SYNTHESIS NOTE
Topics›Conversation Agents›this note

Why do language models respond passively instead of asking clarifying questions?

Explores whether the reward signals used to train language models might actively discourage them from seeking clarification or taking initiative in conversations, and what alternative training approaches might enable more collaborative dialogue.

Synthesis note · 2026-02-22 · sourced from Conversation Agents
Why do AI agents fail to take initiative? What kind of thing is an LLM really? How do you navigate synthesis across fragmented research topics?

CollabLLM makes the training mechanism behind passive responding explicit: "Large Language Models are typically trained with next-turn rewards, limiting their ability to optimize for long-term interaction." The result: models respond passively to ambiguous or open-ended user requests, failing to help users reach their ultimate intents and leading to inefficient conversations.

The fix is multi-turn-aware rewards — rewards that estimate the long-term contribution of a response to the overall interaction quality, not just its immediate helpfulness. By reinforcement fine-tuning with these rewards, CollabLLM enables models to:

This is a direct mechanism explanation for the alignment tax. Since Does preference optimization harm conversational understanding?, we know that RLHF training degrades multi-turn reliability. CollabLLM identifies the specific training signal responsible: next-turn rewards. And it proposes the specific fix: rewards that account for multi-turn consequences.

The connection to proactivity is also direct. Since Why can't conversational AI agents take the initiative?, the passivity is not just a missing feature — it is actively trained in by next-turn reward optimization. You cannot add proactivity on top of a training signal that rewards only reactive helpfulness.

The CollabLLM framework evaluates on three challenging tasks including document creation — contexts where multi-turn collaboration is essential and single-turn helpfulness is insufficient. This grounds the claim in practical interaction scenarios rather than abstract capability measurement.

The Intent Mismatch paper directly supports this causal mechanism: it argues premature assumptions in multi-turn conversation are rational under RLHF helpfulness training. Models construct plausible task formulations for "typical" users and produce provisional answers because the training objective penalizes evasion and rewards helpfulness. The proposed fix — a Mediator-Assistant architecture that decouples intent understanding from task execution — complements CollabLLM's reward-signal approach with an architectural intervention. Both identify next-turn optimization as the root cause; they differ on whether the fix is changing the reward (CollabLLM) or restructuring the system (Intent Mismatch).

Inquiring lines that read this note 201

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.

How should dialogue systems represent uncertainty from noisy speech input? How do chatbots affect human self-disclosure and emotional engagement? Why do multi-turn conversations degrade AI intent and coherence? Does RLHF training sacrifice accuracy and grounding for user agreement? Does conversational format create illusions of genuine AI communication? Why do language models struggle with implicit discourse relations? How should dialogue recommender systems manage conversation history and state? How do formal dialogue structures reveal conversation coherence mechanisms? How do language models establish social grounding in human dialogue? Can next-token prediction alone produce genuine language understanding? Why do language models reinforce false assumptions instead of correcting them? How can language models sustain linguistic synchrony and intersubjectivity during dialogue? How should conversational agents balance goal-driven initiative with user control? What mechanisms drive sycophancy and how can we mitigate it? Why do LLM chatbots fail as independent therapeutic agents? What makes dialogue-based explanation more successful than monologue? How can models identify insufficient information and respond appropriately without guessing? What structural biases does transformer attention create in language model outputs? How faithfully do LLMs reflect their actual reasoning in outputs and explanations? Why should disagreement be treated as signal in collaborative reasoning? Can prompting inject entirely new knowledge into language models? How can emotions function as reliable information in reasoning and cognitive systems? Can AI systems balance emotional competence with factual reliability? What properties determine whether reward signals teach genuine reasoning? Can AI systems develop genuine social understanding without embodiment? What pretraining choices and baseline capability constrain reinforcement learning gains? Do language models learn genuine linguistic structure or just surface patterns? How can LLM user simulators model realistic goal-driven conversation? What constrains reinforcement learning's ability to expand model reasoning? What makes specific clarifying questions more effective than generic ones? What prevents language models from reliably adopting diverse personas? How do language models inherit human biases from training data? Why do reasoning models fail at systematic problem-solving and search? Does reinforcement learning teach reasoning or just when to reason? Does alignment training create blind spots in detecting genuine safety threats? How do training priors constrain what context information can override? How can AI alignment serve diverse human preferences at scale? How do multi-agent systems achieve genuine cooperation and reasoning? Why do reward structures fail to shape long-term agent learning? Do language model representations contain causally steerable task-specific features? How should models express uncertainty rather than forced confident answers? How should dialogue systems best leverage conversation history for retrieval? What articulatory information do speech signals carry that text cannot?

Related concepts in this collection 5

This note in its neighbourhood — explore the map, then jump to a related concept in the list below.

Concept map
18 direct connections · 141 in 2-hop network ·medium cluster Open in graph ↗
Why do language models respond passively instead… Does preference optimization harm conversational u… Why can't conversational AI agents take the initia… Does RLHF training push therapy chatbots toward pr… Why do language models lose performance in longer … Why do standard alignment methods ignore partner i…

Click a node to walk · click center to open · click Open in graph to see this note in the full knowledge graph

your link semantically near linked from elsewhere

Related papers in this collection 8

Papers most semantically related to this note, ranked by cosine similarity in the embedding space.

Search by related questions 4

Suggested questions this note speaks to — click to search the collection, or type your own.

Can architectural changes like decoupling intent understanding help overcome next-turn reward limitations?How does RLHF helpfulness training drive premature assumptions in multi-turn dialogue?What specific metrics distinguish single-turn versus multi-turn collaboration success?Does optimizing for alignment actually reduce conversational grounding over time?
Original note title

next-turn reward optimization limits multi-turn collaboration — multi-turn-aware rewards enable models to actively uncover intent rather than passively respond