papers-please

Good recent paper from Apple on LLMs reasoning capabilities. Their claim is that it still looks more like sophisticated pattern matching, which makes sense intuitively.

We found it to be a pretty robust and intellectually honest evaluation.
https://arxiv.org/pdf/2410.05229

One of the biggest issues with LLMs isn't the technology, but how its sold to the public. Apple may admire it internally with papers like this, however, externally they're positioning it's capabilities in a really lazy way that does a disservice to how it can thoughtfully help people. 

Bridget Carey nails this in her review, and it seems to have resonated with others since it's one of their most viewed videos.

One of the biggest issues with LLMs isn't the technology, but how its sold to the public. Apple may admire it internally with papers like this, however, externally they're positioning it's capabilities in a really lazy way that does a disservice to how it can thoughtfully help people. 

Bridget Carey nails this in her review, and it seems to have resonated with others since it's one of their most viewed videos. 
https://www.youtube.com/watch?v=D0V554NyXWM

off & onchain operations
<--->
Sharing ideas over at /workflow <--->
Problem solving is my jam😎
🎩

seemore.tv/free/fc/adam-

🥑🍋 Farmer / LAMP stack dev / Husband / 🌈 Dad / Web1 dude loving a web3 world / Collector of things digital & physical / 
CdeP ‘94 

Thats exactly it's main use case IMO. Make existing things better. 
Initially Apple seemed to understand this with how they were talking about leveraging LLMs to make Seri more in line with how chatGPT handles nuanced questions. I'm sure they'll get there but currently they're publicly pushing this in the wrong direction.

What if AI isn’t ready to be a consumer app but there are still a few thousand ways these LLMs can be worked into existing solutions etc.