pull down to refresh

Wondering if prompt engineers will swallow software engineers, or if software engineers will swallow prompt engineers.
Asking because I just asked the agent in Cursor to implement DnD support for my wallets, and it did a decent job. I just had to tell it a few times to make the code suck less.
DnD support for my wallets
I wonder how and why you’d play Dungeons and Dragons on btc wallets
reply
8 sats \ 1 reply \ @ek 23 Jun
Oh, haha, I was too deep in my code. I meant drag & drop!
reply
hahah as a programmer myself I shoul've figured that out 😂
reply
Wondering if prompt engineers will swallow software engineers, or if software engineers will swallow prompt engineers
About that, I think the latter is more likely.
reply
0 sats \ 4 replies \ @ek 23 Jun
Why?
reply
293 sats \ 3 replies \ @felipe 23 Jun
Because we (software engineers) are much better at providing the necessary context (both technical and business) to LLMs. We are also much better at breaking problems in smaller sequential steps because we (sort of) understand how the machine works
reply
8 sats \ 2 replies \ @sox 23 Jun
^ completely agree, prompt engineers have to be software engineers for starters
reply
have any of you met someone whose title was prompt engineer?
i never met one, or looked at a job posting, but when people told me about it I assumed that they were still programmers but their task was to optimize prompt inputs to LLMs, which itself can be an iterative optimization process
reply
have any of you met someone whose title was prompt engineer? I never met one
Neither have I. But what I have seen is non-programmers trying to "vibecode" ideas here and there, which is totally fine and even helpful, sometimes. I'm actually instructing some people on how to produce html prototypes using LLMs so they can better describe what they want me to do
Before LLMs it was like this:
To be honest, it still is but it made iterations faster