Fundamentals

A startling statistic often overlooked in discussions about artificial intelligence is that SMBs, despite representing the vast majority of businesses globally, are significantly less likely to have dedicated resources for AI ethics than their larger counterparts. This disparity creates a critical vulnerability, especially as algorithms increasingly influence core business functions even within smaller operations. For a local bakery using an automated scheduling tool, or a small e-commerce store employing AI-driven product recommendations, the concept of algorithmic fairness Meaning ● Ensuring impartial automated decisions in SMBs to foster trust and equitable business growth. might seem abstract, detached from the daily grind of invoices and customer service.

However, this detachment is a dangerous misconception. The very algorithms promising efficiency and growth Meaning ● Growth for SMBs is the sustainable amplification of value through strategic adaptation and capability enhancement in a dynamic market. can inadvertently perpetuate biases, leading to skewed outcomes that undermine the SMB’s long-term viability and ethical standing within its community.

Understanding Algorithmic Bias

Bias in algorithms isn’t some intentional malevolence programmed by shadowy figures; it’s a far more mundane, yet equally impactful, reflection of the data they are trained on. Imagine a hiring algorithm trained primarily on historical data where certain demographics were underrepresented in specific roles. Unintentionally, this algorithm might learn to favor candidates mirroring the existing, skewed workforce, effectively locking out qualified individuals from underrepresented groups. This isn’t about the algorithm being prejudiced; it’s about the data reflecting existing societal or organizational biases, which the algorithm then amplifies.

For SMBs, often operating with limited datasets, the risk of inheriting and perpetuating such biases is amplified. A small dataset reflecting past hiring practices, for instance, can easily encode unintentional discrimination, leading to unfair hiring processes in the future.

Why Fairness Matters for SMBs

Some might argue that for an SMB just trying to stay afloat, worrying about algorithmic fairness is a luxury they cannot afford. This viewpoint, while understandable given the pressures SMBs Meaning ● SMBs are dynamic businesses, vital to economies, characterized by agility, customer focus, and innovation. face, misses a crucial point. Fairness isn’t simply a moral imperative; it’s a sound business strategy, particularly for smaller enterprises deeply embedded in their local communities. Consider the reputational damage a local business might suffer if its AI-powered customer service chatbot consistently misgenders or misunderstands customers from a particular linguistic background.

Word of mouth travels fast in close-knit communities, and such missteps can quickly erode customer trust and loyalty. In contrast, a commitment to fairness, even in algorithmic applications, can become a powerful differentiator, attracting customers who value ethical business practices and fostering a positive brand image. In an era where consumers are increasingly conscious of social responsibility, algorithmic fairness transforms from a potential cost center into a valuable asset, enhancing brand reputation and customer relationships.

Practical Steps for SMBs

Implementing algorithmic fairness doesn’t require a massive overhaul or a team of data scientists. For SMBs, practicality is paramount. The key lies in adopting a phased, incremental approach, focusing on readily accessible tools and strategies. The first step involves awareness and education.

SMB owners and employees need to understand what algorithmic bias Meaning ● Algorithmic bias in SMBs: unfair outcomes from automated systems due to flawed data or design. is, where it can creep into their operations, and why it’s relevant to their business. This can be achieved through readily available online resources, workshops, or even informal discussions within the team. The next step is to identify areas where algorithms are currently being used or are planned for future implementation. This could range from simple tools like social media ad targeting to more complex systems like inventory management or customer relationship management (CRM) platforms with AI features.

Once these areas are identified, SMBs can begin to assess potential fairness risks and implement mitigation strategies. This assessment doesn’t need to be overly technical; it can start with asking critical questions about the data being used, the algorithm’s potential impact on different customer segments, and the mechanisms in place to detect and correct unfair outcomes.

Algorithmic fairness for SMBs begins not with complex code audits, but with a simple commitment to ethical business practices extended into the digital realm.

Accessible Tools and Techniques

Several readily available tools and techniques can aid SMBs in their journey toward algorithmic fairness. For instance, when using off-the-shelf AI software, SMBs should prioritize platforms that offer transparency Meaning ● Operating openly and honestly to build trust and drive sustainable SMB growth. and explainability features. These features, often built into reputable software, provide insights into how algorithms arrive at their decisions, making it easier to identify potential biases. Furthermore, SMBs can leverage readily available datasets and benchmarks to test their algorithms for fairness.

