Comment by chilmers
15 hours ago
Here's a good heuristic for predicting the ability of a modern frontier LLM at any non-automatable task: Is it hugely overrepresented in their pre-training dataset (e.g. code), and were they specifically post-trained on it using reinforcement learning (e.g. coding)? If the answer to one or both is yes, you should expect them to be very good at it. Otherwise, you should expect them to be, at best, as good as an average human who's read a few books on the subject and is good at bullshitting.
I think this is true of any trained anything. It's the fish climbing a tree scenario. The teens on Ellen getting laughed at for not knowing how to fold a paper map.
There's infinite front end code to train on. Virtually infinite python and shell scripts too. As I've worked my way away from such domains, the LLM outputs have gotten worse. It just makes sense.
"average human who's read a few books on the subject and is good at bullshitting" seems like a large portion of white collar workers that make bank.
The most sane comment I've seen in this thread