This post introduces our 2025 article, “The Peircean Theory of AI: Advancing Text Generation through Peirce’s Triadic Model, Speculative Grammar, and Methodeutics,” published in Digital Age in Semiotics & Communication.
The article grew out of a question that continues to shape our work in digital semiotics: what does it mean for AI to produce signs without human understanding?
Large language models can generate fluent and persuasive text. They can translate, summarize, and answer questions with impressive speed. Yet fluency is not the same as interpretation. Our argument is that current AI systems remain limited because they are primarily grounded in statistical pattern recognition. They can produce language, but they do not interpret meaning in the human semiotic sense.
To address this problem, we return to Charles Sanders Peirce’s triadic model of the sign: Sign, Object, and Interpretant. For Peirce, meaning is not a fixed connection between a word and a thing. Meaning emerges through relation, mediation, context, and interpretation.
This is why Peirce is useful for thinking about AI. AI systems can generate outputs that function as signs, but they do not possess consciousness, intention, or lived experience. The article therefore reframes the AI “interpretant” not as a human mental act, but as a functional and relational output produced through interaction, feedback, and contextual adjustment.
A second part of the article focuses on Peirce’s concepts of speculative grammar and methodeutics. Speculative grammar helps explain how signs are structured and related. Methodeutics, by contrast, concerns inquiry, self-correction, and the refinement of interpretation. We argue that this is especially important for AI because future systems should not only generate responses but also revise, test, and refine them through iterative processes.
The article also draws on Claudio Paolucci’s concept of machinic enunciation. This idea helps explain how generative AI produces language without a conscious speaker behind it. AI outputs are not meaningless, but their meaning depends on how they are taken up, interpreted, and circulated within human contexts.
For us, the larger point is not that AI “understands” in the human sense. It does not. Rather, AI produces signs that enter human systems of meaning. This makes semiotics essential for studying AI because it allows us to ask better questions: How are AI-generated signs formed? How do they circulate? How do humans interpret them? And what happens when machines simulate interpretation without possessing interpretive consciousness?
This article is part of our broader effort to develop digital semiotics as a framework for studying artificial intelligence, meaning-making, and knowledge organization in computational environments.
Friedman, A., & Thellefsen, M. (2025). The Peircean Theory of AI: Advancing Text Generation through Peirce’s Triadic Model, Speculative Grammar, and Methodeutics. Digital Age in Semiotics & Communication, VIII, 50–70. https://doi.org/10.33919/dasc.25.8.3
Leave a Reply