Writing used to be simple. You sat down, typed words, and hoped they made sense. Then generative AI waltzed in like Ron Burgundy, pretending to know everything—and often knowing nothing. Some praise it as academic writing’s saviour; others fear it’ll destroy scholarly credibility. Both miss the point. It’s like watching a cat ride a Roomba—it’s definitely going somewhere, but nobody’s quite sure where that is. The real question isn’t whether to use AI, but how to use it without losing your academic soul. Thing is, you have to adapt now as quickly as your AirPods switching between your Apple devices. If you treat AI like a helicopter parent who’s doing all your homework, you’ll feel guilty about it, eventually. There’s a better way to use it.
The problem murmurs not inside the tool
Remember typewriters? Researchers once argued they’d ruin handwriting forever. (Caveat: They did and also didn’t.) But we adapted. AI tools are different, yet somehow the same. They’re not destroying academic writing. They’re changing how we think about authenticity. We don’t want our papers to end up like boy bands from the 90s—different names, same dance moves. Think of AI like a junior research assistant: brilliant but sometimes confused, helpful yet occasionally misleading. You wouldn’t let your RA write your paper alone. Don’t let ChatGPT do it, either.
1. Write first; improve later
Your first draft should flow from your brain, not an algorithm. Start with your raw ideas — messy, incomplete, human. For example, I either jumble unfiltered text together and scramble it into on giant mess, or I talk to Otter AI to transcribe my thoughts perfectly (it is the most accurate transcription software I know), and then I have something to copy over into the LLM of my choice as part of a my mega prompt or my many GPTs, Gems, or Claude projects. If I don’t have much to say, I just outline what I want to discuss before I get started in headings (or use Paperpal's outline function and refine it based on what I want to write about). Then, you can use more AI tools after you’ve mapped your argument. This creates a foundation that’s unmistakably yours.
Another easy example: Draft your methodology section details by hand. Then use AI to check for clarity, not to generate content. In a prompt:
<your text> Make this sound cohesive, clear, concise, and compelling.This maintains your authentic voice while using AI’s strengths. Write with your mind, write with your heart, write with your voice. Your ideas guide the structure. Your expertise shapes the content.
Pro Tip: First, set a timer for 25 minutes. Write or speech-to-text without AI. This creates a pure first zero draft that reflects your genuine thinking. Take it from there.
2. Document every AI interaction
Create an “AI appendix” in your writing process. Record every prompt, every generated response, every edit. This is good practice, but, especially if you’re in undergrad studies, it is also academic survival. Just like citations to previous literature to avoid plagiarism. You need to cite your AI interactions to avoid academic dishonesty. This is your digital paper trail. It shows how you used AI, when you used it, and why. When someone questions your work’s authenticity (and they increasingly will), you’ll have solid evidence. In most courses, where you can use generative AI for your writing, you will have to meticulously document its use. It’s good practice to get into this early. Just your academic insurance policy. And it needs to be more stable than a Jenga tower in an earthquake.
Pro Tip: Keep a simple spreadsheet handy with columns: Date | Tool Used | Purpose | Outcome (see download below). Like a lab notebook, but for your AI interactions and outputs. If you can include URLs to your chat transcripts, even better.
3. Check, double-check, human-check
AI makes mistakes. Big ones. Spectacular ones. It invents citations, mangles statistics, and occasionally writes nonsense with supreme confidence. Don’t trust it. Check every AI-generated sentence. Verify everything. Trust nothing without human oversight. AI is a tool, not a replacement for thinking. Don't put on training wheels when you're heading down a mountain bike trail. It’s your paper’s integrity that's on the line. So, you must read every word, sentence, and paragraph. Use it as your final quality control. You’re the expert. You’re the author. You’re the one responsible for your work.
Each AI-touched section needs three levels of verification:
- Technical accuracy (Are the facts right?). You must check the statistics (which you should still do yourself), the methodology (which you might have enhanced), and the results (which are most important). So, you must verify the AI-generated content against your project knowledge.
- Source validity (Do these citations exist?). This means checking the references and the citations. Really, you should have done this before you used AI. But, now you must do it again. You must ensure the AI-generated content matches the source material. Don’t trust AI’s citations. Ever. It mixes real papers with imaginary ones, real quotes with generated text. Treat every AI-suggested reference as suspect until verified. There are specific tools you can use here (Consensus, Scite AI, SciSpace, Scopus AI, all let you work with the literature and find real references).
