Technology

This article is so confusing because it seems like everyone they're talking to is just using online models and the use of local models is mentioned but it's not clear how many of the people being interviewed are using local models since it's all about laptops. Even the one lady who you can see her screen is a CLI, it's not clear that she's not just using the CLI version of Claude.

I have a mid-range desktop and doing local LLM can be pretty darn slow on it especially with an AMD card and ROCm as opposed to Nvidia and CUDA. I have a relatively nice laptop, but it's specs are well below my desktop and I just can't imagine actually running a local LLM on a laptop.

If they're not using a local model, then they wouldn't need to worry about overheating with the lid closed. Easy to make it so it doesn't hibernate when the lid is closed via CLI (at least in Linux anyway). Because if they're offloading all the work to a remote model, their PC can essentially be relatively idle and draw less power/produce less heat.

Article also seems strangely focused on Macs? All it's mentions of how to make it so you can close the lid are Mac-focused. Did I miss something about the new Apple Silicon being really efficient for local LLMs? Maybe that's what I'm missing here.

It just seems almost weirdly narcissistic, like they want people to ask them about it so they can talk about it. Certainly it seems that way with the kid with a startup business that he runs during classes with tokens paid for by his parents.

Anyway, the whole thing seems odd to me. Either the article is about people who aren't actually super savvy coders or techies, or they would... just switch it so they can close their fucking laptop... or something about making a show of what they're doing is part of it. I dunno, weird. Anyway.