Starting Line

Consider this ● a local bakery, beloved for its custom cakes, begins using an AI-powered tool to predict demand and manage inventory. Sounds efficient, right? Perhaps. But what if this algorithm, trained on past sales data predominantly from wealthier neighborhoods, consistently underestimates demand in lower-income areas?

Suddenly, customers in those areas face empty shelves more often, a subtle yet significant inequity baked into the system. This isn’t some abstract tech dystopia; it’s the reality of algorithmic bias Meaning ● Algorithmic bias in SMBs: unfair outcomes from automated systems due to flawed data or design. creeping into the everyday operations of small and medium-sized businesses (SMBs). Algorithmic fairness, often perceived as a concern for tech giants and government regulators, holds tangible implications for even the smallest enterprises. It’s about ensuring that the digital tools SMBs increasingly rely on don’t inadvertently discriminate or create unfair outcomes for customers, employees, or the business itself.

Unpacking Algorithmic Fairness

Algorithmic fairness, at its core, addresses bias in automated decision-making systems. These systems, driven by algorithms, are now pervasive in SMB operations, from marketing automation Meaning ● Marketing Automation for SMBs: Strategically automating marketing tasks to enhance efficiency, personalize customer experiences, and drive sustainable business growth. and customer relationship management (CRM) to hiring platforms and financial tools. Algorithms learn from data, and if that data reflects existing societal biases ● which it often does ● the algorithms can perpetuate and even amplify those biases.

For an SMB, this can manifest in numerous ways ● biased loan application assessments, skewed marketing campaigns that exclude certain demographics, or hiring processes that inadvertently filter out qualified candidates from underrepresented groups. Ignoring algorithmic fairness Meaning ● Ensuring impartial automated decisions in SMBs to foster trust and equitable business growth. isn’t just an ethical oversight; it’s a business risk.

Why Fairness Matters for SMBs

For SMBs, the immediate question might be ● why should I, a small business owner juggling payroll and marketing, worry about something as seemingly complex as algorithmic fairness? The answer lies in several key areas. First, reputation. In today’s hyper-connected world, news of unfair practices, even unintentional ones, spreads rapidly.

A perception of bias, whether justified or not, can severely damage an SMB’s brand and customer loyalty. Consumers, especially younger generations, are increasingly conscious of ethical business Meaning ● Ethical Business for SMBs: Integrating moral principles into operations and strategy for sustainable growth and positive impact. practices and are more likely to support companies that align with their values. Second, legal and regulatory landscape. While specific regulations around algorithmic fairness are still evolving, the trend is clear ● businesses will be held accountable for discriminatory outcomes resulting from their use of AI and algorithms.

Ignoring fairness now could lead to legal challenges and penalties down the line. Third, business performance. Biased algorithms can lead to suboptimal business decisions. For instance, a hiring algorithm that favors a narrow demographic might overlook highly qualified candidates from diverse backgrounds, limiting the talent pool and hindering innovation.

Similarly, marketing algorithms that exclude certain customer segments miss out on potential revenue streams. Fairness, therefore, isn’t just about doing the right thing; it’s about building a sustainable and successful business in the long run.

Algorithmic fairness for SMBs is not merely about ethical compliance; it’s a strategic imperative for long-term business health and resilience.

Practical First Steps

Implementing algorithmic fairness frameworks might sound daunting, especially for resource-constrained SMBs. However, practical first steps are achievable and don’t require extensive technical expertise. The initial focus should be on awareness and assessment. SMB owners need to understand where algorithms are being used in their business operations.

This involves taking inventory of software and tools that utilize AI or machine learning. Once identified, the next step is to assess the potential for bias in these systems. This can start with asking simple questions ● What data is this algorithm trained on? Could this data reflect existing societal biases?

What are the potential outcomes of this algorithm’s decisions, and could they disproportionately affect certain groups of people? For example, an SMB using a marketing automation platform should examine the data used to segment customers and personalize campaigns. Is this data representative of the entire customer base, or is it skewed towards certain demographics? Are the campaign messages inclusive and unbiased?

Similarly, an SMB using an applicant tracking system (ATS) should understand how the system filters and ranks candidates. Are the keywords and criteria used in the ATS potentially biased against certain groups? These initial assessments don’t require deep technical audits but rather a critical and thoughtful examination of existing processes and tools.

