Disruptive By Design: Preventing AI Slop

Slop (or AI slop) received the controversially coveted 2025 Word of the Year designation by both the U.S. Merriam-Webster Dictionary and the Australian Macquarie Dictionary. 

Recognition like this could be great publicity or the beginning of the end for a trend. But before jumping to conclusions, let’s look at the meaning. AI slop is generally accepted as low-quality, artificial intelligence-generated content—images, videos and text—saturating social media and sometimes insulting our senses. 

Others describe it as digital sludge or garbage—junk-like content. In some cases, it could be viewed as a poor mimicry or a mockery of generative AI’s inability to fully replace human creativity. With that clearer context and connotation, let’s shift to solutions—less slop, more shine.

One consistent variable I’ve observed over the last year with application to prevent AI slop is a solid start—that is, creating prompts that wax and shine.

I subscribe to the idea that AI is like an intern, complete with the enthusiasm, energy and exquisite ability not to read your mind but stick to the script actually given. As ethical teammates to the intern, we must give actionable content. Thus, a prompt with more left and right bounds, top ceilings, bottom floors and specifics for the interior theater show makes for the best performance and standing ovation results. 

A favorite acronym packing about as much as a Mary Poppins carpet bag is RICECO. I credit Max Anh’s August 2025 report for my first encounter with the RICECO prompt framework, but I have since seen it in a dozen or more places. It is guaranteed to give double-digit improvements in results, reduce hallucinations and improve your prompt efficiency (first time is the charm, not the fifth). 

The RICECO acronym breaks down as follows:

Role: What is the AI’s role? Student, teacher, subordinate, leader, scholar, layman.

Instruction: What is the exact task or action that AI will perform? Try verbs: summarize, create, give.

Context: What is the background information, such as the intended audience or setting for output use?

Examples: What is the muse in mind—tables, lists, diagrams, posters, etc., for style or format.

Constraints: What is the background information, such as budget, time, location, headcount or page limit?

Output Format: What should be the look and feel attributes of the final results—paper, tables, bullets, code, cards?

A quick example of RICECO in action for a comptroller (like myself) might read:

Role: You are a federal government chief financial officer (CFO) specializing in budget execution for defense agencies.

Instruction: Identify the top three CFO AI and automation opportunities to help the CFO execute all their current year funding and produce a prioritized unfunded requirement action plan that is designed to maximize time left in the fiscal year and near-term execution rates.

Context: It is currently mid-fiscal year (of an October 1 through September 30 fiscal year). Our execution rates are 55% in X appropriation, 40% in Y appropriation and 20% in Z appropriation. Our unfunded requirements are pasted below in three ranked groups. This description would be extended to plenty more detail with a multi-paragraph background of the specific situation, including sources, current challenges and goals.

Examples: I’d request a few initial products, such as an info paper summarizing at least three courses of action, but no more than five for the top three CFO AI and automation opportunities. I’d also include an example of PowerPoint products containing courses of action. Some ideas I’d heard of would be lists too—dashboards, GovGPT, etc.

Constraints: The proposed solutions must be free, must take less than five hours a week to manage and must not be overly technical yet specifically successful for CFO work.

Output Format: Present the final answers as a prioritized CFO AI and Automation Playbook with three sections: Quick Wins, Core Features and Long-Term Use Outlook.

The first result may need refining, especially if you’re unrehearsed. However, practice makes perfect, and waxing away at better prompts will reduce the slop and increase the shine for first-time AI results more often.