Over the last few hundred years, best practice in some fields of human action—e.g., the treatment of heart disease, the transportation of persons, goods, and messages, and the destruction of landscapes, structures, and lives—has become dramatically more effective. At the same time, best practice in other fields, e.g., the amelioration of poverty or the teaching of reading, writing, or math, has improved more slowly, if at all. I argue that practice and technology (“know-how”) can only improve rapidly under rather special conditions: that, at any given point in time, some fields are more “progressible” than others.I articulate a conceptual framework describing several characteristics of practice in a field that may facilitate rapid progress. These characteristics, while not fixed, tend to remain fairly stable for long periods of time. I argue that know-how can improve more quickly 1) when offline “vicarious trial” of variations in practice is feasible and useful; 2) when practice is formal and standardized; 3) when practice is substantially performed by artifacts rather than by humans; 4) when outcomes of variations in practice may be rapidly evaluated; 5) when goals of practice are consistently agreed upon; 6) when contexts and objects of practice may be treated as, or have been made, consistent for the purposes of intervention; 7) when components of task systems are not heavily interdependent; and 8) when labor is finely and sharply divided. I illustrate and elaborate this framework through comparative case studies on efforts to improve practice in three differentially “progressible” fields. I examine rapid improvement in a COVID-19 testing lab, inconsistent improvement in undergraduate algebra instruction, and ambiguous improvement in regional water modeling to support municipal water management. These cases indicate that my theory may inform judgments about the plausibility of rapid advance within a field of practice, absent disruptive change in methods or problem formulation. My theory may also shed light on which varieties of innovative effort may and may not foreseeably contribute to improving practice in a given field—more formal, theoretical, and context-independent work in high-progressibility domains, more tacit, grounded, and localized work in low-progressibility ones.

While artificial intelligence (AI) has seen enormous technical progress in recent years, less progress has occurred in understanding the governance issues raised by AI. In this dissertation, I make four contributions to the study and practice of AI governance. First, I connect AI to the literature and practices of responsible research and innovation (RRI) and explore their applicability to AI governance. I focus in particular on AI’s status as a general purpose technology (GPT), and suggest some of the distinctive challenges for RRI in this context such as the critical importance of publication norms in AI and the need for coordination. Second, I provide an assessment of existing AI governance efforts from an RRI perspective, synthesizing for the first time a wide range of literatures on AI governance and highlighting several limitations of extant efforts. This assessment helps identify areas for methodological exploration. Third, I explore, through several short case studies, the value of three different RRI-inspired methods for making AI governance more anticipatory and reflexive: expert elicitation, scenario planning, and formal modeling. In each case, I explain why these particular methods were deployed, what they

produced, and what lessons can be learned for improving the governance of AI in the future. I find that RRI-inspired methods have substantial potential in the context of AI, and early utility to the GPT-oriented perspective on what RRI in AI entails. Finally, I describe several areas for future work that would put RRI in AI on a sounder footing.

Science can help inform policy decisions by providing information on the risks and benefits of a technology. In the field of nanotechnology, which is characterized by high degree of complexity and uncertainty, there are high demands for scientists to take an active role in policy debates with regulators, policy-makers and the public. In particular, policy-makers often rely on scientific experts to help them make decisions about regulations. However, scientists’ perceptions about policy and public engagement vary based on their individual characteristics, values, and backgrounds. Although many policy actors are involved in nanotechnology policy process, there are few empirical studies that focus on the establishment of coalitions and their impact on policy outputs, as well as the role of scientists in the coalitions. Also, while the Environmental Protection Agency (EPA) has regulatory authority over nanoscale materials, there is a lack of literature that describes the use of science on EPA’s decision making of nanotechnology.

