Filippos Vlahos, Guillaume Bied, Tijl De Bie
Grokipedia launched in late 2025 as an encyclopedia written entirely by the LLM Grok, pitched as a less biased alternative to Wikipedia. The obvious question: does handing the writing to an AI actually produce neutrality, or does it just swap one slant for another? The authors test this at scale, comparing 1,394 matched article pairs about government officials across nine ideology dimensions, using four different LLMs as judges.
Because the judges could be biased too, they also examined the judges' own patterns. Every judge, including Grok itself, rated Grokipedia as less neutral than Wikipedia. Both encyclopedias came out reading favorably overall, just toward different groups: Grokipedia leaned toward economically right-wing politicians and was harder on socially liberal ones, while Wikipedia tilted the other way.
Keep in mind this rests on LLM judgments of neutrality, and I am summarizing the abstract, so read the paper for the method and its limits.
Online encyclopedias shape political opinion and, through it, democratic discourse. In late 2025, Grokipedia was released, an encyclopedia written entirely by the LLM Grok. One motivation behind the project was to provide an unbiased alternative to Wikipedia, which has faced accusations of "left-wing" and "liberal" bias. But does an encyclopedia written by an LLM deliver greater neutrality, or does it simply embed a different ideology? We conduct a large-scale political bias…
Improving Improved Kernel PLS
arXiv (cs.AI) · July 17, 2026When Do Multi-Agent Systems Help? An Information Bottleneck Perspective
arXiv (cs.AI) · July 17, 2026ToolSciVer: Multimodal Scientific Claim Verification with Visual Tool Augmented Reinforcement Learning
arXiv (cs.AI) · July 17, 2026A Methodology for Auditable Trustworthiness Levels in AI Lifecycle Governance