Where a thesis gets stuck
Most theses do not get stuck for lack of writing. They get stuck for lack of a plan.
The familiar story goes like this. The topic is fixed, a few weeks go on reading papers, then you sit down to write the introduction and three days pass without a single paragraph. A month before the deadline the results chapter has not started, and the data was not collected the way the methodology says.
There is one cure: first the structure of the whole thesis, then a week by week plan, then one chapter at a time.
Structure first, then writing
Most theses stand on six chapters. If your university or department has its own format, follow that first; the one below is the common pattern.
| Chapter | What it holds | The question it answers |
| 1. Introduction | Background, problem, objectives, research question, significance | What you are looking for and why |
| 2. Literature review | Earlier work, organised by theme, ending in the research gap | What is known about this, and what is not |
| 3. Methodology | Design, sample, data collection, analysis, ethical approval | How you looked |
| 4. Results | Tables, charts, test results | What you found |
| 5. Discussion | What the results mean, comparison with earlier work, limitations | What it means |
| 6. Conclusion and recommendations | The main point, recommendations, future work | What comes next |
Outside these sit the Abstract, the references and the appendix. Write the Abstract last, because until the whole work is done you cannot know what goes in it.
The backbone of the whole structure is one sentence: your research question. Every chapter looks back to it. If a part does not connect to that question, however well written, think about cutting it.
A week by week plan
Count back from the deadline and build the plan. Below is a twelve week example; change it to suit your department's timing and how you collect data.
| Week | Work |
| 1 to 2 | Topic, research question and proposal, the supervisor's approval |
| 3 to 4 | Reading papers and the first draft of the literature review |
| 5 | Writing the methodology, fixing the questionnaire or instrument |
| 6 to 8 | Collecting and analysing data |
| 9 | The results chapter |
| 10 | Discussion |
| 11 | The introduction again, the conclusion, the Abstract |
| 12 | Checking references, formatting, proofreading, submitting |
The introduction comes back in week eleven, and that is not a mistake. Write a rough introduction in the first week, and rewrite it at the end once you know the results. Then the introduction and the conclusion say the same thing.
What supervisors look for in each chapter
Before you hand in a chapter, check these five yourself. Most of a supervisor's comments usually come from them.
- Do the objectives and the research question match? Whether the results answer exactly what the introduction said you would look for.
- Does every claim have a source? Beyond common knowledge, a source beside every claim.
- Is there one thread from methodology to results? Whether the results came from the method you wrote down.
- Are the limitations written honestly? Hide a small sample or data from one place, and it becomes the first viva question.
- Is one citation style used throughout? Starting in APA and drifting into another style halfway is a very common mistake.
Which jobs AI can take, and which are yours
In a thesis, AI can be a patient assistant, but not the author. Split by job, it looks like this.
| Job | Where AI can help | What must be your own |
| Topic and question | A few options, sharpening the question | Which question you choose, and why |
| Literature review | Organising the papers you give it by theme | Finding the papers, reading them, checking every source |
| Methodology | The structure, what needs writing | What you actually did, exactly |
| Results | Laying out tables, the language of explanation | The data and every number |
| Language | Fixing it into academic language | The argument and the claims |
| References | Arranging them in the style | Whether the source really exists |
| Viva | Making likely questions | The answers, in your own words |
The right column is your thesis. The left column only saves time. The bigger picture of where AI helps in studies and where it becomes cheating is in How to use AI in your studies.
Fake references: the biggest danger
AI can make up an author's name, a paper's title, a year and a journal so that they sound completely real. Why that happens is in When AI gets it wrong.
If a fake reference is caught in a thesis, it is not treated as carelessness. So check every reference three ways before you submit:
- Search the title on Google Scholar. Whether exactly that paper comes up, with the same authors and year.
- If there is a DOI, open it. A DOI should take you to that very paper.
- Check the paper says what you cited it for. A real paper that never said that thing gets caught too.
How the research cards in the chat work
On the empty page of the BanglaCodes chat there are research cards in two groups: "Research paper and thesis writing" and "Research: reading, comparing, checking". The writing box's "More" menu also has "Research paper and thesis". The full list of what each card does is in the research guide.
The thesis card gives the plan for the whole thesis in its first answer: the six chapter structure, roughly how many words each chapter takes, a week by week plan from your deadline, and a checklist for every chapter. After that it writes one chapter, or a part of one, per answer, because a whole thesis does not fit in one answer. It asks only for what it does not know, such as your department, the word limit or the citation style.
Three rules are built into the cards themselves:
- They never invent a source. Where unsure of an author, a year, a DOI or a statistic, they write [verify] in its place.
- They write no results without your data. If you give no data, the space stays empty, and no made-up number goes in.
- They write a literature review only from the papers you give. If a theme is short of papers, they say so, and tell you what kind of paper to look for.
You can give the papers as files. How much of a message is read, and what happens with a scanned PDF, is in Notes from a PDF book.
Academic integrity, plainly
The rules on using AI are not the same everywhere. Some places allow fixing the language, some allow no AI help at all. Do not guess, ask your supervisor.
What holds everywhere:
- Rewrite the draft in your own words. An AI draft is a structure, not the text you hand in.
- Declare AI help, the way your institution asks.
- Swapping words to slip past plagiarism software is not paraphrasing. Honest paraphrasing means explaining it again in your own words, and keeping the citation of the original source.
- Keep your notes and drafts. They are the simplest proof that the work is yours.
And a last point: in the viva, the examiner asks about your writing. On a chapter you did not understand yourself, it shows within ten seconds.
Questions and answers
Which chapter of the thesis should I write first? The methodology and the literature review first, because they can be written before the data is collected. Keep a rough draft of the introduction at the start, and rewrite it at the end once you know the results. The Abstract comes last.
Is using AI in a thesis cheating? Handing in parts written by AI as your own is cheating. Getting help with the outline, the language, the reference style or viva preparation is allowed in many institutions, but the rules differ from place to place. Ask your supervisor, and declare AI help as the rules require.
Can AI write the whole thesis? It can produce text, but that will not be your thesis. The research question, the data, the results and the argument have to come from you. The BanglaCodes cards do not write the results without your data; they leave the space empty.
Can I trust the references AI gives? No, not without checking them yourself. AI can make up fake references that sound real. Search the title on Google Scholar, open the DOI, and check that the paper says what you cited it for.
How should I organise the literature review? By theme, not one paper summary after another. In each theme, which studies agree, where they disagree, whose method is how strong, and at the end the research gap, which leads to your question.
What do I do if the supervisor keeps sending it back? Put every comment in a list, and with the next draft write one line on where you fixed each. If the same comment comes a second time, the first fix did not work.
How do I prepare for the viva? Make a list of likely questions from your own thesis: the objectives, the gap, why this method, what the results mean, and the limitations. The chat's viva card makes questions and draft answers from your own writing, but practise saying the answers in your own words.
Can a thesis be written in Bangla? If your department allows it, yes. The chat's cards write in both Bangla and English. Technical terms and references usually stay in their original language, so check your department's rules.
In short
- First the whole structure and a weekly plan, then one chapter at a time
- The research question in one sentence, and every chapter looks back to it
- Rewrite the introduction at the end, and write the Abstract last
- AI helps with the outline, the language, the style and the viva; the question, the data, the results and the argument are yours
- Check every reference yourself
- Learn your institution's rules and declare AI help