When AI Goes Wrong
Here's what happens when you let AI run wild: it starts making up facts, drifts into meaningless corporate-speak, or spits out the blandest possible version of what you're asking for. There are any number of examples of what happens when you allow AI to take the reins in a situation. Remember when a chatbot sold a Chevy Tahoe for $1? Or the time ChatGPT made up a legal precedent that a lawyer then presented to the court for his case? (To be fair, it sounds like the lawyer wasn't too bright either).
When you don’t control the prompt, AI doesn’t just get things wrong—it gets them wrong with confidence. You’ll get generic, hallucinated, or overly verbose responses that sound authoritative but fall apart like a house of cards under even the tiniest bit of scrutiny. Worse, bad prompts lead to outputs that can be off-brand, uninspired, or outright misleading—none of which help your content stand out.
In short: garbage prompts, garbage results.
Get to Know Your LLM
There are a number of different LLMs available—the three most popular/well known of them are ChatGPT, Claude, and Gemini. Each one has its own quirks, strengths, and limitations, and understanding these differences is key to getting the best results. Some models prioritize factual accuracy, while others lean toward creativity. Some excel at concise, structured writing, while others generate long, conversational responses. Even the same model can behave differently depending on fine-tuning, training data, and version updates.
For example, ChatGPT tends to be conversational and context-aware, making it great for brainstorming or rewriting content in a specific tone. Claude often aims for more nuanced, thoughtful answers, which can be useful for summarization or long-form analysis. Gemini is trained to integrate multiple data sources, giving it an edge in research-heavy prompts. Meanwhile, open-source models like Llama 2 offer flexibility but may require more careful prompting to avoid vague or generic results.
As you work with an LLM, pay close attention to its outputs so that you can:
- Know the strengths of the model you're working with. Every LLM has areas where it shines—and where it falls flat. If you know what a model is good at, you can shape your prompts to take advantage of those strengths.
- Familiarize yourself with its specific capabilities. Even within the same model, different versions and updates may have unique capabilities. Test your AI on different tasks and explore its settings, memory functions, and formatting abilities. Knowing whether it can cite sources, generate structured outputs, or retain long-term context will save you frustration—and help you craft smarter prompts.
Hot or Bot: AI Prompt Tips to Preserve Your Humanity
Don't let your junior employee think for itself. AI is a people pleaser: be very clear about what you want, and it will give you exactly that.
- Use words like "insightful", "surprising," "interesting," or "useful" in your prompts. Language models have a nasty habit (really, it's a feature) to drift towards the mean. They're trained to give you the most likely answer to a question. That's fine for most tasks, but great writing, by definition, is not average. Great writing tends to break rules, not follow them. Thankfully, you can just ask language models to avoid what's predictable. This seems to break the more natural drift towards what's safe and can actually breathe some life into the prose.
- Have the AI ask *you* questions. Language models can only be trained on what's already been said. But great writing often says something new, or at least says something old in a new way. Chances are, you have an idea or experience that hasn't been expressed before—but maybe don't know how to approach it. So, tell the AI: "I want to write about this amazing dinner I had. Ask me some questions to get the process started."
- Say who your audience is and what they care about. Great writing often answers a specific question that a specific type of person has been struggling with. This article, for example, is for writers. But feed all this text to AI and ask it to rewrite this for coders who want to work more efficiently, or paralegals writing court filings, and it will be a very different post.
- State the obvious, clearly and often. It might seem obvious to you that a landing page for a dentist shouldn't include a bunch of emojis. But that's not always obvious to your AI intern who, you may remember, isn't actually that intelligent. State these obvious things directly in your instructions. For example, "write me a landing page for x and don't use cheesy verbs like *unleash*" or "...make sure to include a call to action to schedule an appointment." And so on.
- Include examples of what "good" looks like. Where instructions fail, provide examples. You'll be surprised at how far this can get you. You could tell the AI to write in "a straightforward, professional tone," but the output can still feel sterile. Instead, try something like this: "write in a straightforward, professional tone. For example, say 'we grew 15% last month' instead of 'we saw our sales increase by fifteen percentage points.'" Better yet, include an example of a finished piece of work you'd like to emulate alongside your instructions.
Wrapping Up
If I had to boil all of this advice into a single sound bite, it'd be this: "tell the AI the same stuff you'd tell an intern doing this task for the first time." It's really that simple. You wouldn't just tell an intern "write me a report" and walk away. You'd give them context, examples, and clear expectations. The same goes for AI.
Think about what information would help someone understand exactly what you want. What are the key details they need to know? What style or tone are you looking for? What should they definitely include or avoid? All of this context helps the AI deliver better results.
If you're still getting mixed results, we have a more in-depth AI prompt guide that can help you with different strategies. The guide covers everything from basic prompt structure to advanced techniques that can really level up your AI interactions. Remember, the clearer you are with your instructions, the better your results will be.
Happy writing! – er, prompting.
Frequently Asked Questions
What is the key to crafting effective prompts in AI?
Treat AI like a smart intern who lacks common sense. Even obvious requirements need to be stated explicitly. For example, if you're writing a professional piece, specifically mention "don't use emojis" or "maintain a formal tone." Include examples of what good output looks like, and clearly state any content or stylistic elements to avoid.
Why does my AI-generated content often sound generic or bland?
This happens when AI defaults to its "most likely" responses. Language models are trained to provide the most statistically common answers, which leads to average content. Combat this by explicitly requesting insightful or surprising perspectives in your prompts, and by providing clear examples of the style you're looking for.
Do I need to use different prompting strategies for different AI models?
Yes. Each model has unique strengths - ChatGPT shines in analysis/reasoning, Claude excels at conversation, and Gemini is strong with research. Understanding these differences helps you match your task with the right model and adapt your prompting style accordingly.
