By Purva Gupta, Co-Founder and CEO of Lily AI
Key Takeaways
- AI slop refers to low-quality media, including copy and images, created using GenAI.
- Low-quality content undermines trust, hurts user experience, and risks LLM model collapse.
- In retail and GEO, brands must prioritize quality content over quantity slop.
Content slop is nothing new and refers to low-quality, filler-heavy content. Think food recipe blogs designed to be found by search engines and inundate poor aspiring chefs with banner ads.
In the AI Era, we are experiencing a new surge of AI-generated content flooding the digital landscape, affectionately referred to as “AI slop.” This new wave of slop content has gotten so out of hand that it was featured on a recent episode of Last Week Tonight.
Zoom Out: Why AI Slop Matters
When left unchecked, AI slop leads to poor consumer experiences. In social, Pinterest recently faced backlash for overwhelming users with repetitive AI-generated pins.
Yet AI slop isn’t limited to social pins and posts. Similar to click farms and MFA (Made-For-Advertising) sites, entire websites of AI slop are on the rise, with a 717% increase in AI slop sites this year, according to Deepsee.io data.
Zooming out, there are even bigger issues at play. Unchecked AI slop poses the risk of model collapse, a phenomenon where AI systems degrade due to training on increasingly lower-quality content.
The critical importance of high-quality data is why Meta invested heavily in Scale AI for its human-labeled training data. Meta’s significant investment underscores the principle of GIGO (Garbage In; Garbage Out). Peak performance in AI relies on human-created content and human-verified data.
Zoom In: Slop vs Creme-of-the-Crop
You might be asking yourself: Is it possible to generate high-quality content with AI? Absolutely!
Is it easy? Absolutely not.
Yet “Creme of the Crop” AI-powered content can absolutely be achieved. To achieve that requires these fundamentals:
- Domain-specific, human-labeled core training data, continuously updated and dynamically assembled for scale
- AI model trainers and engineers with deep domain expertise
- Purpose-built, vertical-specific AI for optimal outputs and performance
- Human-in-the-Loop (HITL) and agentic AI insights and feedback loops
Lily AI has invested in all four of these areas, starting with merchant-led, human-labeled retail training data.
Slop Double Click: SEO vs GEO
One highly visible area where the high-stakes dynamics of slop vs quality content are unfolding is in the disruption of traditional SEO (Search Engine Optimization) and the rise of GEO (Generative Engine Optimization).
While traditional SEO was largely optimized for Google and navigating Google’s core updates, some black hat SEO tactics emerged that favored low-quality content, such as keyword stuffing and structured data spam, to boost rankings.
In today’s GEO world, it’s quality over quantity. Content that is most frequently cited by AI answer engines doesn’t always rank the highest in search engines. In fact, a recent study across three LLMs (Google AI Overviews, ChatGPT, and Perplexity) found that the top 10% of most cited pages receive significantly less traffic. Compared to their SEO-optimized counterparts, these top answer engine citation performers ranked for fewer keywords and received fewer backlinks.
Instead of competing for clicks as you do in SEO, in GEO, it’s a competition for trust.
How to Avoid Slop in Product Content
Let’s explore this concept through the context of retail. From highly visible ad copy and PDP descriptions to machine-optimized metadata, alt text, and schema markup, product content needs to:
- Be both highly accurate and precise; there is no room for error or misinterpretation.
- Include objective attributes, such as dimensions and features that are clearly labeled and ideally as detailed as possible.
- Additionally include subjective attributes, such as style, trend, fit, function, room setting, holiday, or occasion, are also critical.
- Be infused with natural consumer language, capturing how people today, across generations and geographies, naturally speak, shop, and search.
- Optimized and formatted for distribution according to destination and use case.
- Be in complete alignment with brand guidelines and product taxonomies.
This may seem like a tall order… and that’s because it is. It is not easy to use AI to create high-quality content. If it were, we wouldn’t have so much slop!
Fortunately, there are solutions like Lily’s AI-powered Product Content Optimization platform (PCO) that are absolutely worth investing in. With PCO, retail marketing, merchandising, and e-commerce teams can dynamically generate high-quality copy, attributes, descriptions, and metadata in order to achieve peak performance and productivity across on-site and off-site paid and organic search and discovery.
In Summary:
AI slop threatens consumer trust, brand reputation, and LLM model integrity. That doesn’t mean that all AI-generated content is slop, however. Lily AI’s Product Content Optimization platform ensures that brands and retailers can effectively leverage AI for content generation without sacrificing quality, ultimately delivering superior AI discovery.