Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Inferring Implicit Goals Across Differing Task Models
Silvia Tulli, Stylianos Loukas Vasileiou, Mohamed CHETOUANI, Sarath Sreedharan
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use strategic (MDP-optimal) querying instead of Query-All | goal inference agent · 4x4 grids | number of queries: 3.7-4.8 -> 2.0-3.3 queries (22-41% reduction) | Improved |
| use strategic querying instead of Query-All | goal inference agent · 6x6 grids | number of queries: 3.7-7.7 -> 3.2-4.3 queries (13-40% reduction) | Improved |
| use strategic querying instead of Query-All | goal inference agent · 6x6 grids Puddle and Rocks | significance of query reduction: p=0.053 (Puddle), p=0.073 (Rocks) | No clear effect |
| use strategic querying instead of Query-All | goal inference agent · Four Rooms 6x6 | significance of query reduction: p<0.001 (4x4) -> p=0.411 (6x6) | No clear effect |
| use strategic querying instead of Query-All | goal inference agent · 8x8 grids (Maze, Puddle, Rocks) | number of queries: 11.6 ± 2.9 -> 3.2 ± 1.6 (71.9 ± 10.9% reduction) | Improved |
| use strategic querying instead of Query-All | goal inference agent · 8x8 Four Rooms | number of queries: 3.2 ± 1.0 -> 3.2 ± 1.6 | No clear effect |
| reduce number of human models from 10 to 5 | strategic querying · grids, 0.1 obstacle density | runtime: 1.7-1.8s -> 0.8-0.9s bottleneck finding; 4.1-9.2s -> 2.1-4.0s total | Improved |
| increase obstacle density from 0.1 to 0.15 | strategic querying · Maze and Rocks | bottleneck finding time: 1.7-1.8s -> 27.4s (Maze), 13.7s (Rocks) | Hurt |
| increase grid size from 4x4 to 8x8 | strategic querying · grid environments | pruning time: 2-3 seconds query time -> ~308-325 seconds pruning | Hurt |
Closest papers by meaning
- Expectation Alignment: Handling Reward Misspecification in the Presence of Expectation Mismatch (NeurIPS 2024 · similarity 0.87)
- Goal Alignment: Re-analyzing Value Alignment Problems Using Human-Aware AI (AAAI 2024 · similarity 0.86)
- Belief-State Query Policies for User-Aligned POMDPs (NeurIPS 2024 · similarity 0.86)
- Just Do It!? Computer-Use Agents Exhibit Blind Goal-Directedness (ICLR 2026 · similarity 0.85)
- SD-Search: Hindsight Self-Distillation for Search-Augmented Reasoning (NeurIPS 2026 · similarity 0.85)
- AI Alignment with Changing and Influenceable Reward Functions (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
