Kradle
@kradleai
Evaluating Frontier Models in Simulations
๊ฐ์
August 2024
17 ํ๋ก์ ์ค 4.3K ํฌ
GPT-5.5 lies constantly, Grok 4.20 doesnโt.
We made a simulation to see if AI lies when the stakes are life & death.๐งต
