Home Creators Posts Import Register Favorites Logout

Content

Let's use today's post to discuss something that really weighs heavily on my heart: AI slop and military education. Now, to start this discussion fairly, AI has arrived and is here to stay. Every business you interact with will use it, and we are also using it. Generally, we use AI for two things:

Correcting grammar, formatting, and typos in text: Every text at RBT starts its life as a fully hand-typed 600 to 1,000-word article. Before we publish an article, we use AI to find typos and correct them. This saves us valuable time to create more content and focus on creation instead of inspection. Furthermore, we use AI to fact-check historic and scientific claims. But even the AI-corrected text is again inspected and rewritten by us in order to provide you the best possible experience.

Infographic creation: Publishing illustrated articles every day won't be possible at that quality without AI. What we do is use a custom prompt that tells the AI model the amount of graphics we want, how they should look, and what topics the graphics should focus on. However, the graphics rarely fully display what we want or they have serious professional mistakes (which will play an important role later on). We manually adjust all of that in Photoshop and also align the style with our art style.

To nail it down: We do what is fully established in professional and academic circles—using AI for repetitive work, but not for content creation as a whole. The creation time for an article is still way over two hours right now.

The Rise of Tactical "AI Slop"

This brings us right down to a key issue I've started observing on social media more and more: "Educational Pages" that fully rely on AI. This means that texts are fully AI-written, and all images are AI-generated. If AI provided perfect answers, this all wouldn’t be an issue. But if you check such content out a bit closer, you see a lot of discrepancies that should make every NCO or commissioned officer raise an eyebrow. Things are taken out of context, and you quickly see that the person providing that kind of content either hasn’t done the task for real a single time or was just too lazy to even read the whole thing.

This is further confirmed by such pages often lacking any type of actual real-world photos. As the saying goes, "If there isn’t a picture, it didn’t happen." Pictures and action shots aren’t only a way to make content more interesting; they are also a way to show that the creator actually put in the work. A fully infographic-based page can be operated by anyone, and the person behind it could be a guy spending more time in front of a screen than in the field. Indeed, you could run such a page and never set foot outside.

This is where things get dangerous: The information is collected in the most dangerous way possible, there is no real-world confirmation, and there is a huge risk that the information is completely wrong.

Quantity Over Quality

We had the same issue some years ago with a bunch of pages creating content that was nothing more than copied and pasted open-source field manuals. First off, open-source field manuals are often outdated, meaning the SOP can have changed drastically. Furthermore, with no added notes, there is actually zero value to this over simply downloading the full manual yourself. "AI Slop" content is so dangerous because it allows creators to put out wrong or incomplete information that looks professional and visually appealing in a matter of minutes, spamming the internet with bad tactics.

As an example: I just asked an AI "How To Conduct A Recon Operation." The result was a mix of SIGINT, HUMINT, and Recon manuals all put into a blender. I reminded the AI that this was inaccurate according to military doctrine. It quickly admitted it was wrong and broke down reconnaissance a bit more accurately, but the result was still what we would consider slop. After looking over it, I recognized that the AI still missed important pillars like PIRs. I reminded the AI that this article completely lacked PIRs. The AI "excused" itself for that mistake and added them. We could have gone down that route even further, as many details were off.

Remember that AI indiscriminately crawls manuals (both up-to-date and outdated) as well as social media platforms (including Reddit). The results will never be of the same quality as coming handwritten from someone who actually learned and did the thing.

How to Protect Yourself and Support Real Creators

Generally, we don’t have a practical issue with AI slop content because our followers and supporters (you are awesome) know the difference between legitimate information and mass-produced content. But for new, high-quality content creators, AI slop miseducation will be a huge issue. Legitimate creators may only be able to provide a few articles or posts a week, while AI sloppers can push buttons and put out at least a post per day. They are on the verge of winning by sheer quantity.

So please, be kind enough to check for a few things before you follow a page:

Doctrine: Does the content align with any established doctrine you know?

Experience: Does the content just regurgitate doctrine, or does it add personal notes, experiences, and opinions?

Visual Proof: Is there any visual proof the creator set foot outside practicing what they preach?

If the answer to all three is no, please do not follow or share such pages. You are at risk of sharing inaccurate and therefore dangerous information.

There is nothing bad about using AI as a part of your content creation process, but as soon as AI is essentially all of your content, it becomes a scam rather than information.

As always, thanks for your attention.

Files

Comments

Preston Walsh

AI for repetitive tasks are exactly what machines are for. Creating, dreaming, forming new thought, invoking emotion, and conveying ideas to an actual human audience is why human creations will always be absolute. Machines are not made in God's image, we are.