In this dissertation, these research gaps are addressed in three essays that explore the following research questions: (1) how are nano-scientists’ individual characteristics and values associated with their perceptions of public engagement and political involvement? (2) how can the Advocacy Coalition Framework (ACF) can be applied to nanotechnology policy subsystem? and (3) how does the EPA utilize science when making regulatory decisions about nanotechnology? First, using quantitative data from a 2011 mail survey of elite U.S. nanoscientists, the dissertation shows that scientists are supportive of engaging with policy-makers and the public about their results. However, there are differences among scientists based on their individual characteristics. Second, qualitative interview analysis suggests that there are two opposing advocacy groups with shared beliefs in the nanotechnology policy subsystem. The lineup of coalition members is stable over time, while the EPA advocates less consistent positions. The interview data also show a significant role of scientific information in the subsystem. Third, the dissertation explains the EPA’s internal perspective about the use of science in regulatory decision making for nanotechnology. The dissertation concludes with some lessons that are applicable for policy-making for emerging technologies.

Several prominent research strategy organizations recommend applying life cycle assessment (LCA) early in the development of emerging technologies. For example, the US Environmental Protection Agency, the National Research Council, the Department of Energy, and the National Nanotechnology Initiative identify the potential for LCA to inform research and development (R&D) of photovoltaics and products containing engineered nanomaterials (ENMs). In this capacity, application of LCA to emerging technologies may contribute to the growing movement for responsible research and innovation (RRI). However, existing LCA practices are largely retrospective and ill-suited to support the objectives of RRI. For example, barriers related to data availability, rapid technology change, and isolation of environmental from technical research inhibit application of LCA to developing technologies. This dissertation focuses on development of anticipatory LCA tools that incorporate elements of technology forecasting, provide robust explorations of uncertainty, and engage diverse innovation actors in overcoming retrospective approaches to environmental assessment and improvement of emerging technologies. Chapter one contextualizes current LCA practices within the growing literature articulating RRI and identifies the optimal place in the stage gate innovation model to apply LCA. Chapter one concludes with a call to develop anticipatory LCA – building on the theory of anticipatory governance – as a series of methodological improvements that seek to align LCA practices with the objectives of RRI.

Chapter two provides a framework for anticipatory LCA, identifies where research from multiple disciplines informs LCA practice, and builds off the recommendations presented in the preceding chapter. Chapter two focuses on crystalline and thin film photovoltaics (PV) to illustrate the novel framework, in part because PV is an environmentally motivated technology undergoing extensive R&D efforts and rapid increases in scale of deployment. The chapter concludes with a series of research recommendations that seek to direct PV research agenda towards pathways with the greatest potential for environmental improvement.

Similar to PV, engineered nanomaterials (ENMs) are an emerging technology with numerous potential applications, are the subject of active R&D efforts, and are characterized by high uncertainty regarding potential environmental implications. Chapter three introduces a Monte Carlo impact assessment tool based on the toxicity impact assessment model USEtox and demonstrates stochastic characterization factor (CF) development to prioritize risk research with the greatest potential to improve certainty in CFs. The case study explores a hypothetical decision in which personal care product developers are interested in replacing the conventional antioxidant niacinamide with the novel ENM C60, but face high data uncertainty, are unsure regarding potential ecotoxicity impacts associated with this substitution, and do not know what future risk-relevant experiments to invest in that most efficiently improve certainty in the comparison. Results suggest experiments that elucidate C60 partitioning to suspended solids should be prioritized over parameters with little influence on results. This dissertation demonstrates a novel anticipatory approach to exploration of uncertainty in environmental models that can create new, actionable knowledge with potential to guide future research and development decisions.

This dissertation explores the megamachine, a prominent metaphor in American humanist and philosopher of technology, Lewis Mumford's Myth of the Machine series. The term refers critically to dynamic, regimented human capacities that drive scientific and technical innovation in society. Mumford's view of the nature of collectives focuses on qualities and patterns that emerge from the behavior of groups, societies, systems, and ecologies. It is my aim to reenergize key concepts about collective capacities drawn from Lewis Mumford's critique of historical and modern sociotechnical arrangements. I investigate the possibility of accessing those capacities through improved design for Technology Assessment (TA), formal practices that engage experts and lay citizens in the evaluation of complex scientific and technical issues.