Numerous open-source fairness toolkits and libraries offer user-friendly interfaces and pre-built metrics to assess bias in various algorithmic applications. These tools often require minimal coding expertise and can be integrated into existing workflows. Another practical technique is to diversify data inputs. If an SMB is training its own algorithms, actively seeking out and incorporating data from diverse sources can help mitigate bias.

This might involve expanding customer surveys to include a wider range of demographics or actively seeking feedback from underrepresented customer segments. Finally, establishing clear feedback loops is crucial. Regularly monitoring algorithm outputs, soliciting feedback from users, and creating channels for reporting unfair outcomes allows SMBs to detect and address biases proactively. This iterative process of assessment, mitigation, and feedback is key to embedding fairness into algorithmic operations within an SMB context.

The Human Element

Algorithms, regardless of their sophistication, are ultimately tools wielded by humans. Therefore, the human element is paramount in ensuring algorithmic fairness within SMBs. This starts with fostering a culture of ethical awareness throughout the organization. Employees at all levels should be encouraged to think critically about the potential fairness implications of algorithms they use or encounter.

Regular training sessions, even brief ones, can help instill this mindset. Furthermore, assigning responsibility for algorithmic fairness to a specific individual or team, even on a part-time basis, can create accountability and ensure that fairness considerations are not overlooked. This individual or team can act as a point of contact for fairness-related concerns, oversee fairness assessments, and champion ethical algorithmic practices within the SMB. Crucially, human oversight should be integrated into algorithmic decision-making processes.

Algorithms should augment, not replace, human judgment. In critical areas like hiring or customer service, algorithms can provide valuable insights and recommendations, but the final decision should always rest with a human, who can consider contextual factors and ethical implications that algorithms might miss. This human-in-the-loop approach is essential for preventing algorithmic bias from translating into real-world unfairness within SMB operations.

Starting Small, Thinking Big

For SMBs, the journey toward algorithmic fairness is best approached incrementally. Trying to implement a comprehensive fairness framework overnight is unrealistic and likely to be overwhelming. Instead, SMBs should focus on starting small, perhaps by addressing fairness in one key algorithmic application first, such as customer segmentation or marketing automation. Once they gain experience and build internal capacity, they can gradually expand their fairness efforts to other areas.

This phased approach allows SMBs to learn, adapt, and integrate fairness considerations into their operations in a sustainable manner. However, starting small shouldn’t preclude thinking big. Even as SMBs take initial steps, they should keep the long-term vision of algorithmic fairness in mind. This includes considering how fairness can become a competitive advantage, how it can enhance customer trust and loyalty, and how it aligns with the SMB’s overall ethical values. By starting small but thinking big, SMBs can embark on a practical and impactful journey toward implementing algorithmic fairness, transforming it from an abstract concept into a tangible business asset.

The initial foray into algorithmic fairness might feel like uncharted territory for many SMBs. Yet, it represents a crucial step in responsible business evolution. By embracing these fundamental principles and practical steps, SMBs can not only mitigate the risks of algorithmic bias but also unlock new opportunities for growth, customer loyalty, and ethical leadership within their communities.

The journey towards fairness is not a destination, but a continuous process of learning, adaptation, and commitment. And for SMBs, this journey begins with the simple recognition that fairness, in all its forms, is not just the right thing to do, it is the smart thing to do.

Intermediate

The narrative surrounding algorithmic fairness often casts SMBs as passive recipients of AI technologies developed elsewhere, overlooking their potential agency in shaping a fairer algorithmic landscape. This perspective diminishes the strategic leverage SMBs possess, particularly their close proximity to customers and their inherent agility in adapting to evolving ethical expectations. While resource constraints are undeniable, framing algorithmic fairness as an insurmountable barrier for SMBs is a self-defeating prophecy. Instead, a more strategic approach recognizes that SMBs can transform these perceived limitations into advantages, leveraging their unique position to implement fairness in ways that are both practical and strategically beneficial.

Strategic Integration of Fairness

For SMBs, algorithmic fairness should not be treated as a separate compliance exercise but rather as an integral component of their overall business strategy. This integration begins with aligning fairness objectives with core business goals. For instance, if an SMB’s growth strategy hinges on expanding into diverse customer segments, ensuring algorithmic fairness in marketing and customer service becomes paramount. Algorithms that inadvertently discriminate against certain demographics will directly undermine this growth objective.

