WHAT NEW IN X11.06 FRI

X 上的 AI 合集档案 · 23 个合集 · 69,408 条

gemini-4POST

质疑 Gemini 4 Argon 在 Minecraft 写 2783 个程序的可靠性

作者认为 2783 个自写程序只衡量产出量而非可靠性,呼吁公开失败运行次数,才能判断 Argon 写代码的真实水平。

Sebastian Buzdugan
@sebuzdugan17 天前收录原推发布于 2026.10.18
Sebastian Buzdugan @sebuzdugan原推发布于 2026.10.18
在 X 上打开原推

why does gemini 4 argon writing code in minecraft matter? code turns a plan into actions a computer can run, rather than advice someone must follow. the reported 2,783 self-written programs measure output, not reliability. i'd want to see failed runs too

x.com/sebuzdugan/status/2111809523094827047采集于 2026.10.19 10:01