I analyze the components of Mumford's megamachine and align key concerns in two pivotal works that characterize the impact of collective capacities on society: Bruno Latour's Pasteurization of France (1988) and Elias Canetti's Crowds and Power (1962). As I create a model of collective capacities in the sociotechnical according to the parameters of Mumford's megamachine, I rehabilitate two established ideas about the behavior of crowds and about the undue influence of technological systems on human behavior. I depart from Mumford's tactics and those of Canetti and Latour and propose a novel focus for STS on "sociotechnical crowds" as a meaningful unit of social measure. I make clear that Mumford's critique of the sociotechnical status quo still informs the conditions for innovation today.

Using mixed mode qualitative methods in two types of empirical field studies, I then investigate how a focus on the characteristics and components of collective human capacities in sociotechnical systems can affect the design and performance of TA. I propose a new model of TA, Emergent Technology Assessment (ETA), which includes greater public participation and recognizes the interrelationship among experience, affect and the material in mediating the innovation process. The resulting model -- the "soft" megamachine --introduces new strategies to build capacity for responsible innovation in society.

Since its inception in 1973, the Endangered Species Act has been met with both praise and criticism. More than 40 years later, the Act is still polarizing, with proponents applauding its power to protect species and critics arguing against its perceived ineffectiveness and potential mismanagement. Recovery plans, which were required by the 1988 amendments to the Act, play an important role in organizing efforts to protect and recover species under the Act. In 1999, in an effort to evaluate the process, the Society for Conservation Biology commissioned an independent review of endangered species recovery planning. From these findings, the SCB made key recommendations for how management agencies could improve the recovery planning process, after which the Fish and Wildlife Service and the National Marine Fisheries Service redrafted their recovery planning guidelines. One important recommendation called for recovery plans to make threats a primary focus, including organizing and prioritizing recovery tasks for threat abatement. Here, I seek to determine the extent to which SCB recommendations were incorporated into these new guidelines, and if, in turn, the recommendations regarding threats manifested in recovery plans written under the new guidelines. I found that the guidelines successfully incorporated most SCB recommendations, except those that addressed monitoring. As a result, recent recovery plans have improved in their treatment of threats, but still fail to adequately incorporate threat monitoring. This failure suggests that developing clear guidelines for monitoring should be an important priority in future ESA recovery planning.

Life cycle assessment (LCA) is a powerful framework for environmental decision making because the broad boundaries called for prevent shifting of burden from one life-cycle phase to another. Numerous experts and policy setting organizations call for the application of LCA to developing nanotechnologies. Early application of LCA to nanotechnology may identify environmentally problematic processes and supply chain components before large investments contribute to technology lock in, and thereby promote integration of environmental concerns into technology development and scale-up (enviro-technical integration). However, application of LCA to nanotechnology is problematic due to limitations in LCA methods (e.g., reliance on data from existing industries at scale, ambiguity regarding proper boundary selection), and because social drivers of technology development and environmental preservation are not identified in LCA. This thesis proposes two methodological advances that augment current capabilities of LCA by incorporating knowledge from technical and social domains. Specifically, this thesis advances the capacity for LCA to yield enviro-technical integration through inclusion of scenario development, thermodynamic modeling, and use-phase performance bounding to overcome the paucity of data describing emerging nanotechnologies. With regard to socio-technical integration, this thesis demonstrates that social values are implicit in LCA, and explores the extent to which these values impact LCA practice and results. There are numerous paths of entry through which social values are contained in LCA, for example functional unit selection, impact category selection, and system boundary definition - decisions which embody particular values and determine LCA results. Explicit identification of how social values are embedded in LCA promotes integration of social and environmental concerns into technology development (socio-enviro-technical integration), and may contribute to the development of socially-responsive and environmentally preferable nanotechnologies. In this way, tailoring LCA to promote socio-enviro-technical integration is a tangible and meaningful step towards responsible innovation processes.