Introduction
NGEN-AI aims to provide an inclusive, transparent, and constructive review process. Reviewers should consider the nature and maturity of each submission when assessing its merit, including full research papers, short papers, demos, posters, and early-idea contributions.
Reviews should be open-minded, fair, clear, and professional. Every score must be supported by specific comments that help the authors understand both the strengths of their work and the areas that require improvement.
Reviewer Expectations
Program Committee members are expected to write their own reviews. A sub-reviewer may be consulted when appropriate, but the assigned reviewer must critically revise the feedback, express it in their own words, determine the final scores, and remain accountable for the review. Any sub-reviewer must be credited in the designated field of the review form.
Submit a thorough and careful evaluation. Use the bidding process to select papers close to your expertise, and contact the Program Chairs if an assignment falls materially outside it.
If a paper does not yet meet the acceptance standard, explain what would be needed to make it suitable for NGEN-AI while recognizing the space and maturity constraints of its submission category.
Avoid indecisive scoring. Take a clear position and justify it with evidence from the submission. Scores and written comments should be consistent with one another.
Promptly inform the Program Chairs if you identify possible plagiarism, concurrent submission, fabricated results, inappropriate use of data, conflicts of interest, or other ethical concerns. Do not investigate authors independently or raise accusations in comments visible to authors.
Monitor the other reviews and the online discussion for assigned submissions. Reviews and scores may be updated during the review and discussion periods. When appointed as a discussion leader, help the committee identify points of agreement and resolve material differences before a decision.
Paper Review Criteria
Assess how well the paper fits the Call for Papers and whether its contribution matters to next-generation AI systems research or practice. Relevant emerging areas should be considered even when they are not named explicitly in the call. If you judge a paper out of scope, explain why.
- Full research papers should demonstrate meaningful implications for AI systems research or practice and clearly interpret their findings.
- Short and early-idea papers should show that the completed work could have a meaningful impact and should clearly motivate the proposed direction.
- Demo papers should present a relevant working system, tool, or prototype and explain its value to the community.
- Poster papers should communicate a focused, relevant contribution suitable for discussion and feedback at the conference.
Determine whether the claims and contributions are supported through appropriate and rigorous methods. Full papers should describe their methods clearly and address limitations and threats to validity. Preliminary work should still offer a sound proposal, credible motivation, and an appropriate plan for evaluation. Demo claims should be supported by evidence that the system works as described.
Evaluate how the work advances the existing body of knowledge. Originality does not require a surprising result or a complicated solution: a simple, well-supported contribution may still be valuable. Replications and confirmations of earlier findings are welcome when their relationship to prior work and their added value are explained.
Suggest important missing references where helpful, but avoid requesting citations to your own work. If your work is uniquely relevant, consult the Program Chairs and identify suitable alternatives when possible.
Consider whether the paper is clearly written and logically organized, uses understandable English, avoids major ambiguities, presents readable figures and tables, and follows the required format. Distinguish presentation issues that can be corrected from flaws that undermine the contribution.
Reproducibility and Open Science
Where applicable, assess whether the paper provides enough methodological and experimental detail to support verification, replication, or reuse. Consider whether practical artifacts—such as code, data, prompts, models, configurations, evaluation protocols, or teaching material—are shared appropriately.
Artifact sharing may be limited by privacy, de-identification risk, proprietary industry data, trade secrets, security concerns, licensing, or the preliminary nature of the work. Judge availability in the context of the paper type and do not treat a justified absence of artifacts as an automatic reason for rejection.
When artifacts are provided, consider whether access is suitably anonymized for review and whether the paper includes enough instructions to understand and evaluate them.
General Review Guidelines
Every review should:
- Address the review criteria above and relate the assessment to the paper’s submission category.
- Identify the paper’s main strengths before detailing its weaknesses.
- Provide constructive, actionable suggestions, including for papers that are not accepted.
- Explain criticism with clear reasoning and use a polite, supportive tone.
- Suggest relevant standards, literature, benchmarks, or open repositories when they would genuinely improve the work.
- Avoid requiring citations to the reviewer’s own work unless there is a compelling, chair-approved reason and no suitable alternative.
- Keep confidential comments for the Program Chairs separate from comments intended for the authors.