Building a Fairness-Conscious Culture

Beyond specific tools and algorithms, fostering a fairness-conscious culture within the SMB is paramount. This starts with leadership commitment. SMB owners and managers need to explicitly communicate the importance of fairness and ethical considerations in all business operations, including the use of algorithms. This commitment should be reflected in company values, policies, and training programs.

Employee training is crucial. All employees, not just technical staff, should be educated about algorithmic bias and its potential impact. They should be empowered to identify and raise concerns about potentially unfair algorithmic practices. This requires creating a safe and open environment where employees feel comfortable speaking up without fear of reprisal.

Furthermore, diversity and inclusion within the SMB workforce play a significant role in promoting algorithmic fairness. A diverse team is more likely to identify and mitigate potential biases in algorithms and data. Bringing in different perspectives and experiences helps to challenge assumptions and ensure that fairness considerations are embedded throughout the organization. Building a fairness-conscious culture is an ongoing process, not a one-time fix. It requires continuous learning, adaptation, and a genuine commitment to ethical business practices.

The journey towards algorithmic fairness for SMBs begins with recognizing that this isn’t an abstract problem confined to tech giants. It’s a practical business challenge with real-world consequences for SMBs of all sizes. By taking these fundamental steps ● understanding algorithmic fairness, assessing potential biases, and building a fairness-conscious culture ● SMBs can begin to navigate this complex landscape and ensure that their use of technology aligns with their values and business goals.

Navigating The Labyrinth

The initial foray into algorithmic fairness for SMBs, while crucial, merely scratches the surface. Moving beyond basic awareness requires a more structured and methodological approach. Consider a growing e-commerce SMB utilizing algorithmic recommendation engines to boost sales. These algorithms, designed to personalize customer experiences, can inadvertently create filter bubbles, limiting product discovery and potentially reinforcing existing purchasing patterns based on demographic data.

While seemingly benign, this can lead to missed opportunities and even discriminatory outcomes if certain customer segments are consistently steered towards lower-value products or excluded from promotions. Navigating this labyrinth of algorithmic implications demands a more sophisticated framework, one that integrates fairness considerations into the very fabric of SMB operations Meaning ● SMB Operations represent the coordinated activities driving efficiency and scalability within small to medium-sized businesses. and strategic decision-making.

Developing a Fairness Framework ● A Practical Guide

For SMBs, developing a comprehensive algorithmic fairness framework Meaning ● Algorithmic Fairness Frameworks, for small and medium-sized businesses (SMBs), represent structured approaches to ensure that automated decision-making systems do not unfairly discriminate against individuals or groups. doesn’t necessitate reinventing the wheel. Existing frameworks and principles, adapted to the SMB context, can provide a robust foundation. A practical approach involves several key stages ● Assessment, Design, Implementation, and Monitoring (ADIM). This cyclical process allows for continuous improvement Meaning ● Ongoing, incremental improvements focused on agility and value for SMB success. and adaptation as the SMB grows and its use of algorithms evolves.

Assessment ● Identifying Algorithmic Touchpoints and Risks

The assessment phase expands on the initial inventory of algorithms. It involves a deeper dive into identifying specific algorithmic touchpoints within the SMB’s value chain. This includes not only customer-facing algorithms like marketing automation and recommendation engines but also internal algorithms used for human resources, supply chain management, and financial operations. For each touchpoint, the assessment should evaluate the potential for bias and its potential impact.

This requires considering various dimensions of fairness, moving beyond simple demographic parity. Distributive Fairness, for example, examines whether algorithmic outcomes are distributed equitably across different groups. Procedural Fairness focuses on the transparency and explainability of the algorithmic decision-making process. Counterfactual Fairness, a more advanced concept, considers whether an outcome would have been different if sensitive attributes like race or gender were removed.

SMBs don’t need to become experts in all these fairness metrics, but understanding the different dimensions helps to identify potential blind spots and tailor fairness interventions accordingly. This assessment should also consider the specific business context and risk tolerance of the SMB. A high-growth startup might prioritize speed and agility, accepting a higher level of algorithmic risk initially, while a more established SMB might prioritize stability and reputation, demanding a more rigorous fairness framework from the outset.

