No More Slop! We Can Use AI Better.

AI is everywhere. It reads our emails, writes our news articles, it even creates our art, albeit poorly. And that makes sense to some degree. It is a powerful tool, and it's only natural that we want to take it out for a spin to see what it can do. And boy did we do that. Companies and individuals alike have jumped on the chance to integrate AI into everything. Even my baby monitor is AI-enabled to tell me when my child will be ready to sleep next! (I'm good, thanks.) And for a while, it was hard for people to spot AI-written content and folks could just pump out content with almost no overhead.

But, the backlash is here. People are starting to detest hearing the same humanity-free, em-dash riddled voice in their articles and websites, seeing the same sketchy art style in every LinkedIn post, and sitting through hastily thrown together presentations of AI slide decks with subheadings that cut off mid-word. If you use AI badly, we can smell it and there are few more off-putting smells in content these days.

As someone who has led a team of writers and been responsible for QA for large content banks, I can tell you that correcting terrible AI slop passed off as a draft is no longer acceptable, if it ever was. I got tired of giving individualized tips and edits based on the AI content I was getting back and got proactive. I made a training guide to help writers use AI to actually improve content. And the time savings isn't bad either.

If you know how to use it, AI can seriously up your writing game. But it's not just about crafting the right prompt to get the best output. I think it's actually about how you think about the role of AI and your relationship to its content that really makes a big difference. So, I'm not going to give you a simple dos and don'ts of prompt building. Let's talk at an ideological level about how to use AI to improve your content and ditch the slop.

The Takeout Choice Method

When I'm struggling to decide what kind of takeout I want, my husband will quickly pick a random restaurant we like so that I can consider that option and then decide what I actually want by comparison. AI does that, but for basically any content I'm trying to create. It gives me an initial first draft or a list of ideas. Something about it feels right or wrong, and that reaction generates an idea.

There is nothing more intimidating than a blank document. AI eliminates that initial block. Even if you hate what it gives you, the process of deciding why you hate it helps you determine your actual focus and improve your final product.

The Crowd-Sourcing Method

While AI is absolutely not as smart as you are (and don't ever forget it), it is great at knowing what is out there and coming up with a few jumping-off point ideas for your human brain to run with.

I might say something like "Give me 5 ideas for a fun and creative fundraising event for a local historical museum." They might all be mediocre, impractical, or derivative, but they help spark my creativity and provide a direction for me to focus my research and efforts.

The 80% Method

The overall advice I gave my writing team was always "AI can get you 80% there at best. You take it across the finish line." This helped give them a quick and easy reference point for how I wanted them to use AI generally, and was broad enough to apply in a variety of contexts. It also drastically reduced the amount of awful AI slop some contractors tried to pass off as original writing.

The Piecemeal Method

Don't expect AI to create the whole kit and caboodle in one prompt, especially for complex tasks. Instead, use your human brain to give it more direct prompts for specific pieces of what you need.

You know what AI is really, shockingly bad at? Writing multiple choice questions. As it turns out, writing believable but inarguably incorrect distractors is pretty tough. So, I encouraged my team of instructional content writers to break up the content into pieces. Instead of asking it to write a multiple choice question assessing the user's ability to apply the 1st Amendment in a real-world context (social studies was my favorite area), ask it for 5 common misconceptions about the 1st Amendment. Then use those to craft your own distractors.

There will likely come a time when AI can write an entire multiple choice question that is difficult to answer because of the content it is assessing not how confusingly it words the question, but today is not that day. So, play to its strengths and break the task into pieces.

The Template Method

The key to any usable content strategy is scalability. We want to take the time upfront to create tools that can help us create more assets faster. AI can be a big help here. If you have a couple of examples of similar content, or even just one strong piece of writing, give your LLM that piece as an exemplar and ask it to create a template for similar pieces. If you're specific about your prompting, you can get it to create notes to you or your team for future use that you can paste into the document right alongside the template.

Having these scalability tools as a starting point can increase your efficiency and editorial cohesion across writers while retaining the human element and avoiding embarrassing AI-induced factual errors in your copy.

The big takeaway is that AI content is never the final draft. Ever. But, it can help you save time and do better big picture thinking if you know its uses and limitations. AI is not a partner or an employee, and it is certainly not a subject matter expert. It's a power tool. And like any power tool, it can help you build beautiful things if you know how to use it. But if you don't know how to use it, it can make a real mess.