1. LLMs struggle with low‑level reasoning and raw performance
Many commenters argue that LLMs are poor at tasks that require genuine optimization, scheduling, or bin‑packing and that hand‑crafted code can still beat them on efficiency.
“it sounds like a bad abstraction. im llms are really bad at reasoning through efficiency and utilization problems… they fail at scheduling and bin‑packing for the same reason.” – nostrebored
2. Abstractions are essential, but their quality matters
There is consensus that not all abstractions can be dropped; good abstractions dramatically increase productivity, while AI‑generated ones are often brittle or inefficient.
“Abstractions aren't a burden though, they are more like a support… a good abstraction lets you write 10 programs in the time it would have taken to write one.” – vatsachak
“We need abstractions more than ever… The AI will create/impose an abstraction for each piece of logic whether we like it or not.” – socketcluster
3. Credibility of academic claims is viewed skeptically
Several users question whether university research on LLMs carries enough real‑world experience, pointing out the gap between academic prototypes and production‑scale systems.
“I think it's generally fair to assume that you don't become a Stanford CS professor by being bad at writing code and creating abstractions.” – jrflo
“Faculty at Stanford probably have some decent hardware… their research code is very likely not being deployed in production, or at scale.” – raddan