AI Slop: A Closer Look at Its Impact on the Internet
In a study published today in Wired, researchers from Imperial College London and Stanford University have revealed that approximately 35% of all new websites created between 2022 and 2025 are either AI-generated or AI-assisted. This finding, based on data collected through the Internet Archive’s Wayback Machine, underscores a significant shift in internet content.
The research also found that online writing is becoming “increasingly sanitized and artificially cheerful.” Using sentiment analysis, which evaluates words as positive, neutral, or negative, the study discovered that AI-generated websites exhibit an average positive sentiment score 107% higher than non-AI sites. The researchers interpret this heightened positivity as indicative of a sycophantic and overly optimistic nature among existing language models.
The findings also shed light on other characteristics of slop (AI-generated content):
- Artificial cheerfulness: By examining the tone, the study found that AI-assisted websites displayed an average positive sentiment score 107% higher than non-AI sites.
- The range of ideas and viewpoints: The researchers discovered that AI-generated content tends to be more ideologically diverse, with AI-based sites scoring roughly 33% higher on testing for semantic similarity than human-made websites.
However, the study’s findings also contradicted some expectations:
- AI leading to misinformation: Contrary to popular belief, the analysis of evidence did not support the hypothesis that AI-generated content would lead to more misinformation. The study found no link between AI and increased false information.
- Linking to external sources: The research did not confirm that AI writing websites were less likely to connect with external sources or exhibit a more generic writing style compared to human-written sites.
The findings highlight the widespread impact of AI on internet content, presenting both opportunities and challenges for writers and readers alike. While the study suggests that AI may lead to increased positivity and diversity in ideas, it also raises concerns about potential biases and less varied viewpoints.
As with any emerging technology, these results underscore the need for continued monitoring and debate around the ethical implications of AI’s growing presence on the internet.
Key Takeaways:
- The study found that approximately 35% of new websites created between 2022 and 2025 lean on AI-generated or AI-assisted content, indicating a significant shift in internet content.
- Online writing is becoming “increasingly sanitized and artificially cheerful” as sentiment analysis indicates an average positive sentiment score 107% higher for AI-assisted sites compared to non-AI sites.
- The findings also suggest that AI-created content might not always be ideologically diverse, with AI-based sites scoring roughly 33% higher on testing for semantic similarity than human-made websites.
- While AI can lead to increased positivity and diversity in ideas, it has also been found not to link to external sources or exhibit a more generic writing style compared to human-written sites.
Note:
The study’s findings suggest that the internet is becoming more sycophantic and overly optimistic due to AI-assisted content. However, it also raises concerns about potential biases and less varied viewpoints in AI-generated writing.
Originally published at Unknown. Curated by AI Maestro.
