How to Write a Great Essay With ChatGPT Without Cheating

Like any tool, AI can be a big help to a writer. It can come up with an effective outline, suggest sources or a research topic, and help edit. However, it can’t write the essay for you.

Not only would that be a violation of many honor codes and ethical guidelines, but it would also result in a terrible essay! Essays written by ChatGPT are often riddled with factual inaccuracies, comical misunderstandings, and a thesis with enough twists and turns to resemble a preschooler’s self-portrait.

When In Doubt, Seek Help Out

If a strict Honor Code is in play, it might be wise to shoot a quick email to the professor or teacher in charge. It never hurts to confirm that AI tools are acceptable as long as all the writing is human. Schools, colleges, and universities are all grappling with how to deal with the existence of AI writing tools, and there’s probably some form of official policy you can access, or at the very least individual professors and teachers will have their own policies in place.

Find Your Favorite AI

ChatGPT is a great place to start. It is a good all-purpose AI, and there are also custom GPTs available that might suit your needs as well. Or, you could use different AI for each stage of the essay writing process. I think Perplexity AI is the best for research. It will provide sources and you can ask follow-up questions. You also have the option of running a local AI, if your computer has the specs for it.

Fact Check

No matter where you start, always fact-check your AI. Available AI technology is getting better every day, but it still makes mistakes. Frequently. Some mistakes are obvious, and others are not. Using AI to gather research is a quick and efficient way to get started, but you still may have to crack open a book or two. AI works best as a tool to support the human element. Wherever possible, ask the LLM to provide sources for its claims, and the follow up on those sources to ensure they exist and actually say what the LLM states.

Write an Effective Prompt

A good prompt is detailed and specific. An extreme example would be:

Provide some good primary sources from US history," 

Versus:

"List the natural progression of American Colonial ideals using examples from the documents of the 1st Continental Congress, Thomas Paine's Common Sense, and the Declaration of Independence.

AI is a tool that gives only as good as it gets. The first prompt will result in a mishmash of unrelated sources. The second will produce an excellent starting point for an essay because the human has done the complex analytical thinking. The AI can then do what it does best.

Have Your LLM Suggest an Outline

Entering the above prompt into Perplexity AI gave a pretty good answer. After a quick fact check, the next step is to have your LLM suggest an outline. To do this, you’ll need to come up with a thesis on your own. If you’re struggling, ask the AI to suggest a few based on those sources. In my opinion, thesis writing is not a strength of AI, but its suggestions might point you in the right direction.

To get an outline from Perplexity, I prompted:

"Using those 3 sources, compose an outline for a 600-word essay supporting the assertion that the First Continental Congress defined what actions needed to be taken, Paine's work was a call to action, and the Declaration of Independence was the action."

Another option would be to copy/paste the relevant portion from Perplexity AI into ChatGPT and enter the same prompt, this time referencing the “above information.”

Write the Essay Yourself!

Armed with the sources, topic, thesis, and outline, be sure to write the actual content yourself! This is the most important step to avoid cheating. You’re using the groundwork done by the LLM and verified by you as a foundation for the essay. Now it’s up to you to write a coherent piece in your own voice, making arguments you’ve come up with yourself. Remember, you aren’t just being graded on your ability to write, but also on your ability to think, and that’s what you must demonstrate in your essay.

Be Careful with AI Editors

Use AI to help make grammar and spelling corrections, or ask it to make content suggestions, but don’t allow it to re-write your content. AI tends to write every sentence in active voice which is usually considered superior to passive voice. However, most humans don’t write exclusively in active voice. Therefore, AI detectors take this into account when making their determinations.

This is just one example of how seemingly subtle changes suggested by AI can increase the risk of your work getting identified as AI-written. Furthermore, it is still up for debate how much AI can change a sentence before it becomes AI-written. Grammarly AI underlines grammar or spelling errors in red, and other issues in other colors. So, in Grammarly, I only fix those red errors. That said, asking ChatGPT or an AI editor for a few suggestions on how to improve your essay might be a safe way to check your work.

A Word on AI Detectors

The best way to make sure your work will pass an AI detector is to pre-emptively put your work into an AI detector to see if it passes. Generally speaking, AI detectors do not work. Whether or not a writer uses AI, theres a chance they will judged by an AI detector, I’d suggest they make sure their work will pass as human. Human-written work can easily get labeled as AI-written, and writers suffer the consequences. If an AI detector claims your original work is AI-written, let it identify which passages are “AI written,” and re-write them until it is satisfied.

Finally, let me say it one more time: be sure to write all the content yourself, and get clarity on AI policies before using AI tools.