AAAI 2026 accepted papers on Large language models
695 of the 4,417 papers accepted at AAAI 2026 (15.7%) are on Large language models, compared with 14.4% at AAAI 2025. They fall into 7 topics.
See AAAI 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Language, safety and interpretability · 175 papers
- When Truth Is Overridden: Uncovering the Internal Origins of Sycophancy in Large Language Models (9 citations)
- MPMA: Preference Manipulation Attack Against Model Context Protocol (6 citations)
- CharBench: Evaluating the Role of Tokenization in Character-Level Tasks (3 citations)
- DEALT: LLM-driven Diversity-Enhanced Data Augmentation for Long-Tail Text Classification (2 citations)
- Do LLMs Feel? Teaching Emotion Recognition with Prompts, Retrieval, and Curriculum Learning (2 citations)
- Eguard: Defending LLM Embeddings Against Inversion Attacks via Text Mutual Information Optimization (2 citations)
- Not All Inconsistency Is Equal: Decomposing LVLM Uncertainty into Belief Divergence and Belief Conflict (2 citations)
- SlideTailor: Personalized Presentation Slide Generation for Scientific Papers (2 citations)
- SpeakerLM: End-to-End Versatile Speaker Diarization and Recognition with Multimodal Large Language Models (2 citations)
- ViDia2Std: A Parallel Corpus and Methods for Low-Resource Vietnamese Dialect-to-Standard Translation (2 citations)
- Where and What Matters: Sensitivity-Aware Task Vectors for Many-Shot Multimodal In-Context Learning (2 citations)
- AHAMask: Reliable Task Specification for Large Audio Language Models Without Instructions (1 citation)
and 163 more in the interactive map.
Agents, code and math · 126 papers
- A Theory of Adaptive Scaffolding for LLM-Based Pedagogical Agents (8 citations)
- EoH-S: Evolution of Heuristic Set Using LLMs for Automated Heuristic Design (6 citations)
- MedOmni-45°: A Safety–Performance Benchmark for Reasoning-Oriented LLMs in Medicine (4 citations)
- Driving with Regulation: Trustworthy and Interpretable Decision-Making for Autonomous Driving with Retrieval-Augmented Reasoning (3 citations)
- LocalBench: Benchmarking LLMs on County-Level Local Knowledge and Reasoning (3 citations)
- MCTSr-Zero: Self-Reflective Psychological Counseling Dialogues Generation via Principles and Adaptive Exploration (3 citations)
- AMS-IO-Bench and AMS-IO-Agent: Benchmarking and Structured Reasoning for Analog and Mixed-Signal Integrated Circuit Input/Output Design (2 citations)
- Beyond ReAct: A Planner-Centric Framework for Complex Tool-Augmented LLM Reasoning (2 citations)
- Evaluating LLMs for Police Decision-Making: A Framework Based on Police Action Scenarios (2 citations)
- FIXME: Towards End-to-End Benchmarking of LLM-Aided Design Verification (2 citations)
- HouseTune: Two-Stage Floorplan Generation with LLM Assistance (2 citations)
- Incoherence as Oracle-less Measure of Error in LLM-Based Code Generation (2 citations)
and 114 more in the interactive map.
Retrieval and knowledge · 124 papers
- PathRAG: Pruning Graph-based Retrieval Augmented Generation with Relational Paths (12 citations)
- ArchRAG: Attributed Community-based Hierarchical Retrieval-Augmented Generation (8 citations)
- ProgRAG: Hallucination-Resistant Progressive Retrieval and Reasoning over Knowledge Graphs (5 citations)
- Cog-RAG: Cognitive-Inspired Dual-Hypergraph with Theme Alignment Retrieval-Augmented Generation (4 citations)
- From Dialogue to Destination: Geography-Aware Large Language Models with Multimodal Fusion for Conversational Recommendation (4 citations)
- LeanRAG: Knowledge-Graph-Based Generation with Semantic Aggregation and Hierarchical Retrieval (4 citations)
- Personalize Before Retrieve: LLM-based Personalized Query Expansion for User-Centric Retrieval (4 citations)
- ComoRAG: A Cognitive-Inspired Memory-Organized RAG for Stateful Long Narrative Reasoning (3 citations)
- TruthfulRAG: Resolving Factual-level Conflicts in Retrieval-Augmented Generation with Knowledge Graphs (3 citations)
- Beyond Chains: Bridging Large Language Models and Knowledge Bases in Complex Question Answering (2 citations)
- CliCARE: Grounding Large Language Models in Clinical Guidelines for Decision Support over Longitudinal Cancer Electronic Health Records (2 citations)
- Debate over Mixed-knowledge: A Robust Multi-Agent Reasoning Framework for Incomplete Knowledge Graph Question Answering (2 citations)
and 112 more in the interactive map.