Design ● Integrating Fairness into Algorithm Development and Selection

The design phase focuses on proactively integrating fairness considerations into the development or selection of algorithms. For SMBs that develop their own algorithms, this means incorporating fairness metrics Meaning ● Fairness Metrics, within the SMB framework of expansion and automation, represent the quantifiable measures utilized to assess and mitigate biases inherent in automated systems, particularly algorithms used in decision-making processes. into the algorithm design process. This could involve using techniques like Adversarial Debiasing, which aims to train algorithms to be invariant to sensitive attributes, or Fairness-Aware Machine Learning, which directly optimizes algorithms for fairness alongside accuracy. However, many SMBs rely on off-the-shelf algorithmic solutions provided by third-party vendors.

In this case, the design phase shifts to algorithm selection. SMBs should rigorously evaluate vendors based on their commitment to fairness and transparency. This includes asking vendors about their data sources, algorithm design principles, and fairness testing procedures. Vendor Questionnaires and Due Diligence Checklists can be valuable tools in this process.

Furthermore, SMBs should seek vendors who provide explainable AI Meaning ● XAI for SMBs: Making AI understandable and trustworthy for small business growth and ethical automation. (XAI) solutions, allowing for greater transparency into how algorithms arrive at their decisions. Explainability is crucial for identifying and mitigating potential biases, as well as for building trust with customers and employees. The design phase should also consider the broader system within which the algorithm operates. Fairness isn’t solely a technical problem; it’s also a systemic one.

Designing fair algorithms requires considering the social, organizational, and business context in which they are deployed. This might involve redesigning workflows, processes, or even business models to promote fairness and equity.

Implementation ● Deploying and Operationalizing Fairness Measures

Implementation translates the fairness framework into concrete actions. This involves deploying fairness measures throughout the algorithm lifecycle, from data collection and preprocessing to algorithm training and deployment. Data Preprocessing Techniques, such as re-weighting or resampling, can be used to mitigate biases in training data. In-Processing Techniques modify the algorithm training process to directly address fairness constraints.

Post-Processing Techniques adjust algorithm outputs to improve fairness after the algorithm has been trained. The choice of techniques depends on the specific algorithm, data, and fairness goals of the SMB. Implementation also involves establishing clear roles and responsibilities for algorithmic fairness within the organization. This might involve designating a fairness champion or creating a cross-functional fairness committee to oversee the implementation and monitoring of the framework.

Furthermore, SMBs should develop clear guidelines and protocols for handling fairness-related issues or complaints. This includes establishing channels for reporting potential biases, procedures for investigating complaints, and mechanisms for redress when unfair outcomes are identified. Effective implementation requires ongoing communication and collaboration across different departments and stakeholders within the SMB.

Monitoring ● Evaluating and Iterating on Fairness Performance

Monitoring is the crucial feedback loop in the ADIM cycle. It involves continuously evaluating the fairness performance of deployed algorithms and iterating on the framework based on the findings. Fairness Metrics, such as disparate impact ratio, equal opportunity difference, and predictive parity, can be used to quantify and track fairness performance over time. However, relying solely on metrics is insufficient.

Monitoring should also involve qualitative assessments, such as user feedback, employee surveys, and external audits. Regular Algorithmic Audits, conducted internally or by external experts, can help to identify and address hidden biases or unintended consequences. Monitoring should also track the business impact of fairness interventions. Are fairness measures negatively impacting business performance metrics like revenue or efficiency?

If so, adjustments might be needed to balance fairness with business goals. The monitoring phase should inform the next iteration of the ADIM cycle. Findings from monitoring should be used to refine the assessment, redesign algorithms, and improve implementation practices. This iterative process ensures that the fairness framework remains relevant and effective as the SMB evolves and its algorithmic landscape changes.

Developing and implementing an algorithmic fairness framework is not a one-time project but an ongoing journey. It requires a commitment to continuous learning, adaptation, and improvement. By adopting a structured approach like ADIM, SMBs can move beyond reactive fairness interventions and proactively build fairer and more equitable algorithmic systems.

A structured approach to algorithmic fairness, like the ADIM cycle, empowers SMBs to proactively manage bias and build more equitable systems.

The Strategic Advantage of Algorithmic Fairness

Beyond ethical considerations and risk mitigation, algorithmic fairness can be a strategic advantage Meaning ● Strategic Advantage, in the realm of SMB growth, automation, and implementation, represents a business's unique capacity to consistently outperform competitors by leveraging distinct resources, competencies, or strategies; for a small business, this often means identifying niche markets or operational efficiencies achievable through targeted automation. for SMBs. In an increasingly competitive marketplace, businesses that prioritize fairness and ethical AI Meaning ● Ethical AI for SMBs means using AI responsibly to build trust, ensure fairness, and drive sustainable growth, not just for profit but for societal benefit. practices can differentiate themselves and build stronger brand loyalty. Consumers are increasingly discerning and are drawn to businesses that demonstrate a commitment to social responsibility. Algorithmic fairness can be a powerful signal of this commitment.