- Logical coherence (Does this make sense?). You must read the AI-generated content in context. Are the arguments logical? Do the claims follow from the evidence? Does the writing flow?
Pro Tip: Build a “verification checklist” into your writing process (see download below). Check one citation per paragraph immediately after writing. This catches problems early before they multiply.
4. AI is the scaffolding while you pour in the soul
AI excels at organizing ideas, spotting structural weaknesses, and suggesting transitions. Use it there. Let it help arrange your thoughts—but not think them for you. Feed it your outline and ask: “What’s missing?” “Where might a reader get lost?” “Which sections need more support?”
You can even use it as a sparing partner for your argumentation and discuss the paper's findings with it to help you strengthen your rhetoric. Trade those intellectual jabs like Pokémon cards at recess. Or use AI to outline your paper, then fill in the content yourself. Or use a AI to complete sentences while you write other ones. It's like passing the mic during "Sweet Caroline" at karaoke night; you should be having fun with it. Treat your writing as a conversation. It's not a monologue, Marcus Antonius. But keep it away from your core results and implications for now. Let those come from your expertise, your research, your understanding. AI can help you explain your ideas. It can probably even get a rocket ship to parallel park. And when prompted correctly even find arguments for your results against existing literature. But you will have to judge whether this has legs. It's just like looking at IKEA instructions and deciding if they work for you.
Pro Tip: Write your main arguments on paper first. Then type them up. This physical-to-digital pipeline ensures your core ideas remain purely yours.
5. Keep editors in the loop
Tell your editors which AI tools you’ve used. How you’ve used them. Why you’ve used them. Actually do it. Like a person who uses their gym membership in January. This transparency builds trust and helps establish best practices. Have a declaration section in your paper. If you’re using AI to generate content, say so. If you’re using AI to check grammar, say so. If you’re using AI to improve readability, say so. If you’re using AI to find references, say so.
Share this with editors before they ask. Jump on that responsibility like a Wordle player on a 1/6 streak. Being proactive shows professionalism and establishes new standards as we all figure out this new writing world together. I'm sure soon, it'll just be part of the process. But for now, it prepares you for the inevitable questions about AI’s role in your writing. It’s better to be upfront than defensive.
Pro Tip: Create different disclosure templates for different types of work (research papers, grant proposals, book chapters). Each might have unique AI guidelines attached to it.
6. Set your boundaries in stone
Decide what you won’t use AI for. Make these rules non-negotiable for yourself. Mine? No AI-generated methods, analysis, or results for now. These are where your expertise must come through. These boundaries protect your academic integrity and ensure your expertise shines through. They also keep AI firmly in its place where it can support you the most. But to be honest, I wouldn’t be surprised if emerging tools and practices would even facilitate those aspects of your research in the future.
Pro Tip: Write these boundaries down somewhere. In your Apple Notes, Obsidian, Notion, or whatever tool you use to keep track of things like this. Review them before each writing session to stay focused.
7. Stay human-centric
Your academic voice is your fingerprint. It’s unique. Valuable. Irreplaceable. AI should amplify your voice, not replace it. Think of it like autotune for your brain. A backup dancer who is not stealing the spotlight from the main singer but making the whole show unforgettable. Sure, you can copy your style and clone it. But ask yourself: “Could I explain this paper’s key points without using AI?” If not, you’ve maybe drifted too far from your academic core.
Test each section: Does it sound like you? Does it reflect your thinking? Does it demonstrate your expertise? Don't let your personal assistant steal your identity.
Pro Tip: Record yourself explaining your paper’s key points. Transcribe it. Compare this natural explanation to your AI-enhanced written work. The closer your argument flow matches, the more authentic your voice remains. Remove things that sound wrong.
What to do right now
- Audit your current AI use. List every tool and how you use it. Keep tabs on it like Swifties tracking Taylor's private jet.
- Create your AI appendix template. Start documenting every interaction.
- Write your personal AI boundaries. Make them as specific as your Starbucks orders and stick to them.
- Draft your AI disclosure statement for your next submission. Create templates from it.
- Set up a verification system for AI-suggested content. Use a checklist.
Academic writing isn't like ghosting AI; it's like sending a connection request on AI's LinkedIn profile and sliding into those DMs professionally. These seven rules put AI’s power to work without sacrificing your academic integrity. Start with one rule today. Master it. Then move to the next. Repeat.
Downloads
Use my example of this AI interaction tracking spreadsheet below. Also, download my ethical checklist (duplicate to Notion and as PDF) for AI-generated content that will help you assess your article. Here are your downloads:
