains about returning to the fundamentals of leadership. Derrick Mains is an Emmy Award-winning content creator, four-time author, operations podcaster with more than 160,000 monthly listeners, and optimization consultant renowned for his pioneering work in process engineering, design, and transformation. Mains work spans more than 20 years and 150 companies across nearly every industry, from early-stage companies on through the Fortune 10. Mains approach melds essentialism with a keen focus on human-centric system design, emphasizing the need for regular audits, reflection, and reinvestment to achieve optimization. Mains believes that all organizational systems share a fundamental purpose: transforming the input of resources into value, through outputs. He highlights how, without active management, systems degrade, leading to inefficiency and value and margin fade. His philosophy underscores the criticality of understanding the interconnectedness of systems and their natural progression towards entropy.

This video discusses the growing role of hiring algorithms in recruitment processes, highlighting how these digital tools have replaced humans as the initial gatekeepers in job applications. Over 90% of US employers use some form of algorithm to filter candidates, significantly reducing the pool of applicants before any human ever reviews resumes or interviews. Although these systems are designed to improve efficiency and fairness, they often perpetuate and amplify biases inherent in their training data. This results in discriminatory outcomes, particularly against Black and Asian applicants, with some roles exhibiting over 10% bias against Black candidates. A monoculture of algorithms, where many companies utilize the same platforms like HireVue, means that flaws in one system can affect thousands of applicants across multiple employers. The lack of transparency and feedback leaves candidates in the dark about why they were rejected, with no recourse for appeal or improvement. The system, while ostensibly objective, is biased, demoralizing, and can block career opportunities unfairly. The video concludes with solutions: auditing algorithms for fairness at the job level, embracing transparency about AI usage in hiring, establishing appeals processes, diversifying hiring tools, and empowering government and independent audits. Standardizing evaluation protocols for hiring AI and creating public registries of bias results are critical to making algorithms a force for inclusion rather than exclusion. Highlights 🤖 Over 90% of US employers now use hiring algorithms to screen candidates before human review. ⚠️ Hiring algorithms can filter out up to 70% of applicants, often without transparency or feedback. 📉 Algorithms trained on biased data amplify discrimination, particularly against Black and Asian applicants. 🔄 A monoculture of hiring tools means a single flawed algorithm can negatively impact thousands of candidates across many companies. 🚫 Candidates often face blacklisting with no recourse or explanation, harmed both professionally and emotionally. ✅ Solutions include auditing algorithms for fairness, transparency about AI use, and creating appeals processes for candidates. 🌍 Collective action and governmental oversight, alongside public bias audits, are necessary to ensure fair and just hiring algorithms. Key Insights 🤖 Algorithmic Gatekeeping Has Become the Norm: With more than 90% of US employers adopting automated hiring tools, the recruitment landscape has fundamentally shifted. Algorithms act as the first interviewer, screening candidates via resume keyword matching, video analysis, or gamified personality assessments. This introduces a new layer of institutional filtering that shapes who even gets a chance for human interaction, profoundly impacting employment access on a mass scale. ⚠️ Bias Is Inherent and Amplified, Not Eliminated: Contrary to the belief that AI is objective, hiring algorithms inherit and often exacerbate historical biases because they learn from past hiring data. If previous hiring decisions were biased, the algorithm replicates and scales those biases, disproportionately disadvantaging minority groups. The data reveals that over 10% of job openings show evidence of bias against Black applicants, exposing a clear failure to create equitable hiring practices. 🔄 The Problem Multiplies via Algorithmic Monoculture: Many companies use the same handful of hiring platforms. For instance, HireVue is used by over 60% of Fortune 100 firms. This monoculture results in a widespread propagation of the same biases and errors, instead of diversified approaches that could mitigate risks. A single flawed algorithm can effectively blacklist an individual from multiple opportunities, creating career barriers that are difficult to overcome. 🚫 Lack of Transparency and Feedback Creates a Black Box Effect: Job seekers receive rejection notices without explanations, leaving them with no understanding of their shortcomings. They cannot identify whether their resume keywords, video interview responses, or game-based personality tests caused their rejection. Furthermore, many platforms deny candidates the opportunity to retake assessments for extended periods, closing off pathways to improvement and human review. This opacity frustrates applicants, harms mental health, and perpetuates systemic inequality. ✅ Empowering Candidates and Companies Is Crucial: The system’s faults aren’t immutable. Auditing algorithms for fairness should occur at the level of each job, not just in aggregate, to identify role-specific biases. Transparency laws, such as New York City’s Local Law 144, offer a model for making algorithmic processes visible to candidates. Additionally, appeals mechanisms and human review options can provide recourse for unfair decisions. Companies must diversify their hiring tools and establish committees to continuously monitor fairness and predictive validity.

