Peer Review Response

Can I use AI to help respond to peer review?

As an author working on your own manuscript and your own reviews, usually yes, subject to your venue's AI disclosure policy, which you should check and comply with. As a reviewer, usually no: a manuscript under review is confidential, and many publishers prohibit entering it into an AI system entirely. Where a publisher's policy is silent, treat it as prohibited, because the confidentiality obligation exists independently of any rule about a specific technology.

How do I write a response to reviewers?

Address every comment, in the order the reviewer made them, with three elements each: the reviewer's comment quoted in full, your response saying what you did or why you did not, and the exact location of the change in the revised manuscript. The location is the element authors omit and editors most need. Open with a short covering note flagging any request you have not complied with, so the editor is not surprised by it later.

Can I disagree with a reviewer?

Yes, and a well-argued disagreement is a normal part of review and is frequently accepted. Acknowledge the point properly, give substantive reasons (evidence, method, literature, scope), name what would be lost by doing it the reviewer's way, and offer something in return. Most disagreements resolve by narrowing the claim rather than defending it whole.

What do I do when a reviewer misunderstood my paper?

Change the text, and do not tell them they misread it. A competent specialist reading carefully enough to write a review took a different meaning from your passage, which means ordinary readers will too. Acknowledge that the point was not clear, state what the paper intends, and name the specific change you made. Never answer with "as stated on page 6".

What if two reviewers want opposite things?

Name the contradiction explicitly in your response, quote both comments, explain the trade, say which you followed and why in terms of the paper's contribution, address the other reviewer's underlying concern by another route if you can, and ask the editor to adjudicate. That is their role. Trying to satisfy both produces an incoherent paper that fails both.

A reviewer is asking for a whole extra study. Do I have to do it?

Test it: could you meet the request with the data, access and time you have, and does the paper's claim actually require it? If the claim does not require it, this is scope expansion. State the paper's scope, add the limitation the point implies, and identify the request as future work. Cost the requests in aggregate, because five individually reasonable additions can amount to a second paper.

Should I appeal a journal rejection?

Rarely. Appeals are appropriate where a review contains a factual error that was material to the decision, where a reviewer had an evident conflict, or where the decision rests on a checkable misreading. They are not appropriate because the decision was harsh. They take months, they rarely succeed, and a failed appeal costs your relationship with the venue. In most cases, revise using the reviews and submit elsewhere.

How long should I wait before responding to reviews?

At least two days between the first reading and any drafting. This is a method step rather than self-care: the first reading is dominated by threat response, attention narrows to the harshest sentences, and responses drafted in that state are defensive where they should concede and conceding where they should defend.

Does a rejection still tell me anything useful?

Yes, and it is the part most often wasted. Run the same sort on the reviews, execute the must-changes and most of the should-changes, and only then choose the next venue. If two independent sets of reviewers at different venues make the same criticism, that is evidence about the paper, not about the reviewers.