動筆回覆前,建議先把審稿人的「觀察描述」和「結論判斷」分開來看。比如 “This paragraph is repetitive(本段內容存在重複)” 是針對寫作品質的具體意見,而 “The authors used AI(作者使用了人工智慧)” 則是對稿件製備過程的判斷。很多時候,要求補充 AI 使用聲明只是期刊的常規流程,不代表已經定性你違規使用了 AI。
不同類型的質疑,回覆時需要回應的核心資訊也不同:
審稿人關注點
回覆中需明確的核心資訊
The text appears AI-generated. (文本疑似 AI 生成)
使用了哪些工具、用於哪些任務、有哪些寫作過程紀錄可佐證
The writing is generic or repetitive. (表述平淡、內容重複)
你如何修改對應段落、如何強化內容與研究主題的關聯
The AI declaration is missing or incomplete. (AI 聲明缺失或不完整)
如果查核後確認確實使用了 AI,請先對照期刊政策確認使用方式是否在允許範圍內。例如 Taylor & Francis 出版集團政策明確區分了「允許的語言編修」和「禁止的稿件/章節初稿生成」,後者即便揭露也屬於違規。也就是說,主動揭露不代表所有使用方式都能被接受。
情境 1:有限度使用 AI/補充遺漏的 AI 揭露
一份合格的 AI 揭露,必須明確四個要素:工具名稱、具體任務、涉及內容、人工審核過程。如果 AI 實際參與了段落翻譯、論點重組等工作,就不能籠統描述為「grammar only(僅用於語法校對)」。
參考回覆(說明有限使用)
We used [tool and recorded version] to [specific task] in [sections]. We reviewed the resulting changes against [original text or research records], checking [specific features checked]. The tool’s role was limited to [accurate scope]. We expanded the declaration at [page and lines] to describe this use and the authors’ review.
我們使用了 [工具名稱及版本號] 在 [章節名稱] 部分完成了 [具體任務,如「摘要的語言編修與句式優化」]。我們將 AI 輸出結果與 [原始文本/研究紀錄] 逐一比對,查核了 [具體查核項,如「專業術語準確性、論點強度、科學含義」],確認該工具的作用僅限於 [準確範圍,如「語言表達層面的優化」]。我們已在 [頁碼 + 行號] 處補充了相關 AI 使用聲明,詳細說明使用情境與作者審核過程。
參考回覆(致歉並補正揭露)
Our original submission omitted the required disclosure of [specific use]. We apologize for this omission. We added the tool, purpose, affected sections, and review process to [required location] and corrected [applicable submission field]. The revised statement reads: “[Insert the disclosure.]”
我們在此致歉:初次提交時遺漏了針對 [具體使用情境] 的必要揭露。目前我們已在 [指定位置,如「正文 AI 聲明章節、投稿系統資訊欄」] 補充了相關內容,說明所用工具、使用目的、涉及部分與審核流程,並同步修正了投稿系統中的對應欄位。修訂後的揭露內容如下:「插入完整揭露文本」
AI 工具可能生成看似合理、但實際不存在或不匹配的參考文獻。收到相關質疑時,先逐一核對原始文獻與對應論斷,如有 AI 參與需如實說明,再講清更正內容與對全文論點的影響。
參考回覆
We checked [reference] against the original publication and confirmed that it does not support [claim] as written. [Accurate explanation of AI involvement, if applicable.] We removed the unsupported statement and revised [affected passage] at [page and lines]. Our examination of the affected argument found [effect on the interpretation or conclusions]. We also checked [specific related claims or references] and made [additional corrections].
提示:僅查核了一篇文獻時,不要聲稱「所有參考文獻均已檢查」;未完整評估錯誤影響時,不要直接下「the conclusions are unchanged(結論不變)」的定論。
情境 3:說明 AI 在程式碼編寫/資料分析中的輔助作用
這類使用請與語言類編輯分開說明,明確交代驗證過程,並在方法章節補充對應資訊。
參考回覆
[Tool] was used to suggest code for [task]. [Author/role] reviewed and modified the code, then checked its output using [validation procedure]. We added the tool’s role and validation details to the Methods section at [page and lines]. [Code or supporting material] is available at [appropriate location]. The checks identified [findings], and we [resulting action, if any].
After checking the manuscript preparation process with all coauthors, we confirm that generative AI was not used to draft, translate, or revise the manuscript text. The initial draft was prepared by [author/role] and revised through [process]. We retained [available drafts or revision records] and can provide relevant material through the journal’s preferred channel.
Could the editor identify the passages underlying this concern? Identifying these passages would help us address any specific problems with the wording, supporting evidence, or interpretation.
請將回覆重點放在稿件內容本身,不要糾結偵測器的準確性。偵測器分數只是機率估計,不能直接證明寫作過程,過早爭論技術細節反而會偏離論文核心。先簡要說明 AI 使用情況,再聚焦具體的寫作問題優化即可,不必認同「文本由 AI 生成」的判定。
參考回覆(已指出具體段落)
Thank you for raising this concern. We confirm that generative AI was not used to [specific tasks performed]. We carefully reviewed the identified passages and revised [specific wording or section] to improve [clarity, precision, or the connection to the study’s findings]. The changes appear at [page and lines]. We hope these revisions clarify our intended meaning.
Thank you for bringing this concern to our attention. We confirm that generative AI was not used to [defined tasks]. We reviewed the manuscript for clarity and precision and [describe any revisions made]. If there are particular passages that remain unclear, we would appreciate your guidance so that we can address them directly.
可以在同一條回覆裡既澄清事實,也落實修改。注意不要把修改理由說成「為了看起來不像 AI 寫的」,而是針對具體的寫作問題做優化。
參考回覆
Generative AI was not used to prepare this passage. However, we agree that the original paragraph was repetitive and did not clearly connect the finding to our research question. We removed the repeated summary and revised the interpretation to distinguish the observed association from a causal explanation.
The revised passage at [page and lines] now reads: “[Insert the exact revised text.]” This change clarifies the scope of the finding and avoids the broader claim in the original version.
We confirm that [statement about AI use]. We did not retain a complete version history for [section or period], so we cannot provide a continuous record of its drafting. We can provide [existing notes, earlier submission files, or other relevant material] and explain how the passage was developed from [research records or analysis]. Please let us know which part of the preparation process requires further clarification.
我們確認 [關於 AI 使用的聲明]。針對 [部分章節/某段時間] 的稿件,我們未保存完整的版本歷史,無法提供連續的起草過程紀錄。目前我們可以提供 [現有材料,如「研究筆記、早期投稿檔案、資料分析原始紀錄」],並可說明該段落是基於 [研究紀錄/分析結果] 逐步形成的。如果您認為還有哪些環節需要進一步說明,請告知。
何時需要主動向期刊編輯尋求指導?
遇到以下情況時,請直接聯繫責任編輯說明情況,不要自行推定規則:
審稿人要求不明確,無法判斷需要補充哪些資訊
你的實際使用方式與期刊公開政策似乎存在衝突
相關錯誤可能影響研究結論,需要說明處理方案
參考表述
During the preparation of our response, we confirmed that [specific use] occurred in [scope]. We understand that the journal’s policy states [policy provision]. We completed [checks], but would appreciate guidance on how the journal wishes us to proceed regarding [specific unresolved issue].