The Dangers of AI for Writers and How to Use It Wisely

By Doris Booth

CEO, Authorlink.com

If you’re a writer still vexed using Artificial Intelligence (AI) in your work, you are not alone. NPR reports that a recent Authors Guild survey found  only 13% of writers admitted using AI for activities like brainstorming character ideas and creating outlines. In reality, anyone using the Internet to perform the simplest tasks is actually dancing with AI–consciously or unconsciously. 

Whether you are paying a bill, traveling to a new location, or writing the next bestseller, you are interacting with AI. These special algorithms are embedded in every program and device we use, from simple spell-checkers to graphics generation. 

 And the technology is  not going away.

So the question is, how are we writers going to grapple with this monster? 

Writers are actually in a powerful position. We are the very ones who can shape whether technology will help or harm us. 

How  we, as wordsmiths, use AI; what we feed into it; what we allow it to do and not to do with our work; how we explain to others what we must do to control AI. 

The key to that power is understanding what AI can and can’t do, right now, and what it might grow to do to us later.  

Writers who learn to grapple with AI will actually be empowered to help shape their own destiny, and the destiny of the evil parts of technology. 

First, AI has no beating heart. It cannot suffer. 

Most importantly, it cannot use the human form of judgment because it has no real world experience. In its current state, like a jigsaw puzzle of trillions of bits of information, it can fashion an essay or a novel or picture by picking through everything it has gleaned from humans thus far. It can assemble the information in ways that seem like magic because it has access to all the knowledge we have fed to it for the last few decades. 

What is scary is when it becomes self-aware–when it is able to tell us what to do, without our advice or consent.  Experts say it will reach that point in about three to five- years from now. 

So what do we do now? Do we ignore it, hoping it will self-destruct? Or do we face the monster and take control of how we use it. It cannot lash out at us if it is caged inside  the way we design it. But we can still have access through the bars to its vast store of knowledge, while we maintain our creative integrity. 

Whether using Gemini,Chat GPT,Perplexity, Claude or a dozen others for instant responses,  the  key is to use AI not as a butler, maid, or crutch while we lounge on our sofas. Rather, let’s learn to use the systems as mere skeletons that only we humans can flesh out for the real world.

Fear of the unknown should not cripple us.  So let’s take another look at how we can help ourselves use AI–well, intelligently. 

In Authorlink’s analysis of full-length novels, we have learned there are many pitfalls to using AI solely on its own. Here is some of what we have learned from our new feature, Killer Query Letters.

When analyzing your novel or essay using AI, consider these questionable results: 

 Poorly-structured prompts: If you don’t know exactly what you are asking, you will not get useful answers. Poorly-structured prompts or queries return shallow answers.  If you don’t know what structural elements are involved in good writing, you won’t know how to  ask AI for the answer. 

Hallucination: an AI might say in error that a benign skin lesion is cancerous, or say that a verbal agreement is as legally binding as a written contract, or predict a tornado is headed your way when it isn;t. 

Drop out: In analyzing a manuscript, AI might have what we call “Information drop out.” The result might miss a key character or plot point. 

Misinformation: AI might compare your novel to works that have no similarity. 

Mechanical style: AI can write the answer in a flat, mechanical style, or use grandiose words that editors can immediately spot as AI-generated. 

Page Count: AI sometimes can’t count. Be sure you check the length of your submission against what AI reports. 

Systems vary: AI systems vary widely in how recently they were updated, and what information upon which they were trained. It’s a good idea to run your prompt on more than one system and compare them. 

Refusal to play nice: Sometimes AI will refuse to answer your query because it “thinks” your content is somehow damaging. It can also simply say it can’t answer because it hasn’t received enough training. Since some systems are involved in legal battles, owners of big systems may restrict what can and can’t be done with their platforms. 

What specific AI systems are designed for writers? 

