La investigación detrás del método

Trabajos aceptados tras revisión por pares en workshops, simposios y programas bridge, entre ellos los de NeurIPS, ICML e ICLR, 2025-2026.

105 trabajos aceptados en AIXC Research; de ellos, 5 de ciencia de materiales y 42 de métodos para ciencia con IA (registro a 17-10-2026).

El total recoge todo el registro de AIXC Research; aquí solo se listan las dos categorías del vertical de materiales.

Materials science

Destacado

The Architectural Immune System: A Framework for Correcting Synthetic Fallacies in AI-Driven Science

David Scott Lewis, Enrique Zueco, Anar Batkhuu, Haley Yi

Agents4Science 2025, presidido desde Stanford University

Estudio de caso de descubrimiento de materiales con un AI Scientist que detecta y corrige sus propias falacias. Ver en OpenReview

Graph of Thoughts Nanobiomaterials Assistants: Towards Logical, Tool-Augmented, and Multi-Agent Reasoning in Scientific LLMsAAAI 2026 Bridge LMReasoning · bridge
ACHT-World: Causal World Models for Closed-Loop Self-Driving LaboratoriesAI4X-AC 2026 Oral · symposium
Physics-Informed Surrogates in a Verified Agentic Pipeline for Robust Molecular SimulationML4Molecules 2025 Poster · workshop
Recursive Self-Correction in Neuro-Symbolic Retrosynthesis: Grounding LLM Chemistry in Deterministic LogicNEUREASON 2026 · symposium