Furthermore, fairer algorithms can lead to better business outcomes. By mitigating biases, SMBs can tap into previously overlooked customer segments, improve employee engagement, and make more informed decisions. For example, a fairer hiring algorithm can lead to a more diverse and talented workforce, driving innovation and creativity. A fairer marketing algorithm can reach a wider range of customers, increasing market share and revenue.

Algorithmic fairness can also enhance trust and transparency, both internally and externally. Explainable AI solutions, coupled with transparent fairness practices, can build trust with customers, employees, and stakeholders. This trust is essential for long-term business sustainability and growth. In the age of AI, fairness is not merely a cost of doing business; it’s an investment in a more equitable and prosperous future for SMBs.

Navigating the labyrinth of algorithmic fairness requires a strategic mindset and a commitment to continuous improvement. By developing and implementing a robust fairness framework, SMBs can not only mitigate risks and uphold ethical standards but also unlock new opportunities for growth, innovation, and competitive advantage.

The Algorithmic Mandate

The transition from intermediate frameworks to advanced algorithmic fairness implementation for SMBs necessitates a paradigm shift. It moves beyond reactive mitigation and strategic advantage to an algorithmic mandate ● a fundamental re-evaluation of business operations through the lens of algorithmic ethics Meaning ● Algorithmic Ethics, within the realm of SMB operations, concerns the moral considerations regarding the design, deployment, and utilization of algorithms, particularly in automated processes and strategic decision-making impacting business growth. and societal impact. Consider a fintech SMB leveraging AI for credit scoring.

While regulatory compliance and risk management remain paramount, a truly advanced approach delves into the systemic implications of credit access, questioning not merely the fairness of the algorithm but the very definition of creditworthiness embedded within the data and the algorithm itself. This level of engagement demands a critical interrogation of business models, a willingness to challenge industry norms, and a commitment to algorithmic justice Meaning ● Algorithmic Justice, within the framework of SMB growth strategies, pertains to the ethical design, development, and deployment of automated systems and artificial intelligence. as a core tenet of SMB growth Meaning ● SMB Growth is the strategic expansion of small to medium businesses focusing on sustainable value, ethical practices, and advanced automation for long-term success. and automation.

Algorithmic Justice ● Beyond Fairness Metrics

Algorithmic justice extends beyond the technical realm of fairness metrics and statistical parity. It delves into the philosophical and societal dimensions of algorithmic decision-making, asking deeper questions about power, equity, and systemic bias. While fairness metrics provide valuable quantitative measures, they often fail to capture the qualitative nuances of justice. Distributive Justice, in the algorithmic context, moves beyond equal outcomes to consider equitable resource allocation and opportunity distribution.

It questions whether algorithms are perpetuating or exacerbating existing inequalities in society. Procedural Justice expands on transparency and explainability to encompass accountability and redress mechanisms. It asks whether those affected by algorithmic decisions have meaningful recourse and avenues for challenging unfair outcomes. Recognition Justice, perhaps the most profound dimension, focuses on acknowledging and respecting the dignity and agency of individuals and groups impacted by algorithms.

It challenges algorithmic systems that dehumanize or marginalize certain populations. For SMBs, embracing algorithmic justice requires moving beyond a compliance-driven approach to fairness and adopting a values-driven approach to algorithmic design and deployment. This involves engaging in critical self-reflection, stakeholder dialogue, and a willingness to challenge the status quo.

The Business Case for Algorithmic Justice ● Long-Term Value Creation

While the immediate business case for algorithmic fairness might focus on risk mitigation and brand reputation, the long-term business case for algorithmic justice is rooted in sustainable value creation. SMBs that embrace algorithmic justice are not merely avoiding harm; they are actively contributing to a more equitable and just society, which in turn strengthens their long-term business prospects. Enhanced Innovation is a key benefit. By challenging algorithmic biases and seeking to address systemic inequalities, SMBs can unlock new avenues for innovation and develop products and services that are more inclusive and relevant to diverse populations.