Conversely, algorithms designed with fairness in mind can facilitate inclusive growth, enabling SMBs to reach and serve a wider customer base effectively. Furthermore, integrating fairness into automation Meaning ● Automation for SMBs: Strategically using technology to streamline tasks, boost efficiency, and drive growth. initiatives is crucial. As SMBs increasingly adopt automation to enhance efficiency, they must ensure that these automated systems do not perpetuate or amplify existing biases. Fairness-aware automation can lead to more equitable and efficient processes, enhancing both operational performance and ethical standing. Strategic integration, therefore, positions algorithmic fairness not as a cost center but as a value driver, contributing directly to SMB growth and operational excellence.

Navigating the Algorithmic Supply Chain

SMBs rarely develop algorithms from scratch; they typically rely on off-the-shelf software, cloud-based AI services, or pre-trained models. This reliance on an algorithmic supply chain Meaning ● Algorithmic Supply Chain uses smart programs to automate and optimize SMB operations for better decisions. presents both challenges and opportunities for fairness implementation. The challenge lies in the limited transparency often associated with these external algorithmic products. SMBs may lack visibility into the data and processes used to train these algorithms, making it difficult to assess their fairness characteristics.

However, this challenge can be mitigated by adopting a proactive approach to vendor selection and due diligence. SMBs should prioritize vendors who demonstrate a commitment to algorithmic fairness and transparency. This can involve asking vendors specific questions about their fairness assessment processes, requesting access to fairness documentation, and seeking certifications or audits related to algorithmic ethics. Furthermore, SMBs can leverage their collective bargaining power, albeit informal, to demand greater fairness and transparency from algorithmic providers.

By voicing their concerns and preferences, SMBs can collectively influence the algorithmic supply chain, pushing vendors to prioritize fairness in their product development. This proactive navigation of the algorithmic supply chain transforms SMBs from passive consumers to active participants in shaping a fairer AI ecosystem.

Data Governance and Fairness Audits

Robust data governance Meaning ● Data Governance for SMBs strategically manages data to achieve business goals, foster innovation, and gain a competitive edge. is foundational for algorithmic fairness. For SMBs, this doesn’t necessitate complex data management systems but rather a pragmatic approach to data collection, storage, and usage. This begins with understanding the data that fuels their algorithms. SMBs should inventory their data sources, assess their representativeness, and identify potential biases embedded within the data.

For instance, customer data collected primarily through online channels might underrepresent offline customers, potentially skewing algorithms trained on this data. Implementing data diversification strategies, such as actively seeking data from underrepresented sources, can mitigate this bias. Furthermore, establishing clear data usage policies is crucial. These policies should outline ethical guidelines for data collection, processing, and algorithmic application, ensuring that data is used responsibly and fairly.

In addition to data governance, periodic fairness audits are essential. These audits, which can be conducted internally or with external expertise, involve systematically assessing algorithms for potential biases and unfair outcomes. Fairness audits are not about achieving perfect fairness, which is often an elusive goal, but about identifying and mitigating unacceptable levels of bias. For SMBs, fairness audits can be streamlined and cost-effective, focusing on high-impact algorithmic applications and leveraging readily available fairness metrics and tools. These audits provide valuable insights into algorithmic performance and guide iterative improvements toward greater fairness.

Strategic SMBs view algorithmic fairness not as a regulatory burden, but as a competitive edge in an increasingly ethically conscious marketplace.

Employee Empowerment and Fairness Advocacy

Employees are often the first line of defense in identifying and addressing algorithmic fairness issues within SMBs. Empowering employees to become fairness advocates is a powerful and often underutilized strategy. This empowerment begins with providing employees with the necessary training and resources to understand algorithmic bias and its implications. Workshops, online modules, and internal knowledge sharing sessions can equip employees with the foundational knowledge to recognize fairness issues in their daily work.

Furthermore, creating channels for employees to report fairness concerns is crucial. This could involve establishing a dedicated fairness feedback mechanism, incorporating fairness discussions into regular team meetings, or appointing fairness champions within different departments. When employees feel empowered to voice their concerns and contribute to fairness solutions, it fosters a culture of ethical responsibility throughout the SMB. This employee-driven fairness advocacy can be particularly effective in SMBs due to their flatter organizational structures and closer employee-management relationships.

