Public
- Public
- Network
- Groups
- Popular
- People

Important disclaimers: 1. The Al often makes stuff up that isn’t true—proofread everything carefully! 2. This is just a starting point. The writing will be coherent, but a bit bland and mediocre. 3. Sometimes the Al will stop generating text before the post is completely done. If that happens, just type +++ at the bottom and it will continue writing. Keep doing that until it's done.

Download link

Important disclaimers: 1. The Al often makes stuff up that isn’t true—proofread everything carefully! 2. This is just a starting point. The writing will be coherent, but a bit bland and mediocre. 3. Sometimes the Al will stop generating text before the post is completely done. If that happens, just type +++ at the bottom and it will continue writing. Keep doing that until it's done.
https://fediscience.org/system/media_attachments/files/109/961/735/959/366/392/original/0b8d0fd18ecb1ba0.png

Notices where this attachment appears

Carl T. Bergstrom (ct_bergstrom@fediscience.org)'s status on Saturday, 04-Mar-2023 06:51:03 UTC Carl T. Bergstrom

It's interesting how tech companies use language to obscure or minimize the fact that #LargeLanguageModels generate bullshit, plain and simple.
Here's Amazon's Lex, instructing me to "proofread" it's output because it makes stuff up.
We're not talking about typos here. Some much more involved than proofreading — namely, **fact-checking**, is required.

In conversation Saturday, 04-Mar-2023 06:51:03 UTC from fediscience.org permalink