Meaning mediation

Meaning mediation concerns the processes and arrangements through which something becomes significant and available for further experience, communication or action. An expression does not carry a complete, ready-made meaning from one setting into another. Its significance depends on interpretation, context, history and the practical possibilities of its use. In AE, meaning organises relevance and possible continuation, while lived experience, communication and computation remain distinct objects of inquiry. (Watson and Brezovec, 2026, AE2.6)

A familiar educational example

A student receives the sentence “Explain why this transformation is valid.” The wording might function as a helpful invitation, an assessment demand, a reminder of a previous difficulty or an incomprehensible instruction. A teacher’s tone, the student’s experience and the available opportunities for response all matter. The sentence’s existence does not establish which of these roles it plays.

An AI service could help the student formulate a question, compare explanations or find a way to express a partially developed idea. It could also supply fluent language that conceals the student’s uncertainty from an assessor. These are constructed possibilities. Establishing what occurred requires observing the student’s activity and the subsequent interpretation of the contribution, rather than inferring understanding from polished prose. (Watson, Morgner and Brezovec, 2026)

Intellectual sources

The enactive tradition connects cognition with embodied activity and a world of significance that emerges through that activity. Varela, Thompson and Rosch’s The Embodied Mind is a central statement. De Jaegher and Di Paolo extend the inquiry to social encounters, examining how the interaction process participates in the generation and transformation of meaning. These resources help resist an account confined to messages moving between otherwise unchanged individuals. (Varela, Thompson and Rosch, 1991) (De Jaegher and Di Paolo, 2007)

Luhmann’s social theory offers a different analytical emphasis: the continuation of communication through communicative operations. AE brings questions from these traditions into relation while preserving their differences. An account of a person’s experience is not identical to that experience, and neither is identical to an organisation’s classification of the account. (Luhmann, 1995) (Watson and Brezovec, 2026, AE2.6)

Generative AI as a condition of meaning

The critical question is how generated material enters an activity. Does it become a suggestion to question, an apparently authoritative answer, a resource for conversation or evidence in an institutional record? These roles depend on more than the model’s isolated output. They involve interface design, expectations, standards of judgement, access to alternatives and arrangements for correction. (Watson, Morgner and Brezovec, 2026)

Bender and Koller distinguish linguistic form from meaning and challenge inferences from success with form to understanding. Esposito approaches algorithmic communication through the social use of generated contributions, without making human-like understanding by the machine its necessary starting point. These works address different questions; reading them together helps separate a claim about a model’s understanding from a claim about its effects in communication. (Bender and Koller, 2020) (Esposito, 2022)

AE adds an inquiry into how those effects participate in constituting and changing capacities for further activity. A useful suggestion may enable a student to ask a new question. A generated judgement may restrict which answers an institution will recognise. A corrected contribution may change future interpretation, or may leave a retained record untouched. The explanatory object is the organised passage through which the contribution becomes consequential. (Watson, 2026, Operability)

Assigned, enacted and warranted roles

A contribution’s assigned role is what it is presented as being: a provisional suggestion, for example. Its enacted role is what it actually does in subsequent activity. Its warranted role is the use justified by its evidence and the applicable standards. These can diverge. A prominently labelled suggestion might still be treated as a finding; a cautious human judgement can also acquire more authority than its evidence permits. (Watson and Brezovec, 2026, AE2.6)

This distinction makes inquiry more precise. Researchers can inspect instructions and labels, observe uptake, and assess the evidence supporting the resulting decision. None of these sources alone establishes the entire sequence. A log records an action; an interview provides an account of interpretation; an assessment task can supply evidence about understanding. Their relation has to be investigated.

Why this matters for inclusion

Mediation can make a previously unrecognised contribution intelligible, widening participation. It can also privilege particular forms of expression, suppress uncertainty or make challenge harder. Watson, Morgner and Brezovec’s review examines educational AI in relation to inclusion, exclusion and semantic transduction. Its relevance is to how opportunities and criteria are organised, rather than a presumption that AI uniformly includes or excludes. (Watson, Morgner and Brezovec, 2026)

Claims of educational benefit need evidence about learning, judgement and participation over an appropriate interval. Satisfaction with a response, a high volume of feedback or a fluent final text may be relevant observations, but each answers a more limited question.

Related entries

Semantic transduction follows the reconstitution of significance across settings. Ecological hyperknots examines the different capacities implicated in those passages. The history of AE places these questions in their scholarly context.

Sources support the particular claims discussed; their inclusion does not imply endorsement of AE. Publication and manuscript status are identified below.

Watson and Brezovec, 2026, AE2.6

Watson, S., and Brezovec, E. (2026). Autopoietic Ecology: Rethinking Systems, Meaning, and Matter. AE2.6, revised manuscript dated 15 September 2026. Unpublished manuscript. Relevant locators: preface; chapters 3–6, 10–12 and 15–18; appendices A–C.

Watson, Morgner and Brezovec, 2026

Watson, S., Morgner, C., and Brezovec, E. (2026). Generative AI as meaning-mediating infrastructure in education: A critical integrative review of inclusion, exclusion, and semantic transduction. AI & SOCIETY. Source.

Varela, Thompson and Rosch, 1991

Varela, F. J., Thompson, E., and Rosch, E. (1991). The Embodied Mind: Cognitive Science and Human Experience. MIT Press. Link leads to the revised 2016 edition. Source.

De Jaegher and Di Paolo, 2007

De Jaegher, H., and Di Paolo, E. (2007). Participatory sense-making: An enactive approach to social cognition. Phenomenology and the Cognitive Sciences, 6, 485–507. Source.

Luhmann, 1995

Luhmann, N. (1995). Social Systems. Translated by J. Bednarz Jr. and D. Baecker. Stanford University Press. German original published 1984.

Bender and Koller, 2020

Bender, E. M., and Koller, A. (2020). Climbing towards NLU: On meaning, form, and understanding in the age of data. Proceedings of ACL 2020, 5185–5198. Source.

Esposito, 2022

Esposito, E. (2022). Artificial Communication: How Algorithms Produce Social Intelligence. MIT Press. Source.

Watson, 2026, Operability

Watson, S. (2026). Generative AI and the recursive constitution of operability: Rough closure, semantic transduction and ecological hyperknots. Cambridge EdgeLab working paper, version 1, 18 September. Manuscript consulted; not peer reviewed. Constructed case and prospective research propositions.