The short version. The thesis is the by-product of a PhD. The transformation in how you think is the product. Outsourcing the writing to AI skips the transformation and leaves you defending work you did not do.
A doctorate is the only qualification where the point is what happens inside your head. The thesis is the artefact. The mind that produced it is the actual thing. This piece is about how to write your PhD authentically, and why staying in your own thinking is the skill the degree is meant to build in you.
I have been thinking about this all week, because the advice doing the rounds at the moment is that you should use AI to write your PhD. Everyone does. The thesis is just a hoop to jump through. The faster you get through it, the better.
I want to push back. Most of this advice gets passed around in posts written by someone who has never sat in a viva.
What writing your PhD authentically really means
Writing your PhD authentically means the words on the page reflect thinking you have done yourself. The arguments are arguments you can defend because you built them. The literature review is a synthesis your brain made. When your examiner asks you, in your viva, why you chose this framework over another, you have a real answer because you wrestled with it.
It does not mean refusing to use technology, doing things the hard way for its own sake, or proving anything to your supervisor or your examiners or yourself.
That is what authenticity means here. You did the thinking.
The transformation problem
Here’s the thing. Most PhD advice gets one structural fact wrong. It treats the PhD as a project. A long, difficult, complicated project, but a project nonetheless. You set goals, you hit milestones, you produce a deliverable. The thesis is the output.
If that is what you believe, then of course you should use whatever tools make it faster. AI included. Why wouldn’t you?
But the PhD is not a project. The thesis is the by-product.
Five years of reading, writing, getting things wrong, defending your ideas to people who disagree with you, watching your argument fall apart and rebuilding it. That process changes your brain. You leave the PhD able to do something you could not do at the start. You can hold a complex problem in your head, examine it from multiple angles, and produce a defensible position about it. That capacity is the doctorate. The bound thesis is just the certificate the process leaves behind.
When you outsource the writing to a chatbot, you skip the transformation. You walk out with a document that says “Dr” on it and a brain that did not have to do the work the title is meant to mark.
You are paying for a transformation and walking out without it.
What AI is good for, and what it is not
I am not anti-AI. That would be a bad position to hold in 2026. AI is useful for some parts of the doctoral process, and pretending otherwise makes you sound like the people who said the internet was a fad.
The line is about whether the thinking is yours. If the chatbot did the synthesis, you cannot defend the synthesis. If the chatbot wrote the argument, you cannot evolve the argument. You become the spokesperson for a position you do not hold.
Here is where the line sits in practice.
| Use AI for | Do not use AI for |
|---|---|
| Reformatting references | Writing your literature review |
| Catching typos | Drafting your methodology chapter |
| Translating a passage from a language you do not read fluently | Producing your analysis |
| Summarising an article you are deciding whether to read | Generating the argument of your thesis |
| Rephrasing a sentence you have written but cannot quite get right | Naming your contribution |
| Sparring on an argument you have built yourself | Writing the words your viva examiners will ask you to defend |
There is a further reason to hold this line. Recent research from Wharton (Shaw and Nave, 2026, n=1,372) found that users accept 80% of factually wrong AI answers and report an 11.7% boost in confidence while being wrong. That is the exact failure mode a doctorate is supposed to train out of you. Authority built on fluency you did not earn is the thing the viva is designed to expose.
Practical scaffolds for staying in your own thinking
I’ve noticed that knowing the principle and applying it under pressure are very different things. When every part of the system is whispering “just use AI, nobody will know”, the gap between principle and practice gets very loud. Here is what helps.
Write before you research
If you sit down and write what you already think about a topic before reading anyone else, you start from your own ideas. Then you read, and the literature responds to your thinking instead of replacing it. This sounds backwards. It works.
Take handwritten notes when you read
Typing notes is too easy. You end up transcribing. Handwriting forces you to compress, to choose what matters. The slower pace makes the ideas your own.
Talk about your research out loud, often
To anyone who will listen. The act of explaining your work to a non-specialist forces you to understand it. If you cannot explain it in your own words to a friend over coffee, you do not understand it yet, and no AI can fix that for you.
Read your writing out loud before editing
If a sentence sounds like nobody really says it, it probably came from your defensive academic voice. Your real voice sounds different.
Log every AI use, visibly
When you do use it, write down what you used it for. Keeping a log forces you to be honest with yourself. If you cannot justify a use without feeling uncomfortable, that is the signal.
The community that helps you write your own work
The single biggest thing that makes authentic writing possible is community. Community gives you somewhere to bring your half-formed ideas before they harden into bad arguments. Policing has nothing to do with it.
When you write alone, you have no resistance. Your bad ideas grow unchallenged. You start writing things you do not really believe because nobody is around to ask why. AI accelerates this because it agrees with everything you give it.
A real community asks questions you would not have asked yourself. Other people in the room notice when your argument has a hole. They tell you the bit you skipped over too quickly. They make you defend yourself, which is the only way to find out whether you have a position worth defending.
This is what the PhD Common Room is for. It is a community of doctoral researchers who meet to write together, talk through what is hard, and ask each other the questions that make the work better. Members pay between £30 and £75 a month. The community is what makes the writing possible. The writing is what makes the doctorate real.
Your PhD is the slow, deeply unfashionable process of becoming someone who can think rigorously about complex problems. You become that person by doing the work. The work is wrestling with ideas until you understand them well enough to defend them.
That cannot be automated. The interesting question is no longer whether to use AI for the thinking. The interesting question is this. How do you build the conditions that let your mind show up?
The answer is the same one it has always been. Other people, doing the same work, who will sit with you while you do yours.
Common questions
Can I use AI for my PhD literature review?
Not for the synthesis. Use it to find sources, summarise articles you are deciding whether to read, or catch typos. The moment you ask it to decide which ideas matter or how they connect, the thinking stops being yours, and that is the part your viva will test.
Will my examiners know if I used AI to write my thesis?
They will know in the viva. Borrowed thinking collapses the moment someone asks you why you chose this framework over another. Written fluency without held understanding is the tell.
Is there a way to use AI ethically during a PhD?
Yes. Use it as a sparring partner, not a ghostwriter. Log every use. If you cannot explain what you used it for without feeling uncomfortable, that is the signal to stop and do that bit yourself.
0 Comments