Methods for AI-driven science

Neural Cellular Automata and the PD-NCA AutoLab: A Differentiable Artificial Life Testbed for Autonomous Research AgentsAI4Research · workshop
PD-NCA AutoLab: Self-Organizing Neural Intelligence as a Testbed for Autonomous ScienceAI4X-AC 2026 Poster · symposium
Recursive Epistemic Engines for Verifiable Open-Ended Scientific AgentsICML 2026 AI4Math Workshop Poster · workshop
Scientific CI/CD for Self-Modifying Discovery Agents: Statistical Gödel Gates, Capacity Budgets, and Domain VerifiersICML 2026 AI4Math Workshop Poster · workshop
Tacit Verification Apprenticeship for Self-Evolving Scientific AgentsICML 2026 AI4Math Workshop Poster · workshop
RSI for Science: A Verifier-First Framework for AI ScientistsICML2026-AI4Science Poster · workshop
NOVA-Test: Auditing LLM-Generated Research Hypotheses with Structural Analogy, Prior-Work, and Falsifiability TestsICML26-Testing Poster · workshop
Does Recursive Self-Improvement Transfer Across Physical Regimes?ML4PS 2026 Workshop · workshop
Axioms as Scaffolding: Typed Symbolic Regression for Extrapolative Scientific DiscoveryNEUREASON 2026 · symposium
Measure Less, Discover More: Cost-Aware Objective Scheduling for Multi-Objective Self-Driving LabsNEUREASON 2026 · symposium
Choosing Experiments, Replication, and Recalibration under Measurement DriftNeurIPS 2026 Workshop AgenticLS Poster · workshop
Evidence Eligibility Shapes Reward Comparisons in Scientific Experiment AllocationNeurIPS 2026 Workshop AgenticLS Poster · workshop
Execution Signals for Scientific Agent Rewards: Prediction, Selection, and TransferNeurIPS 2026 Workshop AgenticLS Poster · workshop
What Must a Cached Agent Decision Observe? An Evidence-Freshness Benchmark for Molecular WorkflowsNeurIPS 2026 Workshop AgenticLS Poster · workshop
What a Tool-Choosing Molecular Agent Should Retain: Support, Feedback and Tool BudgetNeurIPS 2026 Workshop AgenticLS Poster · workshop
Target Alignment in Sequential Molecular Model ApprovalNeurIPS 2026 Workshop AgenticLS Spotlight · workshop
Count Every Validation: Auditing Hidden Attempts in Recursive AI ScientistsNeurIPS2026-AI4Science Poster · workshop
Escaping the Plausibility Trap: A Verified and Abstract Memory Framework for Trustworthy Agentic Knowledge DiscoveryACM CIKM 2025 Workshop TrustKDD · workshop
Graph of Thoughts for High-Integrity Hypothesis Reasoning and ValidationGMLLM'25 Oral · workshop
LabMemo: Concept‑Level Memory for Autonomous Scientific DiscoveryIEEE IROS 2025 Workshop AIR4S · workshop
Provably Safe Robotics for Autonomous Science: Real-Time Formal Verification with the Verified Agentic Pipeline (VAP)IEEE IROS 2025 Workshop AIR4S · workshop
Teachability@Instruments: Learn-on-the-Job Agentic Operators for Advanced Scientific FacilitiesIEEE IROS 2025 Workshop AIR4S · workshop
The Ascendant AI Scientist: An Empirical Validation of the Reliability Gap on the Path to Autonomous DiscoveryIEEE IROS 2025 Workshop AIR4S · workshop
The Autonomous Experimenter: Closed-Loop Discovery in Self-Driving Laboratories via Active Causal Hypothesis Testing (ACHT)IEEE IROS 2025 Workshop AIR4S · workshop
SOKRATES: Distilling Symbolic Knowledge into Option-Level Reasoning via Solver-Guided Preference OptimizationAAAI 2026 Bridge LMReasoning · bridge
Does LLM Co-Evolution Pay Off in RTL Verification? A Cost-Normalised Audit on Ten RTLLM DesignsAI4ChipDesign 2026 Poster · workshop
A Retention-Gated Self-Evolving Memory for LLM Agents: Pre-Registered Backward-Transfer Evidence and Design RulesCL4FMAgents @ NeurIPS 2026 · workshop
Can Self-Improving Agents Retain the Ability to Improve?CL4FMAgents @ NeurIPS 2026 · workshop
A Self-Improving Forecasting Agent That Chooses Its Own Validation Window: Design Rules from a Two-Stage Pre-Registered StudyFMTS 2026 Poster · workshop
Governed Self-Improvement for Logical Reasoning: Edit-Time Governance for Developmental ConsistencyICLR 2026 Workshop LLM Reasoning · workshop
Memory Transplants for LLM Agents: Disentangling Architecture and Content Transfer under a Code-to-Math ShiftICLR 2026 Workshop MemAgents · workshop
PATCH-BT: Agency-Preserving Behavior-Tree Self-Repair for Shared-Autonomy TelepresenceIEEE ICRA 2026 Workshop Embodied AI and Telepresence · workshop
Which Memory Operation Drives Recovery? A Factorial Study of Retrieve, Write, and Manage Adaptation under Domain ShiftLLA 2026 Poster · workshop
What Validation Loss Misses in Learned Diffusion OperatorsML4PS 2026 Workshop · workshop
Recursive World-Model Self-Repair in PDDL Domains: Verifier Coverage Is the Binding Constraint, and Under-Verified Loops Stall Rather Than DivergeNEmo Poster · workshop
When Memories Collide: Associative Interference Dynamics in Lifelong Agent MemoryNFAM 2026 Poster · workshop
Belief Ledgers: Claim-Level Continual Revision for World ModelsNeurIPS 2026 Workshop CWM · workshop
Ranking Is Not a Deployment Policy: Coverage Drift in Learned Reliability Gates for Prediction-Market InputsNeurIPS 2026 Workshop FMTS · workshop
Replication and Delayed Feedback in Forecasting EnsemblesNeurIPS 2026 Workshop FMTS · workshop
Scenario Diversity Is Not Failure Foresight: A Frozen Audit of Quality-Diversity Scenario LibrariesNeurIPS 2026 Workshop FMTS · workshop
Change Passports: A Machine-Readable Standard for Auditable Artifact Promotion in Continually-Learning AgentsNeurIPS-CLEA 2026 Poster · workshop
Auditing Recursive Model Promotion: Gate Interfaces, Decision Coverage, and Rejection MemoryNeurIPS2026-AI4Science Poster · workshop

Fuente: registro de publicaciones de AIXC Research a 17-10-2026, construido desde OpenReview.