Go to OpenReview News homepageOpenReview Hosts Record-Breaking AAAI 2026 Conference with Pioneering AI Review System
Article: OpenReview is hosting the [AAAI conference](https://aaai.org/conference/aaai/aaai-26/) for the second time, with AAAI 2026 shattering previous records by receiving 30,948 submissions—marking an unprecedented scale for the premier AI conference. The massive growth in submissions underscores the continued expansion of AI research worldwide and presents significant challenges for maintaining high-quality peer review at scale.
To address these challenges, AAAI 2026 is piloting an innovative AI-assisted review system using a large frontier reasoning model from OpenAI. "We are very interested in understanding the capabilities and limitations of state-of-the-art AI tools in assisting with the peer-review process, with human experts in control at every step," says Joydeep Biswas, AAAI 2026 Associate Program Chair, who led the effort to develop and implement the system. The experiment incorporates AI assistance at two key stages: generating clearly-labeled and "non-evaluative" AI reviews that supplement (but do not replace) human reviewers, and facilitating post-review discussions by assessing reviewer agreement and drafting meta-reviews as additional input for senior program committee members. The system employs a multi-step workflow with custom tools for technical accuracy checking, literature search, and results verification, while maintaining strict data confidentiality—with all decisions and recommendations remaining entirely in human hands. Authors, reviewers, and committee members will provide feedback on the AI reviews to inform future improvements, with findings to be shared in a technical report to the broader research community.
OpenReview’s flexible and extensible infrastructure made this large-scale experiment possible. Our platform was customized in close collaboration with the AAAI Program Chairs, who worked directly with the OpenReview engineering team to design and deploy a review workflow that integrates AI components while preserving AAAI’s rigorous review standards. OpenReview’s modular architecture allowed us to incorporate LLM-generated reviews, reviewer-assistance tools, and custom evaluation dashboards into the existing peer review pipeline—without disrupting the standard submission, bidding, and discussion processes. The same fine-grained access controls that protect reviewer anonymity and conflicts of interest were extended to govern AI data access and transparency. Additionally, OpenReview systems proved invaluable for detecting dual and overlapping submissions, both within AAAI and across venues, helping maintain the integrity of the review process at this unprecedented scale. This partnership highlights how OpenReview’s configurable workflows and AI-integration APIs empower conferences like AAAI to innovate responsibly—testing the future of peer review while ensuring that human judgment and community oversight remain at the core of scientific decision-making.
Loading