4 Prompting Techniques That Make AI Writing Sound Human

Summary

This video reveals four advanced AI writing techniques to transform your drafts into polished, human-level books. It covers using focused checks and critiques to clean up AI-generated text, context engineering to provide precise, relevant information for each scene, and narrative physics with sliders to make characters and plot elements dynamic and realistic. Finally, it explores layering multiple LLMs to leverage each model's strengths for dialogue, prose, and dramatic scenes. By combining these techniques, authors can efficiently create coherent, engaging, and nuanced AI-written stories that feel authentically human.

Takeaways

• 😀 Focus on using checks and critiques to improve AI-generated writing by creating specific improvement plans, rather than rewriting entire drafts.
• 😀 Context engineering is crucial for giving AI the right information for a given scene, ensuring it doesn't get overwhelmed by unnecessary context.
• 😀 Narrowing down context to only the necessary elements helps the AI generate better, more relevant results without getting bogged down in too much detail.
• 😀 Narrative physics, or using 'sliders' for different character traits and emotions, allows characters to react more realistically based on their current state.
• 😀 Use sliders to depict emotions, character stress, or magical systems, allowing for more dynamic and believable character development.
• 😀 Layering involves using multiple LLMs for different aspects of writing, such as scene structuring, dialogue creation, and prose generation, to enhance the final output.
• 😀 Layering with different LLMs helps avoid the monotony that can come from relying on a single AI model, resulting in varied and more engaging writing.
• 😀 By using context engineering tools like Novelcfter, you can effectively curate only the necessary character details and plot points for specific scenes.
• 😀 Summarizing chapters or entire books into concise documents makes it easier for AI to process relevant information, improving its performance in writing new content.
• 😀 Running multiple checks (for adverbs, dialogue tags, AISMs, etc.) helps clean up AI-generated drafts and minimizes the need for manual editing, streamlining the writing process.

Q & A

What is the primary goal of using the four prompting techniques discussed in the script?

The primary goal is to improve the quality of AI-generated writing, making it indistinguishable from human writing, by addressing common issues like AI-specific phrases and improving the overall flow and consistency of the text.

Why is the technique of utilizing checks and critiques important in the AI writing process?

Utilizing checks and critiques helps identify and correct issues like overuse of adverbs, repetitive phrasing, and other AI-specific quirks, allowing the writer to improve the draft without rewriting everything manually.

How does the 'Improvement Plan' work in the checks and critiques technique?

The 'Improvement Plan' involves asking the AI to create a plan for how to address specific issues in the text (e.g., removing excessive adverbs), without rewriting the whole draft. This focused approach allows for a more efficient revision process.

What is context engineering, and how does it help improve the AI's writing?

Context engineering is the practice of curating and refining the information provided to the AI. It ensures that only relevant and succinct details are passed to the AI for writing, which improves the quality of the output and prevents it from becoming generic or overly broad.

Why should you avoid giving the AI too much context in your prompts?

Providing too much context can overwhelm the AI, making it difficult for it to focus on the key details. This can dilute the quality of the output, as the AI will have to sift through unnecessary information to identify what's important.

What is narrative physics, and how does it apply to character development in AI writing?

Narrative physics involves using 'sliders' or numerical values to track and adjust different aspects of a character's emotional state or motivations. This technique helps create more dynamic, realistic characters that evolve over time based on their experiences in the story.

Can you give an example of how character 'sliders' might work in practice?

For example, a character’s stress level could be tracked on a scale from -10 (high stress) to +10 (peaceful). As the story progresses, their stress slider might shift based on events, making their reactions and decisions more authentic and varied.

How does the layering technique improve the AI-generated writing process?

Layering involves using multiple AI models for different aspects of writing, such as structuring scenes, generating dialogue, and creating emotional depth. This technique allows each model to focus on what it does best, resulting in a more varied and polished final product.

What are some examples of how different AI models can be used in the layering technique?

For instance, Gemini or ChatGPT might be used to create a scene brief, Grock could be used for generating dialogue, and Claude Opus could handle dramatic or climactic moments. By combining the strengths of various models, the final output becomes more diverse and engaging.

How does context engineering impact the efficiency of writing with AI?

Context engineering streamlines the writing process by ensuring that the AI receives only the necessary information. This reduces unnecessary complexity and improves the quality of the output, allowing the writer to focus on the creative aspects of the story rather than managing AI-related issues.