Key takeaways
- Ask the LLM for targeted suggestions before giving it a vague rewriting task.
- Use IDs and a precise output format to choose which changes the model should apply.
- Always verify the edits, added facts, and sources before publishing.
Whether you are creating a course lesson, a blog article, or an important report, polishing your drafts into professional, high-quality content is an essential step in creating impactful experiences. In this article, we will explore how large language models like ChatGPT and Claude can help you refine your drafts, using a course lesson as our example.
Once you have a first draft of your lesson ready, meaning after you have done your research as we saw in the previous article, it is time to polish it. AI can be a fantastic ally here by helping you with several useful tasks.
Step 1: Ask AI for improvement suggestions
An effective method is to ask AI for suggestions to improve your lesson, then apply the changes that seem relevant to you.
Start by writing a prompt that asks the LLM to provide useful feedback on a lesson you prepared, and add that lesson to the prompt.
The prompt can begin by specifying that the LLM should act as an AI expert, then include a copy of the lesson to improve. Next, add a list of specific details to improve and a clear format for the suggestions.
Here is what that prompt might look like:
You are a knowledgeable LLM expert and a technical writer specializing in AI.
Here is a lesson I am working on:
[insert the lesson here]
Give me suggestions to improve the lesson above. Focus on the following points:
- Suggest improvements to the lesson’s flow. Recommend topics to expand or places where examples would make the ideas clearer.
- Suggest improvements to the grammar and syntax. Identify sentences that could be rephrased for readability.
- Flag potentially inaccurate information. Explain why it is inaccurate and propose a corrected version.
- Identify redundant information.
- Give the 10 most important suggestions you can think of.
Provide the suggestions in the following format:
Suggestion ID: [from 1 to …]
Text to change: [exact text taken from the lesson]
Replacement text: [exact corrected text to insert into the lesson]
Reason for the change: [reason for the change]
Why use this precise format for the suggestions?
First, it makes the reasoning behind every suggestion easy to understand, helping you, as the expert author, choose the relevant ones. Second, it clearly identifies the changes to make. Finally, assigning an ID to each suggestion lets you ask the AI to apply only specific suggestions by their IDs.
Suppose you want to apply changes with IDs 3 and 7. Your next prompt could be:
Apply the changes with IDs 3 and 7.
Rewrite the article exactly as it is, applying only the selected suggestions. Make sure you do not change anything else in the original article that the selected suggestions do not specify. You must not shorten or summarize the lesson.
After applying all the changes, carefully read the final version line by line to find any errors or omissions. If too many details were removed, give feedback and ask for another revision. Splitting the lesson into smaller sections can also improve the results.
Step 2: Add new knowledge to the lesson
Suppose one section of your lesson feels weak and needs more explanation or material. You can ask AI to help you improve it.
Find relevant articles online to explore the topic further, as shown in the article about researching with AI, then ask the LLM:
Here is the lesson I am writing:
[lesson]
I want to improve this section of the lesson:
[section]
Add more information, clearer explanations, and better examples based on the following content:
[content]
Write an improved version of this section.
Why use separators? They clearly distinguish the instructions from the content being processed. Avoid separators that already appear in the pasted content so the model does not get confused.
Step 3: Generate an introduction or conclusion
You can also use AI to automatically write a conclusion that summarizes the lesson or an introduction that quickly presents its content. For example:
Here is the lesson I am writing:
[lesson]
Write a conclusion that summarizes its content. Limit the answer to 300 words.
Step 4: Find sources for the facts you mention
Finding sources for the claims in your lesson both strengthens your audience’s trust and provides additional learning material. To do this, use a prompt like this one:
Here is the lesson I am writing:
[lesson]
Identify the key concepts that need a source, either for verification or to provide additional material. Find relevant sources online and explain why each is relevant.
Conclusion
Using LLMs to help edit lessons can transform your educational content workflow. From suggesting improvements to adding new information and finding reliable sources, these tools can save you time and improve the quality of your work.
However, remember to always verify AI outputs carefully. Errors or unintended changes can happen, but careful review will help you get the results you want.
Try these techniques and see how much more efficient and enjoyable lesson editing can become!
FAQ
How can you use AI to revise a first draft?
Give the model the draft, specify the improvements you want, and request numbered suggestions containing the original text, the replacement, and the reason for each change.
Why ask for suggestions before a complete rewrite?
This keeps you in editorial control and lets you accept only the changes that improve the lesson without losing details or your original intent.
Why assign IDs to the suggestions?
IDs let you ask the model to apply only selected changes, such as suggestions 3 and 7, instead of freely rewriting the entire text.
How can you add new knowledge to a weak section?
Provide the weak section and content from verified sources, then request an expanded version with clearer explanations and better examples.
Why use separators in a prompt?
They clearly distinguish the instructions, draft, and sources, reducing the chance that the model confuses the text to edit with the task it needs to perform.
What should you check after an AI-assisted revision?
Read the final version line by line, confirm every fact and source, and make sure nothing important was removed, summarized, or changed unintentionally.