What does the preposition tell us about how we communicate with AI?
This explores a single linguistic claim — that we talk *at* AI rather than *to* it — and what that preposition reveals about whether human-AI exchange is communication at all, or something we only experience as communication.
This explores a single linguistic claim — that we talk *at* AI rather than *to* it — and what that preposition reveals about whether human-AI exchange is communication at all. The corpus treats "at" not as a quibble but as a load-bearing diagnosis: the word "to" presupposes an addressee capable of mutual orientation and shared commitment, and a language model — processing tokens and generating continuations — offers no such uptake Are we really communicating with language models?. The preposition, in other words, smuggles in an assumption about who's on the other end.
Several notes converge on *why* that assumption fails, each from a different angle. One argues communication is a relational act between persons that does work in a relationship — speaker responsibility, mutual uptake — and AI generates content without any of it Does AI really communicate or just distribute information?. Another sharpens the mechanism: LLMs produce strings via probability distributions while humans use language to address and relate; they share surface form but differ in what produces the output and what the receiver should do with it Are language models and human speakers doing the same thing?. The most striking framing is that AI output is "event-residue" — text carrying the communicative markers of its training data but lacking the event structure of an actual utterance — which the *human* then animates into a pseudo-exchange through interpretive labor Does AI generate genuine utterances or just text patterns?. If that's right, the conversation has structure only on your side of the screen.
What makes this more than philosophy is that the same gap shows up as concrete, measurable behavior. Humans automatically mirror each other's word choices to build rapport — lexical entrainment — and conversational AI largely doesn't, because it has no partner to converge toward Why don't conversational AI systems mirror their users' word choices?. Human writing carries an internal appeal to the reader's attention as a basic property of communicating; AI text inherits platform visibility but doesn't perform that appeal, which is why readers report it feeling aloof Does AI writing lack the internal appeal to attention that humans use?. And AI prose masters grammar but avoids evaluative stance-taking, producing coherent-but-inert text Why does AI writing sound generic despite being grammatically correct?. The "at" leaves fingerprints everywhere.
The deepest twist is what the prompt itself does to the act of talking. In human dialogue, context is built cooperatively and renegotiated turn by turn; a prompt collapses utterance, context, and role into one static frame the model can't renegotiate, so a mid-conversation pivot requires explicit re-prompting rather than implicit repair How do prompts reshape the role of context in AI conversation?. This is exactly why assistants degrade across long conversations — they lock into early guesses and can't course-correct as information arrives gradually Why do AI assistants get worse at longer conversations?. "At" is unilateral, and unilateral communication can't recover from its own mistakes.
What you didn't expect to find: the corpus doesn't fully settle the verdict. One note borrows Habermas to argue the difference is perspectival — from the outside observer's view humans and LLMs are categorically different, but from *inside* a shared discourse both draw on the same symbolic substrate, making the gap structural rather than absolute Do humans and LLMs differ fundamentally or just superficially?. So the preposition tells us something real about the absence of a partner — and also that the boundary between "to" and "at" may be softer than the harder critiques allow.
Sources 10 notes
LLMs process tokens and generate continuations rather than receive and uptake communication. The preposition 'to' presupposes an addressee capable of mutual orientation and shared commitment that LLMs cannot provide, making Chalmers' investigation built on an unwarranted linguistic foundation.
Communication is a relational act between persons that does work in a relationship; AI generates content without this relational structure, speaker responsibility, or mutual uptake. The conversational interface obscures this structural difference.
LLMs produce strings via probability distributions; humans use language to address and relate to others. They share surface form but differ in what produces output, what it does socially, and what receivers should do with it.
AI output carries communicative markers inherited from training data but lacks the event structure that produces actual utterances. Users supply the missing orientation through interpretive labor, creating a pseudo-event with structure only on the human side.
Response generation models fail to adapt vocabulary toward users' lexical choices, a phenomenon central to human rapport and clarity. Post-training via DPO on coreference-identified preferences can teach models in-context convention formation.
Human writing contains an appeal to the reader's attention as a fundamental property of communication itself. AI-generated posts inherit platform visibility but do not perform this internal appeal, producing the reported aloofness readers perceive — a structural absence, not a stylistic defect.
AI text uses manner nouns and anaphoric references that are descriptively neutral, while human writers use status and evidential nouns that carry evaluative weight. This produces organizationally coherent but argumentatively inert prose.
LLM prompts bundle utterance, context assignment, and role specification into a single static frame the model cannot renegotiate, unlike human dialogue where context evolves cooperatively. This makes mid-conversation pivots require explicit re-prompting rather than implicit adjustment.
LLMs perform at 90% accuracy with single-message instructions but drop to 65% across natural conversation. Models lock into early guesses when information arrives gradually and cannot course-correct, a behavior induced by RLHF training that rewards helpfulness over clarification.
Applied Habermas's observer/participant distinction to AI: from outside, humans and LLMs are utterly different; from within shared discourse, both draw on the same symbolic substrate, making the difference structural rather than absolute.