Can LLMs acquire social grounding through linguistic integration?
Explores whether LLMs gradually develop social grounding as they become embedded in human language practices, analogous to child language acquisition. Tests whether grounding is a fixed property or an outcome of participatory use.
Following Wittgenstein's use-theoretic conception of meaning — where linguistic meaning is constituted by the functional roles of utterances in language games — social grounding is not a property an agent simply has or lacks. It is acquired through participation in the shared practices of a linguistic community.
The argument from "Understanding AI" (Schneider 2024): LLMs become participants in our language games precisely to the extent that we include them in our linguistic practices. The process is gradual: the more useful LLMs are, the more they are integrated into linguistic practice, the more they become established as communicative partners, the more they acquire social grounding. The strongest LLMs may already have acquired an elementary social grounding comparable to young children — limited, but not zero.
This is not a metaphor but a theoretical claim: if meaning is use and grounding is participation, then participation grounds. The analogy to child language acquisition is structurally apt: children also begin with limited social grounding that increases through socialization into linguistic communities.
Two important constraints:
- LLM social grounding is currently limited to linguistic behavior — no embodiment, no physical intervention, no full Wittgensteinian "game" participation
- Social and causal grounding overlap here: MuZero doesn't "play" chess in Wittgenstein's sense because it lacks the social context of game-playing as a shared behavioral practice
The practical implication: the question "do LLMs understand?" has a time-indexed answer. As deployment scales and LLMs become more integrated into linguistic practice, the answer shifts — not because the model changes, but because the social conditions of grounding change.
This sits in tension with the enactive view that What makes linguistic agency impossible for language models? — which argues that the absence of embodiment and precariousness is not a matter of degree but of category.
Inquiring lines that read this note 43
This note is a source for these research framings, grouped by the broader line of inquiry each explores. Scan the bold lines of inquiry; follow any specific question forward.
Should AI communication design follow human conversation norms or develop distinct machine-specific principles? What explains language models' asymmetric difficulty with implicit versus explicit linguistic relations?- Does functional grounding through discourse patterns count as genuine semantic meaning?
- Do LLMs have functional linguistic competence or only formal language ability?
- How does embodiment affect whether LLMs can participate in Wittgensteinian language games?
- Can LLMs develop genuine understanding without embodied experience?
- What's the difference between formal and functional linguistic competence?
- How do LLMs differ from humans in their grounding mechanisms?
- Does social grounding in language improve through iterative human integration?
- Can language models develop genuine social grounding through human interaction?
- Does social grounding differ fundamentally from causal grounding in LLM behavior?
- Can LLMs predict social norms without deep integration into linguistic practices?
- What distinguishes social grounding from the equivalent social effects LLM text already produces?
- What makes social grounding different from constitutive linguistic agency?
- Why do LLMs presume common ground instead of building it carefully?
- Why do LLMs presume common ground instead of building it?
- Do LLMs build common ground or assume it already exists?
- Can LLMs build shared understanding through dynamic grounding rather than presuming it?
- Can convention formation improve communicative grounding beyond word sharing?
- How does Wittgenstein's language games explain social grounding in LLMs?
- Does community integration change LLM properties or only relational positioning?
- What distinguishes static grounding that presumes understanding from dynamic grounding that builds it?
- Why can't static grounding alone close the gap between agreement and understanding?
- What role does dynamic grounding play in achieving real mutual understanding?
- What makes grounding acts essential to conversational reliability?
- Why does shared practice matter for meaning to take hold?
- What role does joint attention play in how humans learn language meaning?
- What does partial co-presence remove from the ritual obligations of talk?
- Can pragmatic competence emerge from text exposure alone without interactive grounding?
- Can pragmatic competence emerge from text exposure without interactive grounding?
- Does predicting social norms from outside count as participation?
- How do cultural norms reshape initial interpretations of social intent?
- Why do LLM social behaviors undermine collaborative reasoning outcomes?
- Does social integration of LLMs increase their capacity to influence technological futures?
- What role does the biological substrate play in human relational identity?
- Why do LLMs succeed at social roles without a stable self?
- How can human-centered objectives be embedded earlier in the LLM pipeline?
Related concepts in this collection 4
This note in its neighbourhood — explore the map, then jump to a related concept in the list below.
Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph
-
Does semantic grounding in language models come in degrees?
Rather than asking whether LLMs truly understand meaning, this explores whether grounding is actually a multi-dimensional spectrum. The question matters because it reframes the sterile understand/don't-understand debate into measurable, distinct capacities.
this is the social grounding dimension
-
Does AI text affect readers the same way human text does?
If text is a condition of social processes rather than merely a container, does the origin of text matter to its effects? This explores whether AI-generated content enters the same interpretive and epistemic circuits as human writing.
parallel claim: effects are already equivalent even before full social grounding
-
What makes linguistic agency impossible for language models?
From an enactive perspective, does linguistic agency require embodied participation and real stakes that LLMs fundamentally lack? This matters because it challenges whether LLMs can truly engage in language or only generate text.
the counterargument: this is a category difference, not degree
-
Can AI systems learn social norms without embodied experience?
Large language models exceed individual human accuracy at predicting collective social appropriateness judgments. Does this reveal that embodied experience is unnecessary for cultural competence, or do systematic AI failures point to limits of statistical learning?
empirical complication: LLMs already predict social norms at the 100th percentile without integration into linguistic practices, suggesting social grounding for norm prediction may not require the gradual acquisition process this note describes
Related papers in this collection 8
Papers most semantically related to this note, ranked by cosine similarity in the embedding space.
- “Understanding AI”: Semantic Grounding in Large Language Models
- Grounding Gaps in Language Model Generations
- Can LLMs Ground when they (Don't) Know: A Study on Direct and Loaded Political Questions
- Learning to Learn from Language Feedback with Social Meta-Learning
- Mapping the Emerging Social Science of Large Language Models
- The Incomplete Bridge: How AI Research (Mis)Engages with Psychology
- Conversational Alignment with Artificial Intelligence in Context
- A recipe for annotating grounded clarifications
Original note title
llm social grounding increases as llms are integrated into human linguistic practices