Using AI as a research partner means asking it questions to fill gaps in your knowledge about historical periods, technical details, or cultural contexts before you draft the story itself. The AI excels at synthesizing existing information into accessible summaries, which can anchor your narrative in authentic detail—as long as you fact-check the output.
Using AI to research for creative writing feels risky. Won't it just make things up? Sometimes. But when you know how to use it correctly, AI becomes a research partner that accelerates background building, checks your assumptions, and finds details you'd otherwise miss.
The key is understanding what AI is good and bad at: excellent at synthesizing information and exploring implications, terrible at remembering exact facts. Treat it as a brainstorming partner, not an encyclopedia.
Pattern synthesis: Ask it to explain what a Victorian household routine looked like and it will synthesize general patterns from training data. Good for texture and plausibility.
Implication exploration: "If an alternate history split off in 1850, what would communication technology look like by 1920?" AI can follow logical threads. Not factual, but coherent worldbuilding.
Detail generation: Need to describe a marketplace? AI can generate believable sensory details (smells, sounds, crowds, goods) that fit an era. These feel right even if some details are invented.
Assumption checking: "Would this be plausible?" You can feed AI your plot point and it'll identify logical holes or note what would need to be true for it to work.
Specific dates, names, exact quotes, and technical specifications—AI hallucinates confidently. It will give you a made-up historical figure's name with complete confidence.
This is called "hallucination," and it's the core limitation. The AI doesn't know when it's inventing. So you must verify anything factual with real sources.
Step 1—Brainstorm and texture: Use AI to generate period details, setting descriptions, and plausible-sounding background. This phase is creative, not factual.
Step 2—Fact-check critical elements: Any detail that matters for your plot (specific technologies, political events, real locations, character professions) needs verification through external sources. Wikipedia, books, documentaries, expert interviews.
Step 3—Refine with confidence: Once you've verified the scaffolding, use AI to fill in gaps and add texture. You now know what's real and what's invented-but-plausible.
Step 4—Layer in specificity: Real details make invented ones believable. Once you know the actual streets of Victorian London, you can invent one shop and it feels authentic because it's surrounded by real context.
You're writing a story set in 1930s Shanghai. You ask AI: "Describe the daily life of a Western expat businessman in 1930s Shanghai." It generates plausible details about clubs, housing, business practices. These feel authentic. Then you verify three key things: real neighborhoods (Frenchtown, International Settlement), actual business sectors, period-appropriate transportation. With that grounded, you use the AI texture as authentic backdrop.
Try this: Pick a real historical setting for a story. Ask Claude: "What details would a character notice entering a [specific location] in [specific year]?" Generate 500 words of sensory description. Then pick three factual claims and verify them. You'll quickly learn what AI got right, what it invented, and how to blend verified facts with AI-generated plausibility into seamless worldbuilding.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.