This can lead to the discovery of unmet needs and the creation of new markets. Increased Customer Loyalty is another long-term advantage. Consumers are increasingly drawn to businesses that align with their values and demonstrate a commitment to social justice. SMBs that are perceived as being algorithmically just are more likely to attract and retain customers who value ethical business practices.

Improved Employee Engagement is also a significant factor. Employees, especially younger generations, are increasingly seeking purpose-driven work and are more likely to be engaged and motivated when they believe their employer is making a positive societal impact. Algorithmic justice can be a powerful tool for attracting and retaining top talent. Stronger Stakeholder Relationships extend beyond customers and employees to include investors, partners, and the broader community.

SMBs that are perceived as being algorithmically just are more likely to build trust and goodwill with all stakeholders, fostering stronger and more resilient business relationships. In the long run, algorithmic justice is not merely a cost center; it’s a value driver, contributing to enhanced innovation, customer loyalty, employee engagement, and stronger stakeholder relationships, all of which are essential for sustainable SMB growth and success.

Algorithmic justice for SMBs transcends ethical compliance, becoming a potent driver of long-term value creation Meaning ● Long-Term Value Creation in the SMB context signifies strategically building a durable competitive advantage and enhanced profitability extending beyond immediate gains, incorporating considerations for automation and scalable implementation. and sustainable business growth.

Implementing Algorithmic Justice ● A Multi-Dimensional Strategy

Implementing algorithmic justice requires a multi-dimensional strategy that goes beyond technical fixes and incorporates organizational, ethical, and societal considerations. This strategy can be framed around four interconnected pillars ● Data Justice, Algorithmic Accountability, Participatory Design, and Systemic Advocacy (DAPS). These pillars provide a holistic framework for embedding algorithmic justice into the core of SMB operations and strategic decision-making.

Data Justice ● Reclaiming Data Agency and Equity

Data justice recognizes that data is not neutral; it reflects power dynamics and societal inequalities. Implementing data justice Meaning ● Data Justice, within the purview of Small and Medium-sized Businesses (SMBs), signifies the ethical and equitable governance of data practices, emphasizing fairness, transparency, and accountability in data handling. for SMBs involves reclaiming data agency and promoting data equity. This starts with Critical Data Sourcing. SMBs need to critically evaluate their data sources, recognizing potential biases and limitations.

This might involve seeking out alternative data sources that are more representative and inclusive, or actively working to debias existing datasets. Data Minimization is another key principle. SMBs should collect and use only the data that is strictly necessary for their business purposes, minimizing the potential for misuse or harm. Data Transparency is crucial for building trust and accountability.

SMBs should be transparent about what data they collect, how they use it, and with whom they share it. This includes providing clear and accessible privacy policies and data governance frameworks. Data Ownership and Control are fundamental aspects of data justice. SMBs should respect individuals’ rights to control their own data, including the right to access, rectify, and delete their data.

This might involve implementing data portability mechanisms and empowering users to manage their data preferences. Data justice is not merely about data privacy or security; it’s about ensuring that data practices are aligned with principles of equity, fairness, and respect for human rights.

Algorithmic Accountability ● Establishing Oversight and Redress

Algorithmic accountability focuses on establishing mechanisms for oversight, redress, and responsibility for algorithmic decisions. For SMBs, this involves several key steps. Algorithmic Audits need to move beyond technical assessments of fairness metrics to encompass broader ethical and societal impact Meaning ● Societal Impact for SMBs: The total effect a business has on society and the environment, encompassing ethical practices, community contributions, and sustainability. assessments. These audits should be conducted regularly and transparently, involving diverse stakeholders and external experts.

Explainability and Interpretability are crucial for accountability. SMBs should prioritize the use of explainable AI solutions and develop mechanisms for explaining algorithmic decisions to affected individuals and groups. This includes providing clear and accessible justifications for algorithmic outcomes and allowing for human review and override when necessary. Redress Mechanisms are essential for addressing unfair algorithmic outcomes.

SMBs should establish clear channels for reporting complaints, procedures for investigating and resolving disputes, and mechanisms for providing remedies or compensation when harm is identified. Organizational Responsibility needs to be clearly defined. SMBs should designate individuals or teams responsible for algorithmic fairness and accountability, ensuring that these responsibilities are integrated into performance evaluations and organizational structures. Algorithmic accountability Meaning ● Taking responsibility for algorithm-driven outcomes in SMBs, ensuring fairness, transparency, and ethical practices. is not about blaming algorithms; it’s about establishing human oversight and responsibility for the design, deployment, and impact of algorithmic systems.

