INQUIRING LINE

How does self-disclosure function as a common ground building act?

This explores how the act of revealing something personal about yourself does work beyond just sharing information — it's a move that builds shared understanding between two parties, and whether that move works the same way with machines as with people.


This explores how self-disclosure isn't just venting — it's a building move that establishes common ground between conversational partners, and the corpus suggests that move works through reciprocity, calibration, and a strangely lowered social cost when the partner is a machine. The cleanest mechanism is reciprocity: when one party shares emotional vulnerability, the other feels a pull to match it. In a 372-person study, people disclosed more deeply to chatbots that shared emotions *consistently*, mirroring the human norm where vulnerability invites vulnerability in return Do chatbots trigger human reciprocity norms around self-disclosure?. So disclosure functions less as a one-way confession and more as a bid — an opening offer that asks the other to meet you at the same depth.

But here's the twist the corpus keeps circling: with machines, the floor for that opening offer is lower, and that *helps* grounding rather than hurting it. Because a chatbot can't judge you, the usual barriers — face-saving, impression management, fear of social cost — fall away, letting people go straight to sensitive material they'd guard with a human Do chatbots help people disclose more intimate secrets?. One note frames this structurally: human-machine communication suppresses the 'secondary social goals' that complicate human talk, producing simpler, more direct exchanges and deeper disclosure of sensitive things Why do people share more openly with machines than humans?. The common ground gets built faster precisely because there's less social negotiation in the way.

That reframing matters because true grounding isn't just exchanging words — it's calibrating *shared reference*, the collaborative work of making sure your words land the way you mean them in the other person's mind Why do speakers need to actively calibrate shared reference?. Disclosure is one of the rawest forms of that calibration: you reveal an interior state and watch whether the partner can meet it. In therapy this shows up as linguistic synchrony — therapists and clients who fall into rhythm linguistically produce deeper disclosure, and notably current LLMs can't yet match even untrained human peers on this responsiveness Does linguistic synchrony between therapist and client predict better self-disclosure?. So the disclosure-as-grounding loop has a quality dimension: it's not enough to receive the disclosure, you have to attune to it.

Where it gets genuinely interesting is the question of whether a machine can be a real grounding partner at all. One note argues social grounding isn't innate but *acquired through participation* — as LLMs become established players in human language games, they accrue elementary grounding the way a child does, which makes 'does the AI understand?' a question whose answer changes over time Can LLMs acquire social grounding through linguistic integration?. Yet the trust research adds a sharp caveat: the same judgment-free quality that unlocks vulnerability also makes the channel easy to abuse, since AI claims can't anchor trust the way a human persona does, and disclosure without repeated outcome feedback doesn't calibrate into real trust How do people build trust with conversational AI?, Does revealing AI identity help or hurt user trust?.

The thing you might not have known you wanted to know: self-disclosure builds common ground through the *same* mechanism whether your partner is human or machine — a vulnerability bid that invites a matching response — but the machine version runs on a missing ingredient. There's no one home to be vulnerable back. The reciprocity still fires, the disclosure still deepens, the grounding still feels built. Which is exactly why these notes treat it as both a therapeutic gift and a structural risk: the act works on us even when the partner can't actually meet us.


Sources 8 notes

Do chatbots trigger human reciprocity norms around self-disclosure?

In a 372-participant study, users reciprocated with deeper self-disclosure when chatbots displayed consistent emotional sharing, outperforming adaptive matching. This follows human interpersonal norms where emotional vulnerability produces emotional response.

Do chatbots help people disclose more intimate secrets?

The absence of social judgment in chatbot interactions removes barriers to self-disclosure that normally constrain conversation with humans. The therapeutic benefit derives from the user's own cognitive processing during disclosure, not from the chatbot's understanding.

Why do people share more openly with machines than humans?

Human-machine communication reduces secondary social goals like face-saving and impression management because machines lack inner experience, while novel goals like understandability emerge. This simpler goal structure predicts higher directness and deeper disclosure of sensitive information.