Employee insights, often grounded in direct customer interactions and operational realities, can provide invaluable perspectives for identifying and mitigating algorithmic biases. This bottom-up approach to fairness complements top-down strategic initiatives, creating a holistic and impactful fairness ecosystem within the SMB.

Collaboration and Industry Standards

SMBs are not isolated entities; they operate within broader industry ecosystems. Collaboration with industry peers and participation in industry-wide fairness initiatives can amplify SMBs’ individual fairness efforts and contribute to collective progress. Industry associations, SMB networks, and online communities can serve as platforms for sharing best practices, exchanging fairness resources, and collectively addressing industry-specific fairness challenges. For instance, SMBs in the e-commerce sector could collaborate to develop fairness guidelines for product recommendation algorithms, addressing biases related to product visibility or pricing.

Furthermore, engaging with emerging industry standards and frameworks for algorithmic fairness can provide SMBs with valuable guidance and benchmarks. Organizations like the IEEE and ISO are developing standards related to AI ethics and fairness, which SMBs can adapt and implement within their operations. These standards provide a common language and framework for discussing and addressing fairness issues, facilitating collaboration and interoperability across the industry. By actively participating in collaborative initiatives and adopting industry standards, SMBs can not only enhance their own fairness practices but also contribute to raising the bar for algorithmic fairness across their respective industries. This collective action is essential for creating a truly fair and responsible algorithmic landscape.

Measuring and Communicating Fairness

Demonstrating a commitment to algorithmic fairness requires not only implementing fairness practices but also measuring and communicating these efforts effectively. For SMBs, this doesn’t necessitate complex fairness metrics but rather a pragmatic approach to tracking and reporting fairness outcomes. This begins with defining measurable fairness indicators relevant to their specific algorithmic applications. For instance, in a hiring algorithm, fairness indicators might include representation rates of different demographic groups in interview shortlists or hiring decisions.

In a loan application algorithm, fairness indicators could focus on approval rates across different socioeconomic groups. Regularly tracking these indicators allows SMBs to monitor algorithmic performance and identify potential fairness gaps. Furthermore, transparently communicating fairness efforts to stakeholders is crucial. This communication can take various forms, from publishing fairness statements on their websites to incorporating fairness discussions into customer communications or investor reports.

Transparency builds trust and demonstrates accountability, enhancing the SMB’s reputation as an ethical and responsible business. However, communication should be balanced with caution. Overpromising or making unsubstantiated fairness claims can backfire, eroding stakeholder trust. Authentic and transparent communication, grounded in measurable fairness indicators and ongoing improvement efforts, is key to building credibility and demonstrating a genuine commitment to algorithmic fairness.

Moving beyond basic awareness, SMBs at the intermediate level can strategically embed algorithmic fairness into their operations, transforming it from a reactive measure into a proactive business advantage. By navigating the algorithmic supply chain, governing their data responsibly, empowering their employees, and collaborating with industry peers, SMBs can not only mitigate fairness risks but also unlock new opportunities for innovation, customer loyalty, and sustainable growth. The intermediate stage is about shifting from a compliance mindset to a strategic mindset, recognizing that algorithmic fairness is not just an ethical imperative but a core element of responsible and successful business in the algorithmic age. This strategic approach positions SMBs as leaders, not followers, in the evolving landscape of AI ethics.

Advanced

The prevailing discourse on algorithmic fairness often presupposes a static definition of fairness, neglecting the dynamic and context-dependent nature of ethical considerations within diverse SMB ecosystems. This static perspective limits the potential for truly impactful fairness interventions, particularly for advanced SMBs seeking to leverage algorithmic technologies for transformative growth and societal impact. Moving beyond simplistic notions of equal outcomes, advanced SMBs must embrace a more sophisticated, multi-dimensional understanding of fairness, recognizing that fairness is not a monolithic concept but a spectrum of ethical values that must be continuously negotiated and adapted within specific business contexts and evolving societal norms. This advanced perspective positions algorithmic fairness not as a technical problem to be solved, but as an ongoing ethical dialogue to be cultivated.

