Write Your Thesis with AI: A Guide for Students
If you reading this blogpost at 2025, you probably know this is a confession time in academia. To be spesific, writing a thesis always felt like staring down a mountain of research that never ended. However, now, working with my students this past semester, I got to experiment with the newest AI tools designed to, well, help us conquer that mountain. Well, this is not as dramatic as I probably tries to make it sound but you get the idea. Therefore, to help my (and others) students in the future, my latest and greatest suggestions for thesis with AI writting. To this end, this post is not a sterile... I.e., it is not a "here's what the research says" type of blogpost. Instead, it is a real, raw look at how AI changed the way I think about structuring a thesis, organizing my thoughts, and even fighting that nasty writer's block. Thus, if you find it somewhat messy, this is a feature and not a bug.
The task of writing a thesis, a culminating project in academic life, often presents itself as a daunting and seemingly endless endeavor (learned this word from 'Suits'). The sheer volume of research, the meticulous referencing requirements, and the stringent standards of academic writing can transform the experience into a relentless uphill climb (I use the metaphor to its fullest). Nevertheless, new Artificial Intelligence (AI) developments offer a set of tools and methodologies that, when thoughtfully integrated, can significantly enhance the efficiency and quality of thesis writing.
Phase 1: Foundation – Data Acquisition and Organization
Before we even think about letting an AI loose on our thesis, let us be clear: there is no shortcut around the core of legitimate academic research. The integrity and value of your thesis hinge entirely on the quality and provenance of the data you collect. Forget the promise of instant answers and pre-fabricated analyses – at least for now and the next several years (2028 readers, send me an email if I was right). We are still talking about original research, and that means getting your hands dirty. As such, this first phase is not about AI, rather it is about setting up your future interactions with the tool. It is about being deliberate and setting up the narrative ahead of time. Here is the procedure:
1. Thematic Chapter Design: Start with what you want to say. Before diving into individual figures, outline the overall scope of your thesis into various subsections. This chapter based system allows you to compartmentalize your efforts and not get overwhelmed by various aspects of the thesis. Each chapter should focus on an overall theme of what you want to say.
2. Figure Extraction: While the overall theme of the thesis is in place, extract individual figures (graphs, images, or tables) from your data set that represent a specific idea or potential findings.
3. Density Requirements The old saying "less is more" does not apply to a thesis! Ensure that there is a minimum amount of figures (usually around 12 -15 per chapter) to extract a thesis. There should be a reason for this much figures, as AI will be doing a lot of the work in helping build the connections.
4. The Why Behind the Figure Now we need to talk about your data, that is, what do these figures represent and tell the AI tools that. For example, Figure X represents the morphology of the new electrode composite that demonstrates its impact on increasing the electric conductivity of such devices. The details and description should be very explicit so that the AI vision model has a clear indication of what is in place to help generate the story.
Phase 2: AI-Powered Narrative Construction - Seeing the Story in Your Data
Alright, you've got your figures organized, you know what they represent... Now it is time to bring in the AI muscle. This is where we start transitioning from meticulous groundwork to leveraging AI for creative narrative construction. The key here is not to blindly accept what the AI spits out, but to use it as a thought partner, a catalyst for new ideas and perspectives.
AI can power your narrative, where the key goal is not to blindly accept AI but to use it to help you think. Consider uploading the figures for each chapter into the AI and asking it to produce a potential table of contents. Crisp, clear images are essential for accurate AI vision interpretation. Image qualities and descriptions need to be very deliberate. In addition, do not just upload the raw image files. Be sure to add a description before uploading the image, or even simply give it the description as the file name.
To ensure the AI takes every detail into consideration, use a specific prompt. A vague or poorly worded prompt will yield equally vague and unhelpful results - which you do not want (just making sure we agree on this point). As a template for your prompts, you can use an example like, "I've uploaded [Number] figures that I am grouping to form a chapter in my thesis on [General Thesis Topic]. The figures represent [A brief, high-level summary of the chapter's overall theme/goal]. Please consider the following: 1. Each figure's content: [Quick descriptions of each figure, expanding beyond the figure title]; 2. The figure that is most important; and 3. The most controversial / unique / unexpected aspect of the collection of figures." Now, the AI should have all it needs to help give you an outline.
