FAIR

FAIR is the research-lab context in 133. 对谢赛宁的7小时马拉松访谈：世界模型、逃出硅谷、AMI Labs、两次拒绝Ilya、杨立昆、李飞飞和42 where Xie Saining worked with researchers including Kaiming He. The source presents FAIR as attractive because it concentrated top computer-vision researchers at a time when Xie chose it over OpenAI.

Culture Shift

The interview contrasts FAIR’s bottom-up research culture with a later period after ChatGPT when large AI labs increasingly held alignment meetings to decide one- or two-year research direction. Xie treats that shift as partly contrary to exploratory Research Taste, where a project often pivots through failure before its final story becomes clear.

Connections

• Meta — company context for FAIR.
• Xie Saining, Kaiming He, and Yann LeCun — people connected to the lab in the source.
• ResNeXt, Self-Supervised Learning, and Diffusion Transformers — research lines discussed through FAIR.
• OpenAI and ChatGPT — route and event used as contrast points.
• AI Organization Design — organizational tension between bottom-up research and top-down alignment.