Here are some strategies to effectively integrate AI assistance into a writer’s workflow:

  • Ideation and Brainstorming: Use AI to generate story ideas, character names, plot points, or creative prompts when feeling stuck.
  • First Drafts: AI can help overcome writer’s block by creating initial drafts or outlines, which the writer can then refine and personalize.
  • Editing and Proofreading: Utilize AI tools to check for grammar and spelling errors, improve sentence structure, and enhance overall readability.
  • Content Repurposing: AI can help transform existing content into different formats or styles, saving time and effort.
  • Research Assistance: AI can summarize articles, gather data, or generate reports, aiding in research and information gathering.

Remember, the key is to use AI as a tool to augment your writing process, not replace your own creativity. 

What are some common mistakes writers make when using AI?

To avoid the pitfalls of using AI, writers can:

  • Carefully structure prompts and queries: Clear and specific instructions will help the AI generate more relevant and accurate content.
  • Maintain a healthy skepticism: Always critically evaluate AI-generated content, checking for accuracy, consistency, and originality.
  • Use AI as a tool, not a crutch: Rely on your own creativity and judgment to guide the writing process.
  • Proofread and edit diligently: Don’t assume AI-generated content is flawless. Always proofread and edit carefully before finalizing your work.
  • Be mindful of biases: AI models can perpetuate biases present in their training data. Be vigilant in identifying and correcting any biases in AI-generated content.
  • Collaborate with human experts: Seek feedback from other writers or editors to ensure your work meets high standards of quality and originality.

Finally, as a writer, consider the  ethical issues when using AI in your work:

  • Plagiarism and Originality: Ensure AI-generated content doesn’t infringe on existing copyrights or intellectual property. Always cite sources and avoid presenting AI-generated work as entirely your own.
  • Bias and Fairness: AI models can inadvertently perpetuate biases present in their training data. Writers should be vigilant in identifying and correcting any biases in AI-generated content to ensure fairness and inclusivity.
  • Transparency and Disclosure: Be transparent about the use of AI in your writing process, especially when publishing or submitting work. Disclose the extent of AI involvement to maintain trust with readers and editors.
  • Accountability and Responsibility: Take responsibility for the final output, even if AI tools were used. Don’t blame AI for errors or inaccuracies in your work.
  • Impact on Human Employment:

Checklist  for using AI wisely

  • Check for obvious errors
  • Analyze word choice
  • Look for drop out points
  • Check for faulty conclusions or weak logic
  • Examine judgment in AI’s 
  • Consider ethical questions
  • Get a human expert involved to offer the human touch

Remember, it’s crucial to evaluate AI-generated content responsibility, not blindly.  The best way to use AI systems wisely is to engage a human editor who knows how best to navigate the systems–one with a beating heart. 

Learn more about Killer Query Letters. See real examples. 

About Doris Booth

Doris Booth is founder and CEO of  the award-winning  website, Authorlink.com, named first in its class by Editor & Publisher magazine. She is a journalist and researcher. In the early 90’s she initially programmed her own website and survived the dot com boom. Authorlink.com was among the first to offer ebook conversion services, delivering over 200 titles per year. In 2004 she partnered with iBooks, Inc.(Trade), the first trade publisher to simultaneously publish books in print and online.

Her literary agency, a separate unit from the magazine, represents a small group of bestselling authors both domestically and internationally. Two book projects in which she has been involved became a Netflix documentary series. She has been involved in book sales to HarperCollins, St. Martin’s Press, Simon & Schuster, Random House, Macmillan, Barnes & Noble, Farrar, Straus & Giroux, Mirror Books in the UK, Audible, Tantor Media and others. She has worked as a co-agent with industry icons such as the late Peter Miller and with agents at CAA Doris keeps a low industry profile but is probably best known as the past publisher and current agent for TED BUNDY, CONVERSATIONS WITH A KILLER, on which the recent hit NETFLIX docuseries is based. She is no longer accepting new agency clients, but the magazine, authorlink.com, offers several services for writers.

 Early in her career, her ad agency served Fortune 1000 companies, including high-tech organizations such as SAP America and DacEasy, and she wrote materials for the then chairman of the U.S. House Science and Technology committee.

This article was written with limited assistance from AI. The concepts are largely my own.