After all your prompts are set, evaluate your output. Do not blindly accept the AI's suggestions. Evaluate the logic, coherence, and relevance of the generated outline. Look for insights that resonate with your existing understanding of the data, and thoughtfully integrate them.
Phase 3: Voice-Driven Content Generation
With a structured outline based on both AI insight and your expert knowledge, now its time to fill in the gaps. In this section, you will use your voice to speak, not type! The common thesis writing method is to be tied to your chair, working endless hours writing text for figures. A new method to combat the endless hours, while also making the work easier is by talking. When your thoughts about your research is fresh, it can come out easy without you having to think about it. That is why voice-driven AI can come in handy to make the process a little easier. You can load it into your AI, and then start speaking into your phone. Some key tips include: talking in layman's terms, being conversational, and trying to use proper punctuation and grammar. While it may take some getting used to, remember, AI is just a tool to enhance what you're doing. Not to replace it!
Phase 4: Refinement, Editing, and Ensuring Academic Integrity
Now that the "raw" material is created via AI, it is vital to ensure it aligns with the stringent requirements of academic rigor. This involves a meticulous process of refinement, editing, and verification. It is at this point that you assess your AI. You need to make sure that it has a good understanding of your work, has good formatting, and ensures that the work is not plagiarized. To ensure you have high quality standards, it is recommended to copy the written generated by AI tools into a tool like Paperpal. Think of it as another peer reviewer, providing feedback in all important areas. These changes can include: Language and Style Adjustments, Consistency Checks, and verification of accurate facts in the content. All these changes are done so that you are not just generating content, but generating academic content, that you will be proud to put your name on.
Phase 5: Introduction of a Literature Review
With all that refined, and ensuring that the generated content adheres to academic writing standards, now you can proceed to the creation of a formal Literature Review. While it used to take hours upon hours, AI tools have made it simpler than ever. Previously students had to conduct a manual literature review. This involved meticulously combing through journals, books, and databases to identify relevant sources, summarize their findings, and synthesize their contributions to the field. The process was time-consuming, labor-intensive, and prone to human error. The AI allows you to simply load the resources, give it an outline, and it will give you the key points of the thesis. While AI cannot fully write the work, the AI can extract information from documents in a way that used to be done manually. The two main tools the video highlighted for this are Notetaker LM and Thesis AI. Notetaker LM is particularly helpful to build a foundation structure and Thesis AI can help in the research of the various aspects.
Phase 6: Combining Conclusions and Abstracts
With a structured outline for both your introduction and the body sections, you have to make sure the conclusion and the abstract are all up to snuff! Previously, one would have to comb through your literature review and take hours to think of how best to word these points to make the work easy for people to understand. However, we can now use the tools previously described to do so.
Simply put, copy all of your section conclusions into one ChatGPT or Gemini model and tell it to generate your conclusion section for the entire thesis. The AI will give a synthesized thesis, so make sure to edit for language, style, and consistency like previous steps.
Now that this has been done, go back and summarize the conclusions, and with all that information, you have the foundation you need to write your abstract. The abstract, it seems, is more of a summary and can be the fastest way to make the entire thesis.
Phase 7: AI-based Assessment and Refinement with Thesify
If you think you are finally done writing your thesis, hold on! Before submitting to your supervisor (my students, please read next sentance several times in repeat), it is important to put the thesis in the AI ringer to ensure all your work passes with flying colors.
For academic writing, there is no tool better for this than Thesify. Think of it as a very high quality supervisor who wants your paper to be the best. Thesify will evaluate the text against stringent academic writing criteria and then provide an iteractive feedback loop for you to engage with. Thesify can offer support for checking if your work is up to par for the standards. What else can you ask for?
In essence, using AI tools to craft your thesis involves a strategic blend of meticulous manual groundwork and intelligent automation. It is not about letting the AI write the thesis for you, but rather about using it to boost your own research, enhance your writing, ensure academic rigor, and identify potential areas for improvement. With thoughtful planning and an understanding of each tool's strengths and limitations, AI can transform the arduous process of thesis writing into a more efficient, insightful, and ultimately, successful endeavor.
Personal note: by the way, I did not used LLMs for this blogpost, just like it is 2021 out there. Probably not my best decision, but felt right for this one.