sociotechnical pragmatism

Sociotechnical pragmatism is a practical approach to AI ethics that navigates between two extremes: sociotechnical dogmatism (technology as a panacea) and skepticism (technology as inherently harmful). It posits that an AI system's value and risks are not universal but are defined by its specific social, cultural, and organizational context. Its core tenets are human agency and contextual evaluation, demanding that developers and policymakers assess an AI's real-world impact on all stakeholders. This philosophy directly informs the 'Govern' and 'Map' functions of the NIST AI Risk Management Framework (AI RMF) and aligns with the AI system impact assessment requirements in ISO/IEC 42001 (Annex A.5.2). It encourages moving beyond mere compliance checklists to integrate fairness, transparency, and accountability into the AI lifecycle.

Enterprises can apply sociotechnical pragmatism through three key steps: 1. **Establish a Cross-Functional AI Governance Committee**: Form a team with representatives from legal, ethics, engineering, and business units, as recommended by the NIST AI RMF 'Govern' function. This ensures balanced decision-making. For example, a bank's committee would review an AI lending model for potential bias before deployment. 2. **Conduct Contextual Impact Assessments**: Systematically identify stakeholders and evaluate the AI's potential positive and negative impacts within its specific use case, aligning with ISO/IEC 42001. A retail company, for instance, would assess an AI surveillance system's impact on customer privacy and employee autonomy. 3. **Implement Participatory Design and Continuous Monitoring**: Involve end-users and affected communities in the design process and establish post-deployment feedback loops. A healthcare AI firm that included doctors and patients in development saw a measurable increase in diagnostic accuracy and user trust, leading to higher adoption rates.

Taiwan enterprises face three primary challenges: 1. **Interdisciplinary Talent Gap**: Companies often have strong engineering teams but lack experts in ethics, law, and social sciences needed for holistic risk assessment. The solution is to form an AI ethics board, potentially with external advisors, and initiate cross-functional training programs. 2. **Evolving Regulatory Landscape**: Taiwan's AI-specific regulations are still developing, creating uncertainty. The strategy is to proactively adopt established global standards like the NIST AI RMF or ISO/IEC 42001. This serves as a 'safe harbor' and demonstrates due diligence to international partners. 3. **Difficulty in Quantifying Metrics**: Ethical concepts like 'fairness' and 'trust' are hard to measure. The solution is to use a hybrid approach, combining technical metrics (e.g., statistical bias tests) with procedural metrics (e.g., number of stakeholder consultations, implementation rate of ethics board recommendations). The priority is to establish procedural metrics first while developing relevant quantitative ones.

Winners Consulting specializes in sociotechnical pragmatism for Taiwan enterprises, delivering compliant management systems within 90 days. Free consultation: https://winners.com.tw/contact