Navigating Model Uncertainty and the Rashomon Effect:

From Theory and Tools to Applications and Impact

at AAAI 2026

About

In many real-world AI problems, there is not a single “best” model but rather a multiplicity of models that achieve comparable performance. This phenomenon, known as the Rashomon Effect, raises fundamental questions about uncertainty, diversity, and model choice. Understanding and leveraging model multiplicity has become crucial for building trustworthy, reliable, and responsible AI systems.

This half-day workshop will bring together researchers and practitioners to explore the theoretical foundations, algorithmic methods, and societal implications of predictive multiplicity. We will highlight both classical approaches and emerging challenges in the era of generative models, and will also host community-building sessions (see below).

Call For Papers

Topics of Interest

We welcome contributions about the Rashomon Effect that include:

Position papers offering perspectives on the role of multiplicity in AI practice, including: deployment, governance, and policy.
Connections between the Rashomon Effect and generative models.
Applications and impacts of predictive multiplicity in deployed systems.
Connections between the Rashomon Effect and (i) uncertainty, (ii) explainability, (iii) fairness, (iv) privacy, and (v) robustness.
Theoretical and algorithmic approaches for understanding, constructing, auditing, and navigating Rashomon sets.

Submission guidelines

We invite submissions of 4-6 pages (plus unlimited references and appendices). Submissions should be anonymized, follow the official template and will be handled via OpenReview, with double-blind reviewing. Three tracks are offered:

Main track: new research contributions (including early-stage ideas).
Published paper track: recently published or accepted work for presentation.
Position paper track: commentary of broader interest on the current state of model multiplicity.

Reviewing is double-blind. This workshop is non-archival. Accepted papers will be available on the workshop website and OpenReview.

For the main track, submissions can be exploratory in nature and contain initial novel ideas and results. Submissions should clearly articulate the research problem, methodology, results (if available), and contributions to the field. We welcome submissions undergoing concurrent peer review elsewhere. Papers presented or scheduled for presentation at non-archival venues, such as other workshops, are permitted for submission. However, it is the authors' responsibility to verify compliance with other venues' policies.

We also invite position papers that provide a commentary on the current state of the Rashomon Effect, model uncertainty, or model multiplicity. Such papers may offer perspectives, highlight challenges and blind spots, propose future research directions, or bring interdisciplinary insights. We especially welcome contributions that connect these themes to broader issues of trustworthiness, reliability, and real-world deployment.

Submissions that have been previously published or accepted for publication in peer-reviewed conferences or journals should be submitted to the published paper track. These papers will not undergo the review process and will instead be evaluated by ACs based on the paper's topic and its fit for the workshop.

Please note that while we welcome submissions to the published paper track, acceptance there will be highly selective. Priority will be given to diversity of topics and to submissions in the main track.

Important Dates

Submission Opens: September 15, 2025
Submission Deadline: October 15, 2025
Notification to Authors: November 5, 2025
Workshop Date: January 27, 2026

All deadlines are 11:59PM UTC-12:00 (AoE).

Perspectives from the Community

In addition to paper presentations, we will host a special “Perspectives from the Community” session featuring short talks that highlight researchers’ views on the Rashomon Effect, model uncertainty, and model multiplicity. This session is intended to provide a venue for sharing perspectives on the current state of the field, ongoing challenges, open problems, and other thought-provoking ideas. We also welcome reflections on career paths, including experiences entering the field, finding research directions, or navigating the transition after graduation. We particularly encourage contributions from junior researchers, including those without a formal paper submission.

To participate, please submit a one-page overview (submission link) outlining your perspective. The form is due November 30. Accepted overviews will be invited for short talks during the workshop to foster open discussion and community building.