• 0 Posts
  • 23 Comments
Joined 2 years ago
cake
Cake day: June 21st, 2023

help-circle


  • I recently asked ChatGPT to generate some boilerplate code in C to use libsndfile to write out a WAV file with samples from a function I would fill in. The code it generated casted the double samples from the placeholder function it wrote to floats to use sf_writef_float to write to the file. Having coded with libsndfile over a decade ago, I knew that sf_writef_double existed and would write my calculated sample values with no loss of precision. It probably wouldn’t have made any audible difference to my finished result but it was still obviously stupidly inferior code for no reason.

    This is the kind of stupid shit LLMs do all the time. I know I’ve also realized months later that some LLM-generated code I used was doing something in a stupid way, but I can’t remember the details now.

    LLMs can get you started and generate boilerplate, but if you’re asking it to write code in a domain you’re not familiar with, you have to understand that — if the code even works — it’s highly likely that it’s doing something in a boneheaded way.




  • I’m thinking about getting the new FairPhone 6 when it comes out and running /e/ OS, but I’m so reliant on Google Maps and Gmail (my email account, not necessarily the app … but I do rely on the app).

    I’m afraid that I’ll either install Google apps and end up with a phone just as compromised as a stock Android install, or if I don’t it will be too much of a pain in the ass to use.



  • I only use Windows because I have to work with a corporation’s IT helpdesk staff to get on their VPN if I want to do contract work for them. They are not likely to help me get connected from Linux; they’ll just find another contract dev. Once in, I do everything in Linux because my code will ultimately run in a Linux cloud container of some sort. WSL works well enough for me to do this. I’d rather have Linux on bare metal, but whatever. I’m in; I’m coding; I’m getting paid. I’ll put up with a little bit of suck.








  • As far as I can tell from the article, the definition of “smarter” was left to the respondents, and “answers as if it knows many things that I don’t know” is certainly a reasonable definition – even if you understand that, technically speaking, an LLM doesn’t know anything.

    As an example, I used ChatGPT just now to help me compose this post, and the answer it gave me seemed pretty “smart”:

    what’s a good word to describe the people in a poll who answer the questions? I didn’t want to use “subjects” because that could get confused with the topics covered in the poll.

    “Respondents” is a good choice. It clearly refers to the people answering the questions without ambiguity.

    The poll is interesting for the other stats it provides, but all the snark about these people being dumber than LLMs is just silly.