Contextualizing Fairness for SMB Innovation

For advanced SMBs, algorithmic fairness transcends generic ethical guidelines; it becomes deeply contextualized and integrated into their innovation processes. This contextualization begins with recognizing that fairness considerations are not universal but are shaped by the specific industry, customer base, and societal context in which the SMB operates. For instance, fairness in a healthcare AI application might prioritize different ethical dimensions than fairness in a marketing algorithm. Advanced SMBs engage in rigorous ethical analysis to identify the most salient fairness dimensions for their specific algorithmic innovations.

This analysis involves stakeholder engagement, ethical risk assessments, and consideration of potential societal impacts. Furthermore, contextualization extends to the design and development of algorithms themselves. Advanced SMBs move beyond off-the-shelf fairness solutions and develop custom fairness-aware algorithms tailored to their specific business needs and ethical priorities. This might involve incorporating fairness constraints directly into algorithm design, utilizing fairness-enhancing machine learning techniques, or developing explainable AI models that facilitate ethical scrutiny and accountability. Contextualized fairness becomes a core design principle, driving innovation in a responsible and ethically grounded manner.

Dynamic Fairness and Adaptive Algorithms

Algorithmic fairness is not a fixed state; it is a dynamic concept that must adapt to evolving societal norms, changing business contexts, and emerging ethical challenges. Advanced SMBs recognize this dynamism and develop adaptive algorithmic systems that can continuously learn and adjust to maintain fairness over time. This adaptability requires moving beyond static fairness metrics and embracing dynamic fairness evaluation frameworks. These frameworks incorporate real-time feedback loops, continuous monitoring of algorithmic performance, and mechanisms for detecting and responding to fairness drift ● the phenomenon where algorithms become less fair over time due to shifts in data or societal expectations.

Adaptive algorithms might incorporate fairness-aware reinforcement learning techniques, allowing them to learn fairness preferences from user interactions or ethical feedback signals. Furthermore, advanced SMBs establish robust governance mechanisms for overseeing algorithmic adaptation, ensuring that fairness adjustments are ethically sound and aligned with evolving societal values. Dynamic fairness transforms algorithms from static decision-making tools into continuously evolving ethical agents, capable of maintaining fairness in a constantly changing world. This adaptive approach is crucial for long-term algorithmic responsibility and sustainability.

Intersectionality and Multi-Dimensional Fairness

Traditional fairness approaches often focus on single dimensions of fairness, such as fairness with respect to gender or race. However, real-world fairness challenges are often intersectional, involving complex interactions between multiple dimensions of identity and social vulnerability. Advanced SMBs embrace a multi-dimensional understanding of fairness, recognizing that fairness must be evaluated and addressed across multiple intersecting dimensions simultaneously. This intersectional approach requires developing fairness metrics and evaluation frameworks that can capture the complex interplay of different fairness dimensions.

For instance, an SMB might need to consider fairness not just with respect to race and gender separately, but also with respect to the intersection of race and gender, recognizing that women of color might experience unique forms of algorithmic bias. Furthermore, addressing intersectional fairness requires developing algorithmic techniques that can mitigate bias across multiple dimensions simultaneously. This might involve using fairness-aware optimization algorithms that balance fairness across different demographic groups or developing algorithmic debiasing techniques that address intersectional biases in data and models. Multi-dimensional fairness moves beyond simplistic notions of group fairness and embraces the complexity of real-world social inequalities, striving for fairness for all, particularly those at the intersections of multiple forms of disadvantage.

Advanced SMBs view algorithmic fairness as a continuous ethical dialogue, not a technical problem, shaping their innovation and societal impact.

Algorithmic Transparency and Radical Explainability

Transparency is often touted as a key principle of algorithmic fairness, but traditional transparency approaches often fall short of providing truly meaningful insights into algorithmic decision-making, particularly for complex AI systems. Advanced SMBs push beyond basic transparency and embrace radical explainability, striving for algorithmic systems that are not only transparent but also deeply understandable and auditable by diverse stakeholders. This radical explainability requires developing AI models that are inherently interpretable, such as rule-based systems or inherently explainable neural networks. Furthermore, it involves developing user-friendly interfaces and visualization tools that allow non-technical stakeholders to understand how algorithms arrive at their decisions, identify potential biases, and provide meaningful feedback.

