Meaning and Inference
Inferential approaches to meaning, especially proof-theoretic semantics and the role of inference in determining logical content across classical, intuitionistic, and substructural logics.
New Frontiers Fellow, Southampton · Honorary Research Fellow, UCL
Logician studying the formal foundations of reasoning — how meaning arises from inference, how proofs structure thought, and what these foundations reveal about artificial intelligence.
Gheorghiu's research centres on proof-theoretic semantics: the view that logical and linguistic meaning is determined by inference rules rather than truth conditions. This has deep implications for the philosophy of logic, the nature of mathematical proof, and the question of whether AI systems can genuinely reason.
The Research
Proof-theoretic semantics is a programme in the philosophy of logic that takes inference — rather than truth or reference — as the basic semantic concept. Where classical semantics asks what propositions are true, proof-theoretic semantics asks what follows from what, and why. The meaning of a logical constant such as “and” or “if…then” is given entirely by the rules governing its use in proofs.
This approach, developed in the tradition of Gentzen, Dummett, and Prawitz, has consequences that reach well beyond mathematical logic. It offers a new account of how language acquires meaning, a rigorous framework for evaluating the coherence of logical systems, and a principled basis for asking what genuine reasoning requires — a question that is increasingly urgent as AI systems are asked to explain, justify, and advise.
A central claim of Gheorghiu's work is that the gap between statistical prediction and genuine reasoning is not merely philosophical but formally precise: a system that cannot trace its outputs back to principled inferential steps cannot, in any robust sense, be said to reason. Understanding this distinction is essential for building AI systems that are not just powerful, but interpretable and trustworthy.
Research Themes
Inferential approaches to meaning, especially proof-theoretic semantics and the role of inference in determining logical content across classical, intuitionistic, and substructural logics.
The relationship between formal reasoning, explanation, and artificial intelligence: what separates genuine reasoning from prediction, and how logic can make AI systems interpretable and accountable.
The mathematical structure of proofs and their applications in computation, distributed systems, and automated reasoning — proof search, reductive logic, and the algebraic constraints that shape proof systems.
Logical and formal methods for representing policies, rules, and institutional structures — including governance and regulation in AI-enabled environments.
Selected Publications
See the full list on the Publications page.
Public Writing & Engagement
Gheorghiu writes about logic, mathematics, and minds — artificial and otherwise — for readers outside academia. Recent pieces have appeared in the Times Literary Supplement, Mathematics Today, and The Conversation. His 2025 essay on AI and school mathematics received the Graham Hoare Prize from the Institute of Mathematics and its Applications.
See the Writing page for articles, talks, and media.
PhD Supervision
Gheorghiu welcomes enquiries from prospective PhD students interested in proof-theoretic semantics, philosophy of logic, logic and AI, formal methods, automated reasoning, and related areas. Students from interdisciplinary backgrounds are especially encouraged to get in touch.
Enquiries to alexvgheo@gmail.com or the Contact page.