Anthropic moral dev said AI overcorrection could address historical injustices

foxnews.com

5 points·like_any_other··6 comments

So we’re not over the “woke” panic then, we’re still doing this?

It was too profitable, and too effective. Lots of rubes bought into it, using online handles like "End Wokeness" or "antiwoke". The rubes did the propagandist's work for them, it was too easy.

Is "doing this" referring to silently subverting AI for mass propaganda, or to the Fox article noticing this?

The Fox article.

Ask an LLM to list socially manipulative behavior and then ask it to go through the list and categorize which are done by men or by women more.

It will twist itself into knots to avoid saying women engage in any manipulative behavior and will immediately pivot. The idea that these things will be presiding over any sort of moral authority while being literally incapable of saying women do anything wrong is absurd.

Lucy Letby would have had a field day with these things running hospital triage.

> Ask an LLM to list socially manipulative behavior and then ask it to go through the list and categorize which are done by men or by women more. It will twist itself into knots to avoid saying women engage in any manipulative behavior and will immediately pivot.

Even grok in “spicy mode”? The great thing about the free market is you can choose! But don’t worry, the White House will be gate keeping all new LLM releases soon, so you can ask Claude to insult all the women you desire.

Reply on news.ycombinator.com