In 1962 Thomas Kuhn, an American physicist turned historian of science, published The Structure of Scientific Revolutions, a slim monograph arguing that science does not progress by linear accumulation of facts but advances through long periods of normal science punctuated by revolutionary paradigm shifts in which the very questions, methods, and standards of evaluation change. Kuhn introduced or popularized paradigm, paradigm shift, normal science, revolutionary science, incommensurability, and exemplar into the vocabulary of intellectual life. Within a decade "paradigm shift" had escaped philosophy of science entirely and become one of the most-used and most-misused phrases in business, politics, and culture.
Kuhn's central thesis is that science alternates between normal science and revolutionary science, with paradigms organizing both. A paradigm (Kuhn used the term ambiguously, later distinguishing disciplinary matrix from exemplar) is a constellation of beliefs, values, methods, and exemplary problem-solutions that defines a scientific community, and normal science is puzzle-solving within a paradigm — working out implications, refining measurements, extending the framework, with the paradigm not questioned but used. Anomalies accumulate, are initially explained away (instrumentation problems, special cases, auxiliary-hypothesis refinements), and when they become too central to ignore, crisis sets in, competing paradigms emerge, and the field becomes contentious — established researchers cling to the old paradigm while rising researchers gravitate to alternatives. A paradigm shift occurs when the field as a whole converts, often through generational replacement (Max Planck's "science advances one funeral at a time"). After the shift the old paradigm is not refined but replaced, and Kuhn's canonical examples are the Copernican, chemical (phlogiston → oxygen, Lavoisier 1770s–80s), Newtonian, Darwinian, Einsteinian, and quantum revolutions; his most controversial claim was incommensurability — successive paradigms cannot be evaluated against a paradigm-independent standard, the mass in Newton's laws is not the mass in Einstein's relativity, and paradigm shifts are not strictly cumulative. The standard critiques came swiftly: Karl Popper, Imre Lakatos, and Larry Laudan argued the account was too sociological and not sufficiently normative, that it conflated what scientists do with what they should do, and that it gave too much ground to relativism. Lakatos's research programmes (1970) tried to rescue rationality by analyzing science in terms of progressive versus degenerative programmes; Popper maintained that falsifiability remained the demarcation criterion. Subsequent philosophy of science (Hacking, van Fraassen, Longino) has integrated much of Kuhn while modulating his strongest claims.
Outside philosophy of science, "paradigm shift" has become cultural common currency and largely lost its precision — used to describe everything from product launches to political movements. Within active sciences several recent developments have been called paradigm-shift candidates: the gut microbiome in medicine, epigenetics in genetics, deep learning in AI, amyloid skepticism in Alzheimer's research, the Anthropocene in earth sciences, and the still-pending unification of general relativity with quantum gravity. The largest contemporary candidate may be AI as a paradigm shift in how science is done — foundation models are reshaping practice across many sciences (AlphaFold for protein structure, LLMs for hypothesis generation and literature synthesis, AI-driven materials discovery). The replication crisis in social science has been characterized variously as Kuhnian crisis, normal-science self-correction, or evidence the field needs its first proper paradigm.