GPT-4 is getting worse over time, not better

Re:GPT-4 is getting worse over time, not better by meroes on Wednesday July 19, 2023 @12:50PM (#36789452)

And yet my Google engineer friend tells me to use Bard for my math coursework. I know it’s just an anecdote but what’s with the hype?

I was paying attention and it’s why I refuse to use LLMs for math, but I’m being told by people inside the castle to do so. So it is not so black and white in the messaging dept.

[ Parent | on HN ]

Re:GPT-4 is getting worse over time, not better by csmpltn on Wednesday July 19, 2023 @03:02PM (#36791547)

I have a tenured architect proposing that we stop sending structured data internally between two services (JSON) - and instead pass raw textual data spat out by an LLM which linguistically encodes the same information.

Literally nothing about this proposal makes sense. The loss of precision/accuracy (the LLM is akin to applying one-way lossy and non-deterministic encoding to the source data), the costs involved, efficiency, etc.

All just to tell everybody they managed to shove LLMs somewhere into the backend.

[ Parent | on HN ]