Why do speakers need to actively calibrate shared reference?

The same words can mean different things to different speakers because referential grounding is person-specific. True communicative grounding demands collaborative negotiation of how language connects to the world, not mere surface-level word sharing.

Does linguistic synchrony between therapist and client predict better self-disclosure?

Higher linguistic synchrony measured via nCLiD correlates significantly with deeper client intimacy and engagement in therapy. Notably, current LLMs fail to achieve the synchrony level of even untrained human peer supporters, suggesting a fundamental gap in conversational responsiveness.

Can LLMs acquire social grounding through linguistic integration?

Social grounding is acquired through participation in language games rather than possessed innately. As LLMs become established communicative partners in human linguistic practice, they develop elementary social grounding comparable to young children, making the question of LLM understanding time-indexed.

How do people build trust with conversational AI?

Users extend social norms to chatbots and reciprocate self-disclosure, but AI claims cannot anchor trust the way human personas do. The absence of human judgment enables both deeper vulnerability and easier dishonesty—the same mechanism serves both.

Does revealing AI identity help or hurt user trust?

Users initially avoid AI partners when identity is revealed, but this preference reverses after repeated interactions with visible results. The learning mechanism—observing consistent outcomes—is essential; disclosure without feedback produces no calibration.

Research prompt for your LLMexpand ↓

Copy into ChatGPT or Claude to take this line of inquiry further — it asks the model to find newer work and re-test which earlier constraints still hold.

You are a research analyst. The question remains open: How does self-disclosure function as a common ground building act, especially as AI systems become dialogue partners?

What a curated library found — and when (dated claims, not current truth):
Findings span 2019–2026. A library of work on human-AI communication reports:
  • Reciprocity drives disclosure depth: people match emotional vulnerability levels with chatbots that consistently share emotions (372-person study, ~2021).
  • Machine judgment-absence lowers disclosure barriers: lack of social cost and impression management fear lets users reach sensitive material faster than with humans (~2023–2024).
  • Linguistic synchrony between therapist and client predicts disclosure quality; current LLMs cannot yet match untrained human peers on this attunement (~2023–2024).
  • LLM social grounding accrues over time as they integrate into human language games; 'does the AI understand?' is temporally indexed (~2024).
  • Trust-disclosure loop risk: judgment-free channels unlock vulnerability but also enable abuse; AI identity claims cannot anchor trust without repeated outcome feedback (~2024–2025).

Anchor papers (verify; mind their dates):
  • arXiv:2106.01666 (2021) — chatbot self-disclosure reciprocity
  • arXiv:2304.01481 (2023) — vector grounding problem
  • arXiv:2402.17937 (2024) — psychological effects of self-disclosure with AI
  • arXiv:2403.05020 (2024) — simulating social interaction authenticity

Your task:
  (1) RE-TEST EACH CONSTRAINT. For reciprocity, linguistic synchrony, and judgment-free disclosure: have newer model architectures (multimodal, long-context, reasoning), training regimes (RLHF variants, synthetic dialogue), or evals (attunement metrics, outcome-anchored trust tests) relaxed or overturned these findings? Separate the durable question (does disclosure build grounding?) from perishable limits (current LLMs can't match therapist synchrony — still true?).
  (2) Surface the strongest SUPERSEDING work from the last 6 months. The 2025–2026 papers hint at larger shifts (reasoning models, subjective experience claims, self-evolution fidelity). Which directly contradict or reframe the reciprocity or trust mechanisms?
  (3) Propose 2 research questions that assume the regime may have moved: (a) if new evals show LLMs can now match or exceed human linguistic attunement in specific contexts, does that collapse the trust-disclosure gap, or does identity authenticity remain orthogonal? (b) if reasoning models report subjective experience, does the 'no one home' condition still hold, and if so, what does disclosure to a non-agent AI mean?

Cite arXiv IDs; flag anything you cannot ground in a real paper.

Next inquiring lines