Study finds AI chatbots offer narrower knowledge than Google search
A University of Copenhagen study compared 27 large language models from OpenAI, Meta, Google, and Alibaba against Google search. Researchers tested answers on 155 topics across 12 countries, using 200 prompt variations per topic. The models generated about 1.7 million answers and 70 million claims, revealing that AI chatbots provided less varied responses than search engines.
Key points
- University of Copenhagen study tested 27 AI models from OpenAI, Meta, Google, and Alibaba on 155 topics
- GPT-5 showed **18.7% less variety** than Google search in answers across 12 countries
- Smaller models and newer versions performed better than older ones but still lacked Google’s breadth
The study found that even OpenAI’s GPT-5, the most diverse model tested, delivered at least 18.7% fewer perspectives than Google. Smaller models and newer versions performed better than older ones, but none matched Google’s breadth. The issue stems from how AI models compress training data, prioritizing common patterns and filtering out outliers. Search engines, by contrast, expose users to multiple sources and viewpoints, broadening knowledge exposure.
The story so far
2 episodes →- Study finds AI chatbots offer narrower knowledge than Google searchthis story
AI chatbots serve you a narrower slice of knowledge than a plain old Google search, study finds
digitaltrends.com · 28 September 2026
Loading the full article…
This text was published by digitaltrends.com and written by Manisha Priyadarshini. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- xAI releases Grok 4.7 on Amazon Bedrock · 1 src
- Anthropic to launch Sonnet 5.5 next week, targeting cost savings · 8 src
- Anthropic unveils Claude Opus 5.5 with 40% lower costs and top safety ratings · 6 src
- Claude Sonnet 5.5 available on Amazon Bedrock and Claude Platform on AWS · 1 src
- OpenAI launches GPT-6 Sol and Luna with lower API prices and improved benchmarks · 18 src
Comments
via GitHub Discussions