How Fiction Writers Can Use AI Without Letting it Take Over the Process

In the age of AI, writers of all kinds – journalists, content writers, short story writers and novelists – may be tempted to let LLMs do much of the legwork.

But if a fiction writer wanted to just draw on the efficiency of tools like ChatGPT, without letting them completely take over the process, how could they do so?

Brainstorming

Sometimes ideas for stories don’t come as quickly, but there’s still the itch to write. You can use ChatGPT or another LLM for brainstorming. You could tell it, “I want to write a story about a man who discovers aliens in the forest. Give me five events that might happen once they meet.”

Another example prompt: “I have two characters. One’s from New York and the other is from Philadelphia. Their careers are unrelated. Give me five ways they could meet.”

You’re unlikely to like all its ideas, but it might spur an exciting train of thought. 

Assisting with research

As any writer who has put off writing themselves will know, it’s tempting to read about what other writers do for “research”. You read about their daily routines and their ways of finding inspiration and feel like you’re doing something (and perhaps sometimes it really does help!).

Maybe you want to see if there are any parallels between five writers you like and their routines. Copy and paste five articles into an LLM and ask it to point out any similarities. Perhaps it’ll point to discipline, or that each writer starts working in the early morning, or that they each have a dedicated office solely for writing – or perhaps something you wouldn’t have thought of.

If you’re doing lots of research and you’re not sure about the security of certain sites, a VPN will help by encrypting your internet traffic. Make sure to choose a provider with a no-logs policy. Explore ExpressVPN's global network on their site.

Proofreading

Often, proofreading our own writing is the hardest thing to do – we become blind to grammatical errors and spelling mistakes. We think we’ve done the right thing, so we see the right thing, even when it may be blindingly obvious to anyone else that there’s a mistake. LLMs are useful for pointing out errors. Depending on what model you use and the specific prompt, their default might be to give you a full edit, rather than simply pointing out any mistakes – so tailor your prompts accordingly. 

You can experiment with different options – ChatGPT, Claude, Perplexity, for example – and see which you like better. 

Summarizing papers

Maybe you’re interested in research about creativity, but not used to reading entire scientific papers and untangling all the unfamiliar terms (not many of us are). In these cases, feeding a paper into an LLM and asking it for a summary is surprisingly effective.

It can explain any unfamiliar terms if there’s something you’re particularly interested in or stuck on, turning difficult language (often used in such papers) into something more approachable. While it’s important not to rely on it for answers, the tech has become an effective tutor.

Takeaway

So there are a few ideas: brainstorming, research, proofreading, and summarizing tricky language. You needn’t completely give way to an LLM and hope that it magically blurts out some Shakespeare-level prose, but used strategically it can be a helpful tool.