Call for Papers

The AI Scientist Workshop @ NeurIPS 2026 invites submissions advancing the methodologies that transform Large Language Models and agentic AI systems into rigorous scientific reasoners. We focus on the training strategies and inference-time reasoning required for autonomous discovery across the full scientific lifecycle—from inspiration retrieval and hypothesis generation, to closed-loop experimental refinement and self-correction.

Topics of Interest

We welcome submissions on a broad range of topics, including but not limited to:

  1. Inference-Time Reasoning & Search: Test-time compute techniques (e.g., Tree of Thoughts, Monte Carlo Tree Search) for multi-step logical planning and complex experimental design.
  2. Training, Data & Alignment for Science: Construction of scientific reasoning corpora and synthetic data for low-resource scientific domains (e.g., theorem proving, optimization, experimental design); Supervised Fine-Tuning (SFT), Reinforcement Learning with verifiable rewards (RLVR), and reward models trained from experimental feedback. Methods that align model outputs with physical constraints, mathematical logic, and iterative experimental signals.
  3. Agentic Frameworks & Tool Use: Architectures supporting autonomous tool usage (simulators, code interpreters, lab equipment) and memory mechanisms for long-horizon scientific workflows.
  4. Hypothesis Generation & Ranking: Algorithms for generating novel scientific hypotheses and ranking them based on plausibility, novelty, and verifiability.
  5. Benchmarks & Evaluation: Moving beyond rote memorization to evaluate true reasoning capabilities in out-of-distribution scientific discovery tasks.

Submission Guidelines

  • Paper length: 4–8 pages (excluding references and appendices)
  • Format: Use the NeurIPS 2026 LaTeX template
  • Anonymization: Submissions must be anonymized for double-blind review
  • Dual-submission policy (non-archival): Accepted papers will not be archived in formal proceedings, welcoming ongoing and unpublished work and allowing for future submission to other venues. We welcome submissions currently under review at other venues, including ICLR 2027. However, we cannot consider work that has been previously published or accepted for publication at any peer-reviewed venue. This policy on dual submissions remains in effect throughout the entire reviewing process.

Important Dates (Tentative)

All deadlines are 23:59 AoE (Anywhere on Earth)

  • Submission deadline: August 22, 2026
  • Author notification: September 22, 2026
  • Camera-ready deadline: November 13, 2026
  • Workshop date: December 11 or 12, 2026 (specific day TBD by NeurIPS)

Submission Process

All submissions should be made through the workshop’s OpenReview portal (link to be announced).

Presentation Format

Accepted papers will be presented as posters during the workshop. Selected papers will be invited for oral spotlight talks.

As part of our commitment to recognizing outstanding research, we will present 1–3 Best Paper Award(s) to the most impactful contributions.

Questions?

For inquiries, please contact: ai-scientist-workshop@googlegroups.com.