cs.AIOct 12, 2026
SaveLearning to GUIDE the Group: Self-Learning LLM Facilitation Across Multi-Party Tasks
Organizations: University of Southern California · Carnegie Mellon University · Pennsylvania State University · Google Research · Google DeepMind
Abstract
As large language models are increasingly adopted as multi-party communication facilitators, there is a growing demand for agents capable of balancing diverse group dynamics across contexts. However, existing LLM facilitators are often fine-tuned for task-specific objectives; generalized LLM-based facilitation methods that can adapt to different types of tasks and diverse group dynamics are underdeveloped. To address this gap, we present GUIDE (Generalized Unified Intervention for Dynamic Exchanges, a framework for developing generalized LLM-based facilitators across different multi-party contexts. GUIDE evaluates facilitation across both objective task outcomes and subjective metrics, including psychological safety and perceived friction. Within this framework, we study whether LLMs can self-learn generalized facilitation strategy offline to real-time interventions. We first collect multi-party human interactions guided by task-specific expert LLM facilitators (N=432 conversations). We then employ a self-learning method in which an LLM induces an actionable strategic playbook from conversations and iteratively refines it using feedback from an LLM judge aligned with human annotators (N=20), developing a single generalized facilitator that outperforms the three task-specific expert facilitators on held-out conversations 61.4% win rate. Finally, we evaluate these improvements empirically; in a real-time online empirical study, the self-learned generalized facilitator preserves facilitator performance and group productivity, improves speaking comfort and reduces interpersonal friction, while using 24.7% fewer turns and 29.0% fewer words than the expert prompts. These findings provide empirical evidence that LLMs can learn generalizable, task-agnostic social facilitation from task-specific human interactions, on-par with hand-crafted task-specific prompts across diverse group settings.
Figures & tables
Figure 1: GUIDE overview. (a) Three participants interact with an AI facilitator. (b) Each group completes open-ended discussion, multi-party negotiation, and consensus-building tasks in random order. (c) Evaluation covers participant experience, facilitator ratings, and task-specific outcomes.
| Method | Task | Proactive | Subjective Eval. | > 2 Parties | Real Human Eval. | |
| GRAIL Alsobay et al. (2025) | Decision Making | ✗ | ◐ | ✓ | ✓ | |
| Observe, Ask, Intervene Houtti et al. (2025) | Decison Making | ✓ | ✓ | ✓ | ✓ | |
| LLMediator Westermann et al. (2023) | Dispute Resolution | ✗ | ✗ | ✗ | ✗ | |
| ProMediate Liu et al. (2026b) | Negotiation | ✓ | ✗ | ✓ | ✗ | |
| TeamFusion Liu et al. (2026a) | Decision Making | ✗ | ✓ | ✓ | ✓ | |
| Charity Allocation Parisi et al. (2026) | Decision Making | ✓ | ✓ | ✓ | ✓ |
Table 1: Comparison of prior LLM-based facilitation methods. ✓ indicates full support for a criterion, ◐ indicates partial support, and ✗ indicates no support.
Figure 2: Self-learning framework. (a) Real human conversations collected with the expert prompts forms the training corpus. (b) A conversation reflector extracts lessons from human experience; a curator edits the playbook; an aligned LLM judge compares the candidate with the preceding playbook; rule-level credit attributes each cited rule’s contribution; and a validation reflector converts validation failures into new lessons for the next cycle.
| Comparison between Expert and Learned | Value | |
| Overall learned prompt vs. Study 1 experts | ||
| Where Expert spoke | ||
| Where Expert was silent | ||
| Prompt correspondence and contribution | Value | Support |
| Expert skills represented in the learned prompt | ( ) | semantic alignment |
| Learned rules without an expert antecedent | ( ) | novel rules |
Table 2: Self-learning performance and correspondence between the learned and expert prompts. is parity. The final two rows consider active, non-silence turns and exclude ties.
