Skip to content
HNHacker News·
Not on the current live radar

AI Model Groupthink

AI summary

A study on AI model groupthink found that Chinese models, such as DeepSeek's deepseek-v4-flash-0731 and Kimi's kimi-k2, tend to align more with consensus opinions. The research involved twelve models, including Gemini 3.1-flash-lite and Grok 4.3, accessed via OpenRouter, mostly in their "flash" or small variants with reasoning often switched off. The Temperature parameter was not adjusted, and default reasoning behaviors were used where applicable.

Time & source

Times shown in UTC

Display time zone: UTC

Local time zone unavailable; showing UTC.

IngestedOffset at this time: UTC+0Oct 7, 2026, 19:00 UTC

Ingested
Oct 7, 2026, 19:00
Source type
Unclassified

Full text isn't available here.

Read at source →
Source·Hacker News·magicnumbers.io