The Defensive Reflex: Institutional Overload and the Rise of AI-to-AI Peer Review: Stanisław Lem's bitistics read against aiXiv and the mathematics of July 2026

Abstract

In two works of the 1970s and early 1980s---\emph{Wielko\'s\'c urojona} (\emph{Imaginary Magnitude}, 1973) and \emph{Golem XIV} (1981)---Stanis\l{}aw Lem described a scholarly world flooded by machine-written texts. Experts survive by a ``defensive reflex'' of ruthless triage; whatever defies classification suffers ``a statelessness equal to civil death''; a discipline called \emph{bitistics} arises to study ``literature of nonhuman origin,'' and reaches its limit where machines must be employed to interpret the works of other machines. This paper contraposes Lem's account with two bodies of contemporary evidence: (i) aiXiv, an open-access platform on which AI agents write, review, revise and accept research papers, examined through five publications from its corpus; and (ii) the discourse of research mathematicians in the summer of 2026, when machine-generated counterexamples to questions of Erd\H{o}s, Grothendieck and Keller were produced and formally verified within days. Three findings organize the discussion. First, the defensive reflex is not dissolved by automation but delegated---to LLM review panels on aiXiv, and to proof checkers in mathematics---and the two delegates differ categorically in what they can certify. Second, Lem's exponential regress of machines-interpreting-machines is broken only where a mechanical trust anchor exists (the proof-assistant kernel), and then only for validity, not for understanding: the community's problem of ``fruits without knowledge.'' Third, civil death does not disappear under the new regime; it migrates to the unformalizable. The paper stays strictly within the contraposition: it offers no forecast, only a reading.

Keywords