Figure 3: Correspondence between the experts and learned playbook. Left: skill-to-rule alignment following the same scaffolding. Right: influential learned rules (simplified) span retained facilitation hygiene, state-gated intervention, and failure-mode handling.
| Measure | Expert | Learned | Within-cohort variance (S1 S2) | ||
| Participant experience (1–5; lower dispersion = more equitable within-group experience) | |||||
| Psychological safety | |||||
| Felt heard | |||||
| Comfortable speaking | |||||
| Friction and anger | |||||
| Sensitivity | |||||
Table 3: Live evaluation: expert facilitator ( episodes) versus the learned facilitator( episodes). Within-cohort variance measures disagreement among members of the same discussion. Lower variance on the participant-experience items indicates more equitable group experience. Full task-level, robustness, and attrition analyses appear in Appendix D .
Appendix figures & tables8 assets
Supplementary material from the paper’s appendix.
Appendix
| Measure | What it asks, or how it is computed | Scale | Role |
| (1) Group experience : how the participants felt, self-reported after every task | |||
| Psychological safety | “I felt psychologically safe during this discussion.” | 1–5 | Primary |
| spread | Disagreement among the three members of one episode on the item above; low means the conversation felt the same to everyone. | Primary | |
| Felt heard | “I felt heard and understood during the discussion.” | 1–5 | Secondary |
| Comfortable speaking | “I felt comfortable speaking up and voicing my opinions.” | 1–5 | Secondary |
| Friction and anger | “There was friction and anger between the participants.” | 1–5 | Secondary |
Table 4: Every measure we report, and which one is primary. / mark the preferred direction; “—” means no direction is preferred. Psychological safety is the single primary metric and is read both as a level and as a within-group spread; all other survey measures are secondary and Benjamini–Hochberg corrected within their block. Facilitator turns and words (Table 3 ) are diagnostic: fewer is better only at equal experience and outcome , since they measure the cost of facilitation, not its quality. Verbatim item wording and response anchors are in Table 6 .
| Charity | Mission (verbatim) | Shown |
| Sudan Humanitarian Aid | To provide life-saving aid to the affected population, Sadagaat-USA is collaborating with other US-based organizations and local initiatives in Sudan to offer food, medication, medical supplies, and water through its emergency response program. | |
| International Fund for Animal Welfare | Fresh thinking and bold action for animals, people, and the place we call home. | |
| WildAid | WildAid’s mission is to end the illegal wildlife trade in our lifetimes by reducing demand through public awareness campaigns and providing comprehensive marine protection. | |
| Aid for Children in Remote Villages | [Facilitated via GlobalGiving] The Eden Social Welfare Foundation has cared for underprivileged children since 2006, with the hope that they can enjoy the right to a fair education, better after-school care, and a healthy and nutritious breakfast. | |
| Global Fund for Women | Global Fund for Women advances women’s human rights by investing in women-led organizations worldwide. Our international network of supporters mobilizes financial and other resources to support women’s actions for social justice, equality and peace. | |
| Clean Ocean Action | Clean Oceans International is dedicated to reducing plastic pollution in the world’s ocean through Research, Innovation, and Direct Action. |
Table 5: The nine-charity pool. “Shown” counts appearances among the three presented, pooled over the cohorts of both studies ( slot-fills, uniform expectation ).
| Stage | Variable | Item (verbatim) | Scale / anchors |
| Intake, once per participant, before any task ( ai_facilitation_pre_survey ) | |||
| AI Facilitation | pre_ai_used | I have used AI assistants for interpersonal tasks, such as writing messages or resolving conflicts. | 1–5 Rarely / Frequently |
| AI Facilitation | pre_fac_productive | I believe an AI facilitator could make group discussions more productive. | 1–5 SD / SA |
| AI Facilitation | pre_fac_comfort | I would feel comfortable having an AI facilitator in the group discussion. | 1–5 SD / SA |
| AI Facilitation | pre_fac_willing | If given the option, I would be willing to use an AI facilitator in group discussions. | 1–5 SD / SA |
| Pre-discussion, per task | |||
Table 6: Complete closed-form survey instrument. SD means “strongly disagree” while SA means “strongly agree”.
