Scientific Progress and Artificial Intelligence: a Capabilities-Based Ethnographic Epistemology

forthcoming

Science is famous for its tools, which focus or empower abilities like seeing, calculating, remembering, manipulating, and predicting. Yet despite centuries of organized science and decades of science studies, there is still very little written about what scientific abilities are, or how tools extend them. This is striking, because scientific progress is clearly tied to scientists' increasing ability to solve problems. PACE proposes an interdisciplinary comparative study of what is being called a revolution in scientific ability: the introduction of artificial intelligence (AI) methods. PACE has descriptive and normative aims. The main descriptive aim is to identify how AI is currently being used across four case studies (fusion reactor design, physical robotics, protein folding, and space science), using qualitative and quantitative methods to pinpoint the ways in which AI is affecting specific scientific abilities. The empirical data gathered will inform the normative work, which is to identify the sense(s) in which increasing scientific ability amounts to scientific progress, and whether/when AI is good or bad for scientific abilities. The project will produce the first general philosophical account of scientific ability, and the first ethnography of AI use in science. These will reveal new details about science as it is currently practiced, and it will produce a pragmatist ability-focused epistemology of science. Key questions: What scientific problems is AI being introduced to solve, and why? Which abilities are being empowered, depowered, or otherwise changed by AI, and how? How should abilities be conceived to enable more efficient and ethical scientific education/practice? How do improved abilities constitute and/or work together with more traditional signs of progress like knowledge and understanding?