Philosophy of information

Working Definition

The philosophy of information is a discipline that investigates the conceptual nature of information, its dynamics (creation, flow, and processing), and its utilization, while simultaneously applying information-theoretic methods to classic philosophical problems. In Luciano Floridi’s technical formulation, semantic information consists of well-formed, meaningful, and truthful data; falsehood is not “bad information,” but is not information at all. In everyday usage, the term simply refers to philosophical commentary on the internet and digital media; the crucial difference is that Floridi’s program possesses its own ontology (the world as the infosphere) and its own method (levels of abstraction), rather than merely a topical focus.

Genesis and Global Lineage

The field was named by Luciano Floridi—a philosopher from Rome educated at La Sapienza and the University of Warwick—in his article “What Is the Philosophy of Information?” published in Metaphilosophy (2002). He laid out the full research program in The Philosophy of Information (Oxford University Press, 2011), followed by The Ethics of Information (2013) and The Logic of Information (2019). Its intellectual lineage runs from Claude Shannon’s mathematical theory of communication (1948) and Norbert Wiener’s cybernetics, through Yehoshua Bar-Hillel and Rudolf Carnap’s theory of semantic information (1953), to Fred Dretske’s naturalistic epistemology (1981) and Jon Barwise and John Perry’s situation logic. Floridi added two distinctive elements to this tradition: a veridicality (truthfulness) condition built directly into the definition of information (rejecting alethic neutrality), and informational structural realism, according to which entities at an appropriate level of abstraction are informational objects. In recent years, the concept has been partially co-opted by business and technology discourse, where “philosophy of information” often means any loose reflection on data—a circulation in which Floridi’s central question of truth disappears.

Central European Lineage

While the canonical author does not come from Central Europe, the foundational question he poses has an earlier, little-known Polish counterpart. Marian Mazur, an electrical engineer who graduated from the Warsaw University of Technology (Politechnika Warszawska) in 1934 and worked after the Second World War at the Electrotechnical Institute and the Praxeology Unit of the Polish Academy of Sciences (Zakład Prakseologii PAN) in Warsaw, published Jakościowa teoria informacji (Qualitative Theory of Information) in 1970 with the Warsaw scientific and technical publisher Wydawnictwa Naukowo-Techniczne (WNT). In that book, Mazur proposed a concept of information separated from Shannon’s quantitative measure: information is a relation of transformation between messages in a cybernetic control pathway, and the number of bits says nothing about what is actually being transmitted. This shift from asking “how much” to asking “what” closely anticipates Floridi’s later distinction between the mathematical theory of communication and the semantic philosophy of information. Although Mazur formulated a parallel insight three decades earlier, there is no evidence that Floridi knew his work, and the book’s Russian (1974) and French (1976) translations did not bring it into Anglo-American philosophical circulation. The connection remains a documented anticipation rather than a direct influence.

Why Now

Between 2018 and 2026, the question “what is information?” ceased to be merely academic. The entry into force of the European Union’s General Data Protection Regulation (GDPR) in May 2018 forced legal systems to define personal data as objects with their own distinct identity, while the Cambridge Analytica scandal (2018) demonstrated that such informational objects wield direct political power. The public release of large language models from late 2022 onward posed the problem in a sharp new form: these systems generate texts that are syntactically well-formed and semantically meaningful, yet lack any intrinsic mechanism guaranteeing truthfulness—producing precisely what, under Floridi’s definition, does not qualify as information at all. Debates over model “hallucinations,” mandatory labeling of synthetic content under the EU Artificial Intelligence Act (2024), and the boundary between misinformation and disinformation in reports surrounding the 2024 global election cycle all unfold within the conceptual grid that the philosophy of information prepared two decades earlier. Floridi, based at Yale since 2023, has become one of the most frequently cited voices in AI governance, carrying his philosophical program directly into regulatory practice.

Explanatory Power

The philosophy of information makes it possible to distinguish three things that digital discourse routinely conflates: data (diaphoric differences that can be recorded), information (data that are well-formed, meaningful, and truthful), and knowledge (information embedded in a network of explanatory justifications). First, the question of whether a large language model “knows” anything breaks down into three separate questions, only the first of which (that it processes data) receives an affirmative answer. Second, the method of levels of abstraction explains why debates over whether an algorithm “discriminates” often talk past each other: opposing sides describe the exact same system at different levels of abstraction (source code, training data distribution, or downstream social outcomes), and both can be right relative to their chosen interface. Third, the concept of the infosphere allows us to describe a smartphone, a neural prosthesis, and a state archive as components of a single environment in which a human being is an informational organism (inforg) rather than an external user of tools—explaining why being cut off from the network feels like the loss of a sensory modality rather than the mere absence of an appliance.

Axes of Debate

Underappreciated Angle

The least discussed element of Floridi’s program is its method rather than its substantive theses. Floridi borrowed the method of “levels of abstraction” from formal specification in theoretical computer science (defining a system through a typed set of observables) rather than from traditional metaphysics, deploying it as an analytical tool to resolve philosophical disputes that remain intractable when participants speak about different observables under the same name. This shift has an immediate practical consequence: the question “is this AI system fair?” must always be preceded by the question “at what level of abstraction are we evaluating it?”, which in AI regulation translates into a concrete requirement to document which observables an auditor is actually inspecting. A second overlooked thread comes from Central European cybernetics: in Marian Mazur’s work, the qualitative question about information was a question about steering and control rather than truth, suggesting that Floridi’s veridical definition is not the only conceptual path leading beyond Shannon.