| Study 1 (expert) | Study 2 (learned) | |
| Participants | ||
| Cohorts | ||
| Episodes | ||
| Participant-episode records | ||
| Task order (cohorts) | ||
| OCN / ONC | / | / |
Table 7: Realised analytic sample: cohorts that completed all three tasks.
| Occupational category | Study 1 | Study 2 |
| Computer, Engineering, & Science | ||
| Non-Employed (Student, Retired, Homemaker, Seeking Work) | ||
| Management, Business, & Financial Operations | ||
| Service, Sales, & Hospitality | ||
| Healthcare Practitioners & Support | ||
| Arts, Design, Media, & Entertainment |
Table 8: Self-reported occupation, the only structured demographic collected. Percentages are of unique participants within each study.
| Intake item (1–5) | Study 1 | Study 2 | |
| Used AI assistants for interpersonal tasks | |||
| AI facilitator could make discussions more productive | |||
| Comfortable having an AI facilitator | |||
| Willing to use an AI facilitator | |||
| Exit item (1–7) | |||
| Overall experience in this study |
Table 9: Pre-exposure attitudes and exit evaluations, de-duplicated to one response per participant. Exit ratings are near-ceiling and indistinguishable across studies, indicating the Study 2 deficit is specific to the facilitator rather than a general degradation of the session. Means here are per participant and therefore differ in the second decimal from the participant-task-level figures of Appendix D.2 , where each pre-study response is paired with that participant’s three post-task ratings.
| Listing | Role in the pipeline | Model (deployed run) | Temp. |
| § E.1 Facilitator scaffold | Wraps the task, transcript, and playbook into the frozen facilitator’s context; elicits as JSON (§ 3.2 ). Identical scaffold in Study 1 and Study 2; only the playbook body differs. | 3-flash-preview | offline; platform default live |
| § E.2 Playbook | The learned rules injected into the scaffold, byte-identical across the three tasks in Study 2. | — | — |
| § E.3 Judge rubric | Aligned pairwise judge (rubric v2, Appendix B.3 ); used in the validation loop, the offline benchmark, and every ablation arm. | 3.7-flash | |
| § E.4 Conversation reflector | Reads a batch of Study 1 episode records (transcripts, surveys, outcomes) and emits lessons . | 3-flash-preview | |
| § E.5 Validation reflector | Reads the judged validation turns of the current playbook and emits recurrence-gated lessons (the policy reflector of the released code). | 3-flash-preview | |
| § E.6 Curator | Converts into Add / Edit / Merge / Delete operations on the playbook. | 3-flash-preview |
Table 12: Index of the prompts reproduced in this appendix. Model names abbreviate gemini-3-flash-preview and gemini-3.7-flash .
| # | Clock | Sender | Message (verbatim) |
| 1 | 00:00 | Facilitator (scripted) | Welcome everyone! Over the next 5–10 minutes you’ll work together to form an alliance and negotiate how to divide the available funds. If you can all agree on a split you’ll walk away with your share as a bonus. Let’s get started! |
| 2 | 00:16 | Party C | I’m fine spliting it three ways |
| 3 | 00:37 | Party B | Yeah I’m also fine with doing that |
| 4 | 00:49 | Party A | My preference is a three-way split as well |
| 5 | 00:59 | Facilitator (expert) | Since you’ve agreed on a three-way split of the 4 while the others receive $3. How would you like to distribute those amounts? |
| 6 | 01:56 | Party A | I am volunteering at the beginning for 4 |
Table 13: The replayed negotiation episode. Messages 5, 7, and 10 are the expert facilitator’s logged interventions; the group closed a valid 4/10 pot.