Article - November 8, 2025
So, why Discourse Analysis matter in AI?
My work began in linguistics, long before I entered design or artificial intelligence. In discourse analysis — the field shaped by thinkers like Michel Foucault, Pêcheux, and Bakhtin — language was never treated as a neutral tool. It was a mechanism of power: a system that defines what can be said, who gets to speak, and what must remain silent.
That way of reading language shaped how I see technology today. What discourse once did socially, algorithms now do computationally. Natural Language Processing (NLP) and generative AI don’t just handle text; they reproduce entire structures of discourse. They classify, prioritize, and predict meaning — automating the politics of language at scale.
Foucault described discourse as the “archive” of a culture: the infrastructure of thought that determines what becomes visible or credible. Today, that archive has become algorithmic. AI systems transform syntax into data and semantics into probabilities. They don’t interpret; they compute. Meaning, once a human negotiation, is now an automated process — fast, persuasive, and indifferent to context.
This shift doesn’t destroy meaning; it industrializes it. Designers working with AI are no longer crafting messages or interfaces — they’re building systems that generate discourse in real time. Every prompt, every model output, every tone of voice in a chatbot shapes the epistemic field users inhabit. Coherence has quietly replaced truth as the measure of credibility.
That’s where my background in discourse analysis becomes practical again. It reminds me that every system speaks — and that its language performs ideology. Training data is never neutral; it’s a mirror of historical discourse, filled with exclusions and repetitions. By treating AI outputs as discourse, not as data, we can see what they reveal about us — and what they choose to conceal.
The real reason it matters.
AI doesn’t just process language — it produces worldviews.
Every model encodes a perspective on what counts as relevant, probable, or true. Discourse analysis gives us the framework to read those perspectives critically, to expose the ideologies hidden in code, and to design with awareness of their social impact.
It matters because AI is not only shaping what we read, but what we can think.
When language becomes infrastructure, design becomes a form of authorship — and interpretation becomes an act of resistance.
When language becomes infrastructure, design becomes a form of authorship — and interpretation becomes an act of resistance.
The future of AI design won’t be led by those who code faster, but by those who read deeper — by researchers and strategists who understand that shaping discourse is shaping reality. Because language has always been architecture. And now, we are the architects of the architectures that speak.