Hello! Welcome to the sixth module of our course, Editing and Polishing Your Manuscript.
In the last lesson, we put the finishing touches on your book's exterior by designing a complete, professional cover using your AI-generated art and Canva. Now, it's time to turn our attention back to the manuscript itself and elevate the quality of the writing.
You've mentioned that you've found the quality of raw AI-generated text to be lacking for your book-writing goals. This lesson directly tackles that problem. We will focus on how to use AI tools to identify and correct common issues in AI-generated text, such as repetition and logical inconsistencies. You will learn to move beyond being a simple prompter and become an editor who uses AI as a powerful diagnostic and revision assistant.
1. Why AI-Generated Text Needs a Human Editor
Before we fix the problems, it's useful to understand why they occur. Large Language Models (LLMs) like ChatGPT are often described as "hyper-advanced autocompletes." They predict the next most likely word based on statistical patterns in the vast amount of text they were trained on. This can lead to some common flaws:
- Repetition: The AI may fall into loops, using the same phrases or sentence structures because they are statistically probable, making the text monotonous.
- Logical Inconsistencies: An AI doesn't "understand" your story. It may forget a detail mentioned in a previous chapter or have a character act on information they don't yet possess. These are plot holes.
- Factual Hallucinations: The model can confidently state incorrect facts. (We will cover fact-checking in-depth in a later lesson).
Just as you might suffer from "author blindness"—where you're so close to your story you miss obvious gaps—an AI has no context beyond the data it's given. This is where you, the author, step in as the lead editor, using AI as a co-pilot with a perfect memory.
How to Use AI for Fixing Plot Holes and Save Your ...
The article 'How to Use AI for Fixing Plot Holes and Save Your ...' from Sudowrite provides an excellent overview of why these errors are hard for authors to spot and how AI can act as an analytical partner.
Please read the sections 'What is a Plot Hole, Really? (And Why They’re So Hard to Spot)' and 'The AI Co-Pilot: How AI for Fixing Plot Holes Actually Works'. Focus on understanding the concept of 'author blindness' and the analytical tasks the AI performs to spot inconsistencies.
2. Tackling Repetition and Generic Phrasing
One of the most common signs of unedited AI text is its repetitive nature. The fix for this involves a multi-step process: generating, rephrasing with AI assistance, and then polishing with your unique voice.
The video below offers a practical workflow for this, emphasizing that you should never just copy and paste raw AI output. It highlights free tools that can help you rewrite and rephrase text to make it more original.
The CORRECT Way to Write a Book with ChatGPT - Do This NOW
In his video, Sean Dollwet demonstrates a workflow for turning raw AI text into more unique content. This directly addresses the issue of repetition.
Watch the segments from 02:03 to 03:51, 03:51 to 04:44, and 05:49 to 06:22. Pay attention to: The three-step process: Generate, Customize, Final Check. How to use a free paraphrasing tool like QuillBot to rewrite text. The strategy of combining tools (like QuillBot and Grammarly's free version) to refine the output.
Your new workflow for defeating repetition:
- Generate the initial draft from your detailed outline (as we covered in previous modules).
- Rewrite & Rephrase: Take the generated text, section by section, and run it through a paraphrasing tool like QuillBot. Its free version is sufficient for this. This will break up the repetitive sentence structures.
- Refine & Polish: Paste the rephrased text into your manuscript document. Now, read through it. The text will be more varied, but may sound a bit awkward. This is where you apply your human touch—correcting phrasing, infusing your authorial voice, and ensuring it flows naturally. We'll cover this human editing process in more detail in the next lesson.
3. Making the AI Your Debugger: Finding Logical Errors
Fixing logical inconsistencies is more complex than fixing repetition. You can't just ask an AI to "rephrase" a plot hole away. Instead, you need to prompt the AI to act as a system for self-correction. In software development, this is akin to writing test cases or running a debugger. Here, we'll make the AI debug its own (and your) narrative logic.
The "Open-Close" Method
A powerful and intuitive way to do this is called the "Open-Close" or "Expansion-Compression" method. Instead of asking for one answer, you make the AI compete with itself.
I TESTED Every AI Novel Workflow for Authors (One CRUSHED It)
The channel Writing Secrets has a video that brilliantly explains this workflow. It's a fundamental shift in how you interact with the AI, moving from hope to strategy.
Watch the first minute (00:00 - 01:06) and the segment from 03:10 to 03:52. Focus on the core idea: generating multiple options ('open') and then using the AI to select the best one based on clear criteria ('close').
This "Open-Close" method is a practical application of a formal prompt engineering technique called Self-Consistency.
For someone with your IT background, the principles behind this will be quite intuitive. It's about building a more robust system by not relying on a single, fragile line of reasoning.
Mastering Self-Consistency Prompting
The article 'Mastering Self-Consistency Prompting' offers a fantastic deep dive into the 'why' and 'how' of this technique. It moves from basic principles to advanced applications.