This research explores the phenomenon of algorithmic monoculture in the labor market, where a high concentration of employers relies on the same few vendors for automated hiring tools. Research into millions of applications suggests that while vendors may claim overall fairness, disaggregated data reveals significant racial bias at the individual position level. This widespread dependency creates a systemic exclusion effect, where an applicant rejected by one algorithm is likely to be automatically disqualified across many different firms. The research argues that this lack of vendor diversity and transparency undermines legal protections and economic productivity by trapping qualified candidates in a cycle of unemployment. To address these vulnerabilities, the research advocates for regular bias audits, increased regulatory oversight, and the implementation of human-centered oversight in the recruitment process. Ultimately, the research warns that unchecked algorithmic consolidation transforms localized hiring errors into structural barriers for marginalized job seekers.

The current landscape of artificial intelligence (AI) in human resources (HR) is often misunderstood, with many companies marketing AI “agents” that are essentially sophisticated, AI-powered workflows rather than truly autonomous systems. A 2024 Deote survey reveals that although 65% of large enterprises claim to use AI agents in HR, less than 15% deploy real autonomous agents. This distinction is crucial because workflows and agents serve fundamentally different purposes, cost structures, and risk profiles. Highlights 🤖 Many HR “AI agents” are just AI-powered workflows, not truly autonomous systems. 💸 Agents cost 3 to 10 times more than workflows, making cost-benefit analysis critical. 🔄 Workflows excel at structured, rule-based HR tasks like onboarding and promotions. ⚖️ Transparency, fairness, and compliance are stronger with workflows than autonomous agents. 🚨 Johnson & Johnson replaced biased AI agents with human-guided workflows to restore fairness. ✅ Leading companies use agents only for complex, high-stakes HR problems. 🧑‍💼 The best approach combines human judgment with AI workflows for routine tasks. Key Insights 🤔 Distinguishing AI Workflows from True AI Agents: A Critical Strategic Decision The distinction between AI-powered workflows and autonomous AI agents is essential yet often blurred due to marketing hype. Workflows automate predefined, transparent steps and are predictable and auditable, making them ideal for HR processes that require consistency and fairness. Autonomous agents operate with greater flexibility but introduce unpredictability and opacity. This difference influences cost, compliance risk, and employee trust. Misclassifying workflows as agents can lead to unnecessary expenditure and risk. 💰 Cost Implications and ROI Must Drive AI Adoption in HR The financial impact of substituting workflows with autonomous agents can be substantial, with agent-based systems costing up to five times more initially and additional unpredictable ongoing expenses. This calls for rigorous cost-benefit analysis. Most HR tasks, being routine and rule-based, do not justify the high costs of agents. Only high-value, complex problems warrant the investment in autonomous agents, ensuring return on investment and minimizing waste. 🛠️ Structured Tasks Are Best Served by AI-Powered Workflows Onboarding, promotions, benefits enrollment, and similar processes follow a clear sequence, akin to a recipe. Workflows automate these reliably with human control and predictable outcomes. This approach supports compliance requirements by maintaining clear audit trails and predictable decision-making. AI workflows excel where rules and outcomes are well-defined and repeatable, contributing to operational efficiency without sacrificing transparency. 🔍 Autonomous Agents Can Introduce Risk, Bias, and Erode Employee Trust AI agents, due to their flexible and opaque nature, can generate decisions that are difficult to explain or defend, complicating regulatory compliance. Hidden biases within their models can harm diversity and inclusion efforts, as evidenced by the Johnson & Johnson case, leading to employee disengagement and loss of trust. This risk necessitates keeping humans in the decision loop when autonomy increases the stakes. 👨‍💼 Human-Guided AI Is the Optimal Balance for Future HR Models The future of HR automation depends on augmenting human judgment with AI tools that preserve control, fairness, and clarity. Human-guided AI workflows combine the efficiency of automation with human oversight, promoting fairness and employee confidence. This balanced model lets humans focus on strategic decisions and complex interpersonal dynamics while delegating repetitive tasks to AI, driving both productivity and employee satisfaction. 🧩 Four Guiding Questions for AI Deployment in HR HR leaders should evaluate AI tools by assessing: (1) task complexity—can it be mapped via clear if-then logic for workflows? (2) cost-value alignment—does the problem’s value justify agent expenses? (3) AI capabilities—can the system handle nuanced or ambiguous decisions autonomously? (4) compliance needs—is transparency and auditability a priority? These questions provide a framework for making data-driven, practical decisions that align technology with organizational goals. 🌟 Industry Leaders Demonstrate Best Practices Companies like IBM, Salesforce, and Unilever showcase effective AI integration by reserving autonomous agents for truly complex scenarios while widely deploying AI workflows for standard HR tasks. Their approach underscores that AI’s power is maximized through strategic application aligned with task nature and human oversight, mitigating risk and optimizing cost. Such examples affirm that sweeping agent adoption for routine tasks is neither necessary nor advisable.

