body { -ms-overflow-style: scrollbar; overflow-y: scroll; overscroll-behavior-y: none; } .errorContainer { background-color: #FFF; color: #0F1419; max-width: 600px; margin: 0 auto; padding: 10%; font-family: Helvetica, sans-serif; font-size: 16px; } .errorButton { margin: 3em 0; } .errorButton a { background: #1DA1F2; border-radius: 2.5em; color: white; padding: 1em 2em; text-decoration: none; } .errorButton a:hover, .errorButton a:focus { background: rgb(26, 145, 218); } .errorFooter { color: #657786; font-size: 80%; line-height: 1.5; padding: 1em 0; } .errorFooter a, .errorFooter a:visited { color: #657786; text-decoration: none; padding-right: 1em; } .errorFooter a:hover, .errorFooter a:active { text-decoration: underline; } #placeholder, #react-root { display: none !important; } body { background-color: #FFF !important; }

JavaScript is not available.

We’ve detected that JavaScript is disabled in this browser. Please enable JavaScript or switch to a supported browser to continue using twitter.com. You can see a list of supported browsers in our Help Center.

Terms of Service Privacy Policy Cookie Policy Imprint Ads info © 2023 Twitter, Inc.

Conversation

Here we goooooo

19

38

257

separately, it makes me feel a little weird that they're training out the "toxic" stuff

Aligning Language Models to Follow Instructions

We’ve trained language models that are much better at following user intentions than GPT-3 while also making them more truthful and less toxic, using techniques developed through our alignment...

10:53 PM · Jan 29, 2022

Replying to

I agree. It implies a values judgement about what is toxic, which was implicitly made by the many inputs to the dataset, now controlled by OpenAI corporate bureaucracy. I don't trust any tech company to have a principled approach to this.

1

5

Replying to

Yup, I still miss her:

1

6

Replying to

it makes sense, though. as soon as they publish a new model tech j*urnalists try to make it say something offensive.

Replying to

Yeah I think we need to get to the point where the AI can understand why something is toxic and make independent value judgments about why that's bad.

Michael P. Frank

Replying to

Yeah… In my limited experience so far, Instruct feels more “soulless” than plain GPT-3. It may be more articulate, but is that a good trade? I’m not so sure…

1

Replying to

This is exactly why a true AI will never exist

Significant Fig

Replying to

Very useful models that teach us a lot about language and humans. Sad that they are made corporate PC

Show additional replies, including those that may contain offensive content

Show