Please read the following sections: 'Layer 1: The Linear Path – Chain-of-Thought (CoT) Prompting': This is the foundation—making the AI think step-by-step. 'Layer 2: Embracing Diversity – Self-Consistency Prompting': This explains the 'majority vote' method, which is great for problems with one right answer. 'Layer 3: Unlocking Flexibility – Universal Self-Consistency (USC)' and its example in 'Action Card 3': This is the most relevant part for us. It explains how to make the AI judge its own open-ended creative outputs (like story summaries), which is exactly what we need for finding plot holes.
In essence, Universal Self-Consistency (USC) is the engine behind the "Open-Close" method for creative tasks. You generate several versions of a scene or summary, and then you give all those versions back to the AI with a final prompt asking it to evaluate them and pick the best one according to your criteria (e.g., "Which of these is most logical?" or "Which has the strongest emotional impact?").
4. Your Practical Plot Hole Detector
Now, let's turn theory into a practical tool for your manuscript. By combining a summary of your book with a USC-style prompt, you can create a powerful "plot hole detector."
The article from Sudowrite that we looked at earlier provides a step-by-step guide with excellent prompt templates for this exact purpose.
How to Use AI for Fixing Plot Holes and Save Your ...
Let's return to the 'How to Use AI for Fixing Plot Holes' article for a practical guide on implementation.
Read the sections 'Your Step-by-Step Guide to Using AI for Fixing Plot Holes' and 'Case Study: Plugging a Narrative Leak with AI'. Focus on: Step 2: Prepare Your Manuscript: The importance of working with summaries, not the full draft. Step 3: Master the Art of the Prompt: Study the prompt formulas for continuity, character knowledge, and forgotten elements. These are your new debugging tools.
Here are the key prompt structures you can adapt:
-
For Continuity/Timeline:
Analyze the following chapter summaries for timeline inconsistencies. Create a chronological list of events and flag any that seem out of order or logically impossible.[Paste your detailed chapter summaries here] -
For Character Knowledge:
Based on this text from Chapter 5, create a list of everything Character X knows about the stolen artifact. Then, review the attached text from Chapter 8 and identify if their actions are consistent with only having that knowledge.[Paste relevant text from both chapters] -
For Forgotten Elements (The "Chekhov's Gun" Detector):
I am providing my full novel summary. Please identify any significant objects, characters, or subplots that are introduced early in the story but are not resolved or mentioned again in the later parts.[Paste your full book summary here]
Test your understanding!
You've written two scenes. In Scene 1, your main character, a tech CEO, loses her smartphone in a taxi. In Scene 10, she is trapped in a server room and immediately calls for help. You suspect this is a plot hole.
Based on the prompt templates you just learned, write a specific prompt for an AI like ChatGPT or Claude to help you identify and potentially solve this inconsistency.
Show answer
Here is a strong prompt that combines identification and brainstorming:
Role: You are a meticulous continuity editor.
Task: Analyze the provided scene summaries for a logical inconsistency and brainstorm solutions.
Context: I have identified a potential plot hole in my story. Please confirm if it is a logical contradiction and then suggest three ways to fix it.
Summary 1 (Scene 1): My protagonist, Jane (a tech CEO), is in a rush and leaves her smartphone in a taxi. She is distraught because it contains sensitive company data.
Summary 2 (Scene 10): Jane gets accidentally locked in a secure server room. The text says she immediately calls for help.
Analysis & Brainstorming:1. First, confirm if Jane's action in Scene 10 contradicts the event from Scene 1. Explain your reasoning step-by-step.2. Second, brainstorm 3 creative and logical ways that Jane could call for help from the server room *without* her original smartphone. The solution should be consistent with her character as a resourceful tech CEO.
This prompt is effective because it uses the role-task-context framework, asks for step-by-step reasoning (CoT), and directs the AI to both identify the problem and propose solutions (USC).
Conclusion
Today, you've added a crucial new layer to your AI authoring workflow: using AI as an editorial assistant. You've moved from simply generating text to actively diagnosing and correcting its flaws.
Key Takeaways:
- AI-generated text often suffers from repetition and logical inconsistencies due to its probabilistic nature.
- You can combat repetition by using a workflow of Generate -> Rephrase (with tools like QuillBot) -> Polish (with your human touch).
- To fix logical errors, use self-correction techniques like the "Open-Close" method or Universal Self-Consistency (USC), where you make the AI generate multiple options and then evaluate its own work based on your criteria.
- By providing the AI with chapter summaries and specific prompts, you can turn it into a powerful "plot hole detector" to check for continuity errors, character knowledge contradictions, and unresolved plot points.
Preview of the Next Lesson:
While AI is a phenomenal tool for spotting these issues, it cannot replace the nuance and artistry of a human editor. In our next lesson, "Apply human editing to improve the flow, transitions, and narrative coherence of the manuscript," we will focus on the essential human-led process of taking your refined text and polishing it into a seamless, engaging reading experience.