Reasoning and chain of thought · 119 papers
- TokenPowerBench: Benchmarking the Power Consumption of LLM Inference (8 citations)
- MedS³: Towards Medical Slow Thinking with Self-Evolved Soft Dual-sided Process Supervision (3 citations)
- NeSTR: A Neuro-Symbolic Abductive Framework for Temporal Reasoning in Large Language Models (3 citations)
- BadThink: Triggered Overthinking Attacks on Chain-of-Thought Reasoning in Large Language Models (2 citations)
- CP-Router: An Uncertainty-Aware Router Between LLM and LRM (2 citations)
- DeepOR: A Deep Reasoning Foundation Model for Optimization Modeling (2 citations)
- ExtendAttack: Attacking Servers of LRMs via Extending Reasoning (2 citations)
- GenPRM: Scaling Test-Time Compute of Process Reward Models via Generative Reasoning (2 citations)
- Joint Evaluation of Answer and Reasoning Consistency for Hallucination Detection in Large Reasoning Models (2 citations)
- A Reasoning Paradigm for Named Entity Recognition (1 citation)
- AdaCuRL: Adaptive Curriculum Reinforcement Learning with Invalid Sample Mitigation and Historical Revisiting (1 citation)
- AdaMCoT: Rethinking Cross-Lingual Factual Reasoning Through Adaptive Multilingual Chain-of-Thought (1 citation)
and 107 more in the interactive map.
Instruction and fine-tuning · 52 papers
- RAG-Enhanced Collaborative LLM Agents for Drug Discovery (3 citations)
- STAR-1: Safer Alignment of Reasoning LLMs with 1K Data (3 citations)
- PoeTone: A Framework for Constrained Generation of Structured Chinese Songci with LLMs (2 citations)
- A Course Correction in Steerability Evaluation: Revealing Miscalibration and Side Effects in LLMs (1 citation)
- A Robust Unlearning Method with Adaptive Knowledge Guidance and Memory Preservation (1 citation)
- AsFT: Anchoring Safety During LLM Fine-Tuning Within Narrow Safety Basin (1 citation)
- Evaluating, Synthesizing, and Enhancing for Customer Support Conversation (1 citation)
- From Macro to Micro: Probing Dataset Diversity in Language Model Fine-Tuning (1 citation)
- Ghost in the Transformer: Detecting Model Reuse with Invariant Spectral Signatures (1 citation)
- PRAGWORLD: A Benchmark Evaluating LLMs’ Local World Model Under Minimal Linguistic Alterations and Conversational Dynamics (1 citation)
- Selection of LLM Fine-Tuning Data Based on Orthogonal Rules (1 citation)
- Towards Acyclic Preference Evaluation of Language Models via Multiple Evaluators (1 citation)
and 40 more in the interactive map.
Alignment and preferences · 51 papers
- Multi-Value Alignment for LLMs via Value Decorrelation and Extrapolation (3 citations)
- Lost in Benchmarks? Rethinking Large Language Model Benchmarking with Item Response Theory (2 citations)
- Safety Alignment of Large Language Models via Contrasting Safe and Harmful Distributions (2 citations)
- Align to Structure: Aligning Large Language Models with Structural Information (1 citation)
- Align³GR: Unified Multi-Level Alignment for LLM-based Generative Recommendation (1 citation)
- ALPHA: Action-Based Learning for Pluralistic Human Alignment in Large Language Models (1 citation)
- Beyond Step Pruning: Information Theory Based Step-level Optimization for Self-Refining Large Language Models (1 citation)
- Multi-Metric Preference Alignment for Generative Speech Restoration (1 citation)
- ParetoHqD: Fast Offline Multiobjective Alignment of Large Language Models Using Pareto High-Quality Data (1 citation)
- Preference Robustness for DPO with Applications to Public Health (1 citation)
- RankList – a Listwise Preference Learning Framework for Predicting Subjective Preferences (1 citation)
- SMPRO: Self-Supervised Visual Preference Alignment via Differentiable Multi-Preference Multi-Group Ranking (1 citation)
and 39 more in the interactive map.
AI and society · 48 papers
- Assessing the Quality of AI-Generated Exams: A Large-Scale Field Study (5 citations)
- Benchmarking XAI Explanations with Human-Aligned Evaluations (2 citations)
- Too Sure for Our Own Good: A User Study on AI Confidence and Human Reliance (2 citations)
- Characterizing AI Manipulation Risks in Brazilian YouTube Climate Discourse (1 citation)
- Democratizing Writing Support with AI: Insights from One Year of Real-World Interactions with an Open-Access Writing Feedback Tool (1 citation)
- F.A.C.U.L.: Language-Based Interaction with AI Companions in Gaming (1 citation)
- A Tale of Two Identities: An Ethical Audit of AI-Crafted Synthetic Personas
- Align When They Want, Complement When They Need! Human-Centered Ensembles for Adaptive Human-AI Collaboration
- AncientBench: Towards Comprehensive Evaluation on Excavated and Transmitted Chinese Corpora
- Attribution Analysis-based Concept Alignment: A Human-in-the-loop Data Debugging Framework
- AutoGameUI: Constructing High-Fidelity GameUI via Multimodal Correspondence Matching
- AutoMalDesc: Large-Scale Script Analysis for Cyber Threat Research
and 36 more in the interactive map.