Participatory Design ● Engaging Stakeholders in Algorithm Development

Participatory design recognizes that algorithmic fairness is not solely a technical problem to be solved by engineers; it’s a social and ethical challenge that requires the active participation of diverse stakeholders. For SMBs, this involves engaging stakeholders throughout the algorithm lifecycle. Stakeholder Mapping is the first step. SMBs need to identify all relevant stakeholders who are affected by their algorithmic systems, including customers, employees, community groups, and regulatory bodies.

Consultation and Dialogue are crucial for understanding stakeholder perspectives and concerns. SMBs should establish channels for ongoing communication and dialogue with stakeholders, seeking their input on algorithmic design, deployment, and impact. Co-Creation and Co-Design involve actively involving stakeholders in the algorithm development process. This might include participatory workshops, user testing, and community advisory boards.

Feedback Loops and Iteration are essential for ensuring that algorithmic systems are responsive to stakeholder needs and concerns. SMBs should establish mechanisms for collecting and incorporating stakeholder feedback into ongoing algorithm refinement and improvement. Participatory design is not merely about gathering input; it’s about empowering stakeholders to shape the design and deployment of algorithmic systems in ways that are fair, equitable, and aligned with their values.

Systemic Advocacy ● Contributing to Algorithmic Justice at Scale

Systemic advocacy recognizes that algorithmic justice is not solely an individual SMB responsibility; it’s a collective challenge that requires broader societal and industry-level action. For SMBs, this involves contributing to algorithmic justice at scale. Industry Collaboration is crucial for sharing best practices, developing common standards, and advocating for policy changes that promote algorithmic fairness across the industry. SMBs can participate in industry associations, consortia, and working groups focused on algorithmic ethics and justice.

Policy Engagement is essential for shaping the regulatory landscape. SMBs should engage with policymakers and regulators to advocate for policies that promote algorithmic fairness and accountability, while also ensuring that regulations are practical and proportionate for SMBs. Public Education and Awareness are important for raising awareness about algorithmic bias and its societal impact. SMBs can contribute to public education initiatives, share their experiences and learnings, and advocate for greater algorithmic literacy among the public.

Ethical Sourcing and Procurement can be powerful tools for promoting algorithmic justice throughout the supply chain. SMBs should prioritize vendors and partners who demonstrate a commitment to algorithmic fairness and ethical AI practices. Systemic advocacy is not merely about corporate social responsibility; it’s about recognizing that SMBs have a collective responsibility to contribute to a more just and equitable algorithmic ecosystem.

Implementing algorithmic justice is a complex and ongoing endeavor, but it is also a profound opportunity for SMBs to lead the way in building a more ethical and equitable future for AI. By embracing the algorithmic mandate and adopting a multi-dimensional strategy grounded in data justice, algorithmic accountability, participatory design, and systemic advocacy, SMBs can not only navigate the challenges of algorithmic fairness but also unlock new pathways for sustainable growth, innovation, and societal impact.

1. Prioritize Explainability ● Choose algorithmic solutions that offer transparency into their decision-making processes.
2. Establish Feedback Loops ● Create mechanisms for users and employees to report concerns about algorithmic fairness.
3. Invest in Training ● Educate your team on algorithmic bias and ethical AI practices.
4. Seek Diverse Perspectives ● Involve diverse teams in the design, development, and auditing of algorithms.

References

• O’Neil, Cathy. Weapons of Math Destruction ● How Big Data Increases Inequality and Threatens Democracy. Crown, 2016.
• Barocas, Solon, et al. Fairness and ● Limitations and Opportunities. MIT Press, 2023.
• Holstein, Kenneth, et al. “Improving Fairness in Machine Learning Systems ● What Do Industry Practitioners Need?” Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, ACM, 2019, pp. 1-16.
• Mehrabi, Ninareh K., et al. “A Survey on Bias and Fairness in Machine Learning.” ACM Computing Surveys (CSUR), vol. 54, no. 6, 2021, pp. 1-35.
• Selbst, Andrew D., et al. “Fairness and Abstraction in Sociotechnical Systems.” Proceedings of the 1st Conference on Fairness, Accountability and Transparency, ACM, 2018, pp. 59-68.