This research analyzes the strategic choice between deterministic workflows and autonomous agents within human resources technology. While current market trends favor highly complex agentic AI, the author argues that structured workflows are superior for the vast majority of HR tasks due to their lower costs, greater transparency, and predictable audit trails. To guide technology selection, the research introduces a four-part diagnostic framework assessing task complexity, economic value, AI reliability, and the potential impact of errors. By prioritizing human-supervised workflows for routine processes, organizations can reserve expensive autonomous systems for high-value scenario planning that requires dynamic decision-making. Ultimately, the research cautions that over-engineering AI solutions can lead to budget overruns and a loss of stakeholder trust through opaque, "black-box" results.

This research analyzes the strategic choice between deterministic workflows and autonomous agents within human resources technology. While current market trends favor highly complex agentic AI, the author argues that structured workflows are superior for the vast majority of HR tasks due to their lower costs, greater transparency, and predictable audit trails. To guide technology selection, the research introduces a four-part diagnostic framework assessing task complexity, economic value, AI reliability, and the potential impact of errors. By prioritizing human-supervised workflows for routine processes, organizations can reserve expensive autonomous systems for high-value scenario planning that requires dynamic decision-making. Ultimately, the research cautions that over-engineering AI solutions can lead to budget overruns and a loss of stakeholder trust through opaque, "black-box" results. See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

In this podcast episode, Dr. Jonathan H. Westover talks with Derrick Mains about returning to the fundamentals of leadership. Derrick Mains is an Emmy Award-winning content creator, four-time author, operations podcaster with more than 160,000 monthly listeners, and optimization consultant renowned for his pioneering work in process engineering, design, and transformation. Mains work spans more than 20 years and 150 companies across nearly every industry, from early-stage companies on through the Fortune 10. Mains approach melds essentialism with a keen focus on human-centric system design, emphasizing the need for regular audits, reflection, and reinvestment to achieve optimization. Mains believes that all organizational systems share a fundamental purpose: transforming the input of resources into value, through outputs. He highlights how, without active management, systems degrade, leading to inefficiency and value and margin fade. His philosophy underscores the criticality of understanding the interconnectedness of systems and their natural progression towards entropy. See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.