Автоматизация науки. Концептуальный взгляд

Вестник Томского государственного университета. Философия. Социология. Политология. 2022. № 65. С. 37–50.

Kham T.

The problems of automation of research activities are considered, primarily from the perspective of the epistemic potential of gaining new scientific knowledge without the participation of a human as a subject of science. The bases of automation are considered from the standpoint of jointly working methodological approaches to cognition: empiricism and logical positivism. The author shows that scientific empiricism and positivism, the structural tissue in the intellectual Western thought, predetermine foundation of science automation. He analyzes the reasons for the need to eliminate the subject in mathematics and science through the hierarchy of a formal-axiomatic approach, a semantic view, and computational technologies. The current state of mathematics is shown as a confrontation between two opposite approaches: axiomatic and analytical. The axiomatic approach allows and stimulates the automation of science and the elimination of a human as a subject of science. The analytical approach defends the position of the fundamental impossibility of automating science. The author evaluates the role of a human in the new conditions of scientific activity, in which the main part of epistemic categories is delegated to computational algorithms in the system of automated science. These categories are based on Mieke Boon's list, which he compiled while investigating the impact of machine learning on the automation of science.

Keywords: искусственный интеллект, машинное обучение, эпистемология, эвристика, компьютерные симуляции, автоматизация науки