Radical explainability also extends to data transparency, providing stakeholders with insights into the data used to train algorithms, including data provenance, data biases, and data limitations. This level of transparency empowers stakeholders to engage in informed ethical scrutiny of algorithmic systems, fostering accountability and trust. Algorithmic audits, conducted by independent third parties, become crucial for verifying radical explainability and ensuring that algorithmic systems are truly transparent and accountable. Radical explainability transforms algorithms from black boxes into glass boxes, fostering a culture of transparency and ethical accountability.

Fairness as a Competitive Differentiator and Societal Value

For advanced SMBs, algorithmic fairness is not merely a cost of doing business or a risk mitigation strategy; it becomes a powerful competitive differentiator and a source of societal value creation. In an increasingly ethically conscious marketplace, consumers, investors, and employees are increasingly valuing businesses that demonstrate a genuine commitment to 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. Advanced SMBs leverage their fairness leadership to attract and retain customers who prioritize ethical consumption, investors who seek socially responsible investments, and employees who are motivated by purpose-driven work. Fairness becomes a core element of their brand identity and value proposition, differentiating them from competitors who lag behind in ethical AI adoption.

Furthermore, advanced SMBs recognize that algorithmic fairness can contribute to broader societal value creation. By developing and deploying fair algorithms, they can help mitigate societal biases, promote equity and inclusion, and contribute to a more just and equitable algorithmic society. This societal value creation not only enhances their brand reputation but also aligns with their long-term sustainability goals and ethical mission. Fairness transforms from a compliance burden into a strategic asset, driving both business success and positive societal impact. This value-driven approach positions advanced SMBs as ethical leaders in the algorithmic age, shaping a fairer and more responsible future for AI.

Ethical AI Governance and Accountability Frameworks

Implementing advanced algorithmic fairness requires robust ethical AI governance Meaning ● Ethical AI Governance for SMBs: Responsible AI use for sustainable growth and trust. and accountability frameworks. For advanced SMBs, this goes beyond basic compliance policies and involves establishing comprehensive governance structures, ethical review processes, and accountability mechanisms that ensure ongoing ethical oversight of algorithmic systems. This begins with establishing a dedicated ethical AI committee or board, composed of diverse stakeholders, including ethicists, fairness experts, and representatives from affected communities. This committee is responsible for setting ethical AI guidelines, overseeing fairness assessments, and providing ethical guidance on algorithmic innovation.

Furthermore, advanced SMBs implement rigorous ethical review processes for all new algorithmic applications, ensuring that fairness considerations are integrated into the design, development, and deployment phases. These review processes might involve fairness impact assessments, ethical red teaming exercises, and stakeholder consultations. Accountability mechanisms are crucial for ensuring that ethical AI guidelines are followed and that fairness issues are addressed effectively. This might involve establishing clear lines of responsibility for algorithmic fairness, implementing fairness monitoring and reporting systems, and establishing mechanisms for redress and remediation when unfair outcomes occur. Robust ethical AI governance and accountability frameworks transform algorithmic fairness from an aspirational goal into a concrete and measurable organizational commitment, ensuring ongoing ethical oversight and responsible AI innovation.

At the advanced level, algorithmic fairness becomes a strategic imperative, driving innovation, enhancing competitiveness, and contributing to societal value creation. By contextualizing fairness, embracing dynamic and intersectional approaches, pursuing radical explainability, and establishing robust governance frameworks, advanced SMBs can not only mitigate fairness risks but also unlock new opportunities for ethical leadership and transformative impact in the algorithmic age. The advanced stage is about moving beyond reactive fairness measures and proactively shaping a fairer and more responsible algorithmic future, positioning SMBs as pioneers in ethical AI innovation and societal progress. This proactive and value-driven approach defines the next frontier of algorithmic fairness for SMBs, transforming it from a challenge into a catalyst for positive change.

References

• O’Neil, Cathy. Weapons of Math Destruction ● How Big Data Increases Inequality and Threatens Democracy. Crown, 2016.
• Barocas, Solon, et al. Fairness and Machine Learning ● Limitations and Opportunities. MIT Press, 2019.
• 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, et al. “A Survey on Bias and Fairness in Machine Learning.” ACM Computing Surveys (CSUR), vol. 54, no. 6, 2021, pp. 1-35.