The Generic ML/AI 2026 review form, explained

How Generic ML/AI reviewers read your paper

A generic format that combines the fields most major ML/AI conferences share: a faithful summary, substantiated strengths and weaknesses across soundness, presentation, significance and originality, questions that could change the assessment, a limitations check, three 1–4 sub-scores, a 1–10 overall recommendation and a 1–5 confidence. It is not any venue's official form; for a specific venue, check its reviewer guidelines.

• Constructive, specific and substantiated; every criticism names what, where and why.
• Lead the weaknesses with the one or two issues that drive the recommendation.
• Do not reject solely for not beating state of the art, or claim a lack of novelty without naming the prior work.
• Every fair/poor sub-score must be justified by a weakness.

Where the bar sits. Generic calibration, not a venue statistic: selective ML/AI conferences accept roughly 20–30% of submissions. On this 1–10 scale a mean of 6 or more is competitive, 5–6 is borderline, and below 5 is a likely reject.

The reviewer form, field by field

In the order reviewers see it, with every score and its official wording.

1. 01

   Summary

   Free text

   Summarize the problem, approach and main claimed contributions in your own words. No critique; the authors should agree with it.

2. 02

   Strengths

   List of points

   Reasons to accept: soundness, clarity, significance and originality, each substantiated with evidence from the paper.

3. 03

   Weaknesses

   List of points

   Reasons to reject, most important first. Be specific: say what is wrong or missing, where, why it matters for the claims, and what would fix it. Separate fatal flaws from fixable issues.

4. 04

   Questions

   List of points

   Questions and requests for the authors whose answers could change your assessment; say how.

5. 05

   Limitations

   Free text

   Are limitations and potential negative societal impacts adequately discussed? Write "yes" if so; otherwise give constructive suggestions. Do not penalize candor.

6. 06

   Soundness

   Score · 1–4

   Are the claims supported by correct and rigorous theory or experiments?

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
7. 07

   Presentation

   Score · 1–4

   Clarity, organization, and positioning relative to prior work.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
8. 08

   Contribution

   Score · 1–4

   Significance and originality of the contribution to the field.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
9. 09

   Overall

   Score · 1–10

   Overall recommendation. Use the borderline scores (5, 6) when the case really is balanced.

   Score	Meaning
   10	Award quality technically flawless, with groundbreaking impact.
   9	Very strong accept excellent paper, among the best at the venue.
   8	Strong accept technically strong, with high impact on its sub-area.
   7	Accept technically solid, with clear impact; minor weaknesses only.
   6	Weak accept reasons to accept outweigh reasons to reject.
   5	Borderline merits and weaknesses roughly balance.
   4	Weak reject reasons to reject outweigh reasons to accept.
   3	Reject significant flaws in soundness, evaluation or clarity.
   2	Strong reject major technical flaws or well-known results.
   1	Trivial or wrong fundamentally flawed or out of scope.

10. 10

    Confidence

    Score · 1–5

    How certain you are about your assessment.

    Score	Meaning
    5	You are absolutely certain about your assessment. You are very familiar with the related work and checked the math/other details carefully.
    4	You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.
    3	You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.
    2	You are willing to defend your assessment, but it is quite likely that you did not understand the central parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.
    1	Your assessment is an educated guess. The submission is not in your area or the submission was difficult to understand. Math/other details were not carefully checked.

Area chair meta-review

Decide from the arguments in the reviews, not the mean score.

• Accept (oral)
• Accept (poster)
• Reject

Policy on LLMs in reviewing