Call for Papers

The Exploration in AI Today Workshop will focus on the question of how exploration can enable progress in modern AI. Answering this will require addressing both fundamental algorithmic questions, as well as application-driven challenges, such as (but not limited to):
  • How might we leverage prior knowledge—for example, in the form of a simulator or pretrained model—to enable effective exploration, and how might biases present in such prior knowledge affect the resulting exploration?
  • How can we effectively explore high-dimensional search spaces? In particular, how might we exploit the structure inherent in many high-dimensional settings of interest to enable effective exploration?
  • Uncertainty quantification historically has played a critical role in exploration: how can we quantify uncertainty effectively to direct exploration in high-dimensional settings such as language?
  • How can modern learning techniques (e.g. diffusion, in-context learning) be utilized in exploration?
  • How can we develop agents—LLM agents, robots, and beyond—that effectively explore their environment online to determine how to accomplish a desired goal?
  • Which experiments should we run in order to collect the most informative data to update our models—whether a language model, world model, model of biological interactions, or more?
  • What role can exploration play in reasoning, e.g., in searching and generating chains of thought?
We welcome paper submissions addressing such questions, as well as other challenges related to exploration in AI, across domains such as:
  • LLMs, AI agents, and AI alignment.
  • Robotics and control.
  • AI for science and drug discovery.
  • Reinforcement learning.
  • Active learning, experiment design, and Bayesian optimization.
  • Recommendation systems and A/B testing.


Best Paper Awards

We will be awarding a total of four best paper awards, with one award given to a paper under each of our four submission tracks: Language Modeling, Robotics, AI for Science, and Theory. Each winning paper will be awarded a cash prize of $500 USD, and will be highlighed with a spotlight talk at the workshop.


Important Dates

  • Submissions open: 13 April 2025
  • Submission deadline: 26 May 2025 AOE (NeurIPS supplementray deadline is 22nd)
  • Notification of acceptance: 30 June 2025
  • Posters and camera ready: 9 July 2025 AOE


Submisison Guidelines

All submissions must be in PDF format, and in a single PDF file include, in this suggested order:

  1. The submitted paper content
  2. Paper references

The authors may optionally choose to include some or all of the technical appendices in the same PDF above. But those included parts cannot be changed after the full submission deadline.

The main text of a submitted paper is limited to seven content pages, including all figures and tables. Additional pages containing references, and the optional technical appendices do not count as content pages. The maximum file size for a main paper submission is 50MB.

You must format your submission using the ICML 2025 EXAIT LaTeX style file (Based on NeurIPS 2025 LaTeX style file), which includes a “preprint” option for non-anonymous preprints posted online. Submissions that violate the ICML 2025 EXAIT style (e.g., by decreasing margins or font sizes) or page limits may be rejected without further review. Papers may also be rejected without consideration of their merits if they fail to meet the submission requirements, as described in this document.