Only 11% of Domains Cited by Both ChatGPT and Perplexity in Analysis of 680M Citations
Key Takeaways
- ▸Only 11% of domains are cited by both ChatGPT and Perplexity across 680 million analyzed citations
- ▸The low overlap indicates fundamental differences in how AI systems select and prioritize information sources
- ▸This divergence may result in users receiving different information from different AI assistants for the same query
Summary
A new analysis of 680 million citations reveals a striking divergence in how different AI search and chat systems source their information. According to research shared on social media, only 11% of domains are cited by both ChatGPT and Perplexity, suggesting these AI systems have fundamentally different approaches to information retrieval and attribution. This finding raises important questions about the consistency and reliability of AI-generated responses across platforms.
The research highlights how AI companies are making distinct choices in their information architecture, potentially influenced by factors including web crawling strategies, partnership agreements, content licensing deals, and algorithmic preferences. This divergence could have significant implications for content creators, publishers, and websites seeking visibility in the AI-mediated information ecosystem.
The low overlap in cited sources suggests users may receive substantially different information depending on which AI assistant they use, even when asking identical questions. This variability underscores ongoing challenges in AI transparency and the need for users to understand that different AI systems may draw from vastly different knowledge bases, potentially affecting the comprehensiveness and balance of responses.
- The finding raises questions about transparency, consistency, and potential bias in AI information retrieval systems
Editorial Opinion
This research reveals a concerning fragmentation in the AI information ecosystem that most users likely don't recognize. While competition between AI systems can drive innovation, such dramatic differences in source selection raise questions about whether any single AI assistant can provide truly comprehensive answers. The industry needs greater transparency about how these systems select sources, and users should be aware that their choice of AI assistant significantly shapes the information they receive.
