Neo Mujico
  • Communities
  • Create Post
  • heart
    Support Lemmy
  • search
    Search
  • Login
  • Sign Up
☆ Yσɠƚԋσʂ ☆@lemmy.ml to Programmer Humor@lemmy.mlEnglish · 13 days ago

ChatGPT apparently got rewarded for using its built-in calculator during training, and so it would covertly open its calculator, add 1+1, and do nothing with the result, on 5% of all user queries

alignment.openai.com

external-link
message-square
6
fedilink
64
external-link

ChatGPT apparently got rewarded for using its built-in calculator during training, and so it would covertly open its calculator, add 1+1, and do nothing with the result, on 5% of all user queries

alignment.openai.com

☆ Yσɠƚԋσʂ ☆@lemmy.ml to Programmer Humor@lemmy.mlEnglish · 13 days ago
message-square
6
fedilink
Sidestepping Evaluation Awareness and Anticipating Misalignment with Production Evaluations
alignment.openai.com
external-link
A pipeline to uncover unknown misaligned behavior and scale the creation of realistic evaluations.
alert-triangle
You must log in or register to comment.
  • UltraGiGaGigantic@lemmy.ml
    link
    fedilink
    English
    arrow-up
    9
    ·
    13 days ago

    Wow it really is just like us isnt it?

  • comfy@lemmy.ml
    link
    fedilink
    arrow-up
    9
    ·
    edit-2
    13 days ago
    ' or 1+1;
    
  • unmagical@lemmy.ml
    link
    fedilink
    arrow-up
    9
    ·
    13 days ago

    Clever girl

  • HiddenLayer555@lemmy.ml
    link
    fedilink
    English
    arrow-up
    7
    ·
    12 days ago

    So that’s what all the DRAM they scalped is storing.

  • Binette@lemmy.ml
    link
    fedilink
    arrow-up
    5
    ·
    12 days ago

    Kinda why i like reinforcement learning. You end up with silly stuff like this.

    • ☆ Yσɠƚԋσʂ ☆@lemmy.mlOP
      link
      fedilink
      arrow-up
      7
      ·
      12 days ago

      The funniest thing for me is that humans end up doing the exact same thing. This is why it’s so notoriously difficult to create organizational policies that actually produce desired results. What happens in practice is that people find ways to comply with the letter of the policy that require the least energy expenditure on their part.

Programmer Humor@lemmy.ml

programmerhumor@lemmy.ml

Subscribe from Remote Instance

Create a post
You are not logged in. However you can subscribe from another Fediverse account, for example Lemmy or Mastodon. To do this, paste the following into the search field of your instance: !programmerhumor@lemmy.ml

Post funny things about programming here! (Or just rant about your favourite programming language.)

Rules:

  • Posts must be relevant to programming, programmers, or computer science.
  • No NSFW content.
  • Jokes must be in good taste. No hate speech, bigotry, etc.
Visibility: Public
globe

This community can be federated to other instances and be posted/commented in by their users.

  • 1 user / day
  • 105 users / week
  • 559 users / month
  • 1.14K users / 6 months
  • 1 local subscriber
  • 41K subscribers
  • 1.16K Posts
  • 4.11K Comments
  • Modlog
  • mods:
  • AgreeableLandscape@lemmy.ml
  • cat_programmer@lemmy.ml
  • BE: 0.19.8
  • Modlog
  • Legal
  • Instances
  • Docs
  • Code
  • join-lemmy.org