4 ms·

Neal Wu recently posted a video of him attending Meta Hack Coding challenge. This time around, Meta has two score board, one for all human and the other one, I

Neal Wu recently posted a video of him attending Meta Hack Coding challenge. This time around, Meta has two score board, one for all human and the other one, I believe is AI assisted. Shockingly, the leader board for human was significantly faster than leaderboard for folks who used AI. We have a long way to go before AI surpasses human.

discordance 2y ago
A competition like that is for the top 1% and very difficult problems. For most people and most problems, AI is very helpful and speeds up the process. As an example, I made a browser extension in 5 minutes this weekend using chatgpt by telling it what I wanted. I didn't know how to make a browser extension before that, so you could say AI surpassed this human quite easily.
- krater23 2y ago
  So AI is the next way to give people that don't know what they do new tools to generate and release shit where they don't know what it does to customers that don't know how much shit they got. I'm sure this will bring the mankind further...
  - Ozzie_osman 2y ago
    People said similar stuff about the printing press.
- jppope 2y ago
  As expected, AI is great for stuff thats been done a million times. AI will slow you down for anything even slightly novel.
  - 93po 2y ago
    for now
    - hatefulmoron 2y ago
      How long until the statement is false? You seem to have something in mind.
    - tovej 2y ago
      for all of time as long as machine learning = AI If it's not in the dataset the AI won't handle it correctly (unless it's trvial and a linear model is good enough, but then why even use AI)
  - JamesBarney 2y ago
    99% of coding is doing something that's been done a million times and gluing it together in a novel way.
- beeflet 2y ago
  Yeah I guess it helps for applications where you otherwise need to read a ton of documentation to get started. For web stuff it helps me find useful APIs sometimes, but other times it wastes my time with deceptively wrong answers. The task of programming or engineering is generally one where you spend 80% of the time solving bugs in 20% of the code. LLMs are good for suggesting possible solutions based on existing things, but not for verifying solutions or doing any sort of real problem solving (they generally waste more time than they save because they give plausible but wrong answers). If you just want to reproduce some boilerplate or example code it is great though because documentation tends to lack good examples, and LLMs have a ton of real code stored in their latent space. Sometimes I even just use it as a thesaurus, if I have something on the tip of my tounge.
rsynnott 2y ago
> Shockingly Yeah, see I don't think that's at all shocking.

discordance 2y ago

A competition like that is for the top 1% and very difficult problems. For most people and most problems, AI is very helpful and speeds up the process. As an example, I made a browser extension in 5 minutes this weekend using chatgpt by telling it what I wanted. I didn't know how to make a browser extension before that, so you could say AI surpassed this human quite easily.

krater23 2y ago
So AI is the next way to give people that don't know what they do new tools to generate and release shit where they don't know what it does to customers that don't know how much shit they got. I'm sure this will bring the mankind further...
- Ozzie_osman 2y ago
  People said similar stuff about the printing press.
jppope 2y ago
As expected, AI is great for stuff thats been done a million times. AI will slow you down for anything even slightly novel.
- 93po 2y ago
  for now
  - hatefulmoron 2y ago
    How long until the statement is false? You seem to have something in mind.
  - tovej 2y ago
    for all of time as long as machine learning = AI If it's not in the dataset the AI won't handle it correctly (unless it's trvial and a linear model is good enough, but then why even use AI)
- JamesBarney 2y ago
  99% of coding is doing something that's been done a million times and gluing it together in a novel way.
beeflet 2y ago
Yeah I guess it helps for applications where you otherwise need to read a ton of documentation to get started. For web stuff it helps me find useful APIs sometimes, but other times it wastes my time with deceptively wrong answers. The task of programming or engineering is generally one where you spend 80% of the time solving bugs in 20% of the code. LLMs are good for suggesting possible solutions based on existing things, but not for verifying solutions or doing any sort of real problem solving (they generally waste more time than they save because they give plausible but wrong answers). If you just want to reproduce some boilerplate or example code it is great though because documentation tends to lack good examples, and LLMs have a ton of real code stored in their latent space. Sometimes I even just use it as a thesaurus, if I have something on the tip of my tounge.

rsynnott 2y ago

> Shockingly Yeah, see I don't think that's at all shocking.