AI automation’s potential in notification pipelines and kernel optimization
A user argues that frontier AI models could cut notification-pipeline costs to a bare minimum with the right guidance, and that kernel optimization need not demand much creativity.
TLDR
In two replies, a user argues that even “novel” tasks can be highly automatable with frontier AI models. They describe notification pipelines as having obvious success criteria and say the right guidance could bring costs down to a bare minimum. For kernel optimization, they argue that a baseline ability to exploit symmetries can be enough to keep optimization running. They also report getting image-processing code for Allwinner neural processing units to run very fast.
AI automation’s potential in notification pipelines and kernel optimization
A user argues that frontier AI models could cut notification-pipeline costs to a bare minimum with the right guidance, and that kernel optimization need not demand much creativity.
TLDR
In two replies, a user argues that even “novel” tasks can be highly automatable with frontier AI models. They describe notification pipelines as having obvious success criteria and say the right guidance could bring costs down to a bare minimum. For kernel optimization, they argue that a baseline ability to exploit symmetries can be enough to keep optimization running. They also report getting image-processing code for Allwinner neural processing units to run very fast.