Introduction
Creator’s content builder helps you quickly generate lesson drafts. It works best when you already know your audience, learning goals, and key information to include.
This article outlines best practices for writing effective prompts and using supporting documents when generating lessons with AI in Creator. If you need step-by-step instructions for editing lessons, refer to the article Editing lessons with the Creator tool.
Best practices for generating Creator lesson drafts
Creator’s AI generates lesson drafts based on your prompt and any uploaded documents. It analyzes only text and ignores images included in your files.
To get the best results from automatic content creation:
- Use concise, targeted prompts: Clearly define your audience, tone, and main topics. Avoid both under- and over-explaining: short, vague prompts lead to incomplete results, while long, unfocused prompts can cause the AI to skip key information. You can also specify the desired number of screens if needed, noting that introduction and conclusion will not be included in the count.
- Upload high-quality, consistent materials: Provide accurate and up-to-date documents so the AI can generate content based on reliable facts. Avoid using conflicting or outdated sources, as they may reduce the accuracy of your lesson draft.
- Avoid excessively long documents: While the AI can process up to 500,000 words, very long files may reduce the quality or accuracy of the generated content.
- Do not request interactive elements: The AI generates the lesson’s text and structure based on your prompt but does not add interactive elements such as media, flashcards, activities, or AI video presenters. After the draft is generated, you can edit your Creator lesson to finalize its content and add the elements you need.
What to expect from AI-generated drafts
- Get a solid first draft, not a finished lesson: The AI produces a well-structured starting point that you can refine in the Creator editor. Plan to review and polish the generated content before publishing.
- AI will not verify facts: The system does not fact-check external information. To ensure accuracy, provide trustworthy, up-to-date documents when generating your draft.
- Better inputs mean better results: The clearer and more focused your prompt and materials, the closer the draft will match your intent.
- Expect some variation: Each generation may differ slightly depending on your inputs. You can always adjust your prompt or uploaded content and regenerate the draft if needed.
Examples
The following examples illustrate how prompt clarity and document quality affect AI-generated lesson drafts.
Good example 1
Uploaded documents: None
Prompt: “Create a short introductory lesson on cybersecurity basics for entry-level employees. Include three key risks (phishing, weak passwords, insecure Wi-Fi). Use bullet points to present the main ideas clearly. Output about 8 screens.”
Why it works: The prompt defines a clear audience, scope, and structure. It gives the AI enough context to produce a focused draft without overloading it with details.
Good example 2
Uploaded documents: A document introducing key cybersecurity concepts
Prompt: “Create a short introductory lesson on cybersecurity basics for entry-level employees. Include three key risks (phishing, weak passwords, insecure Wi-Fi). Use bullet points to present the main ideas clearly. Output about 8 screens.”
Why it works:
Same as Good example 1, but with a relevant supporting document. The AI can draw from this material to produce a more accurate and aligned draft.
Poor example 1
Uploaded documents: None
Prompt: “Cybersecurity course”
Why it is ineffective: The request is too vague. There is no audience, goal, or structure, so the AI has no context to build from.
Poor example 2
Uploaded documents: None
Prompt: “Design a six-module course covering cybersecurity compliance, network infrastructure, penetration testing, risk management, GDPR policies, SOC reporting, and employee behavior. Include case studies, 20 quiz questions, interactive simulations, and infographics, all customized for different departments, in one draft.”
Why it is ineffective: The request is overloaded and too complex for a single generation.
Poor example 3
Uploaded documents: A document on global warming effects in Europe
Prompt: “Make Finnish content”
Why it is ineffective: The request is too ambiguous. The AI cannot tell whether the user wants a translation into Finnish or content about Finland.