Measuring the Machine: Evaluating Generative AI as Pluralist Sociotechnical Systems
Summary: arXiv:2604.20545v1 Announce Type: new
Abstract: In measurement theory, instruments do not simply record reality; they help constitute what is observed. The same holds for generative AI evaluation: benchmarks do not just measure, they shape what models appear to be. Functionalist benchmarks treat models as isolated predictors, while prescriptive approaches assess what systems ought to be. Both obscure the sociotechnical processes through which meaning and values are enacted, risking the reification of narrow cultural perspectives in pluralist contexts.
This thesis advances a descriptive alternative. It argues that generative AI must be evaluated as a pluralist sociotechnical system and develops Machine-Society-Human (MaSH) Loops, a framework for tracing how models, users, and institutions recursively co-construct meaning and values. Evaluation shifts from judging outputs to examining how values are enacted in interaction.
Key Contributions
- Conceptual Contribution: MaSH Loops reframes evaluation as a recursive, enactive process.
- Methodological Contribution: The World Values Benchmark introduces a distributional approach grounded in World Values Survey data, structured prompt sets, and anchor-aware scoring.
- Empirical Contribution: The thesis demonstrates these concepts through two cases: value drift in early GPT-3 and sociotechnical evaluation in real estate.
A final chapter draws on participatory realism to argue that prompting and evaluation are constitutive interventions, not neutral observations. This perspective challenges the traditional view that benchmarks and evaluations are merely tools for measurement.
Challenges of Static Benchmarks
The thesis argues that static benchmarks are insufficient for generative AI. Responsible evaluation requires pluralist, process-oriented frameworks that make visible whose values are enacted. Traditional methods often ignore the complex interactions and cultural contexts that shape AI systems, leading to a skewed understanding of their implications.
The Role of Evaluation in Governance
Evaluation is therefore a site of governance, shaping how AI systems are understood, deployed, and trusted. By recognizing the sociotechnical dimensions of generative AI, stakeholders can better navigate the ethical and practical challenges that arise in development and implementation.
In summary, the thesis provides a comprehensive framework for evaluating generative AI that emphasizes the importance of context, interaction, and the values embedded within the systems. This pluralist approach not only enhances our understanding of AI technologies but also promotes more responsible and inclusive practices in their evaluation and deployment.