"There's a new generation of empirical deep learning researchers, hacking away at whatever seems trendy, blowing with the wind" [D]

A pointed critique from an AI researcher has gone viral, sparking a heated debate about the current state of machine learning. The post calls out a perceived new generation of "empirical deep learning" practitioners who are "hacking away at whatever seems trendy" and "blowing with the wind." This criticism targets a research culture that prioritizes quick experiments on fashionable topics—like the currently buzzy concept of "agentic AI" (systems that can autonomously take actions)—over developing deeper theoretical understanding or reproducible, foundational science.

The discussion, which originated on social media platform X and spilled into forums like Reddit, resonates with many in the field who are fatigued by rapid hype cycles. Commenters note the struggle to even define terms like "post-agentic AI," suggesting the vocabulary is evolving faster than the substance. The core concern is that this trend-chasing approach, while generating impressive demos and papers, may be creating a fragile knowledge base and diverting talent from solving harder, less glamorous problems that are essential for sustainable advancement.