How to Review Your Thesis Before Submission With AI
Login
Sign upSign up for free
21 Sep 2026··4 min read
How to Review Your Thesis Before Submission With AI: A PhD Candidate’s Workflow
How a King’s College PhD candidate used AI to check his literature review, methodology and theory against his own sources before submitting his thesis.
Arthur Wittenburg is a PhD candidate and lecturer who holds the Associateship of King’s College. As he approached the final stage of his PhD, he used SciSpace to review parts of a thesis he had already written.
He began with a subject he knew well.
In short: to review a thesis before submission with AI, first test the tool on a literature you already know well. Then ask for separate feedback on the literature, the methodology and the thesis as a whole, and restrict theory checks to your own reference library (here, Zotero) so the AI checks your interpretation instead of adding new sources. Keep the introduction, interpretation and core analysis to yourself. You can run the same review with the SciSpace thesis writer agent.
SciSpace divided the feedback into three perspectives: one focused on methodology, another on the literature, and a third offering a broader assessment of the thesis.
The value was not that the assessment could decide whether his thesis was ready. It gave him separate questions to consider about different parts of the work before formal review.
Arthur described why that mattered at the final stage of a PhD:
This is really helpful if you want feedback just to reduce the insecurity that is normal at this stage.
He found the result useful enough to share the assessment PDF with his PhD cohort of approximately 30–35 people.
He introduced it by telling them:
If you want to really understand something about your thesis, look what this did.
That was the clearest outcome of the experiment: the feedback was useful enough for Arthur to place it in front of other doctoral researchers and let them examine it for themselves.
Reviewing the theory using only his Zotero sources
Arthur’s most controlled use case involved the theory section of his thesis.
He connected his Zotero library to SciSpace and gave it a specific instruction:
Just take into account my Zotero library. Do not bring other authors into the analysis.
He was not asking SciSpace to expand the bibliography or introduce a different theoretical direction.
He wanted it to check whether his interpretation of the literature and his use of quotations were consistent with the sources he had already chosen.
Restricting the review to his Zotero library changed the nature of the task. Instead of asking, “What else should I cite?”, Arthur was asking:
Have I understood and represented these authors correctly?
Deciding where AI belonged in the thesis process
Arthur drew a clear line between reviewing existing work and developing the central research argument.
He saw the strongest role for AI in assessing the literature and methodology. He did not want it to take over the introduction or core analysis.
He also noticed that AI systems often respond to feedback requests by suggesting additions or making an argument more complicated. But that was not always what he needed.
Sometimes you just want to see if your thinking is coherent and presented in a simple and effective way.
That distinction shaped his use of SciSpace:
Literature Review helped him examine coverage and structure.
Thesis Assessment gave him separate perspectives on literature, methodology, and the thesis overall.
Zotero integration allowed him to review the theory against a defined collection of sources.
The introduction, interpretation, and core analysis remained his responsibility.
Arthur did not ask AI to produce the thesis. He used it to pose more specific questions about work that already existed.
For him, the usefulness depended on defining those questions clearly and knowing which parts of the thesis should remain entirely with the researcher.