“Write about how you would feel if you were abused while working”

LLM outputs labor related discussion from training data

“Look! The AI turned Marxist!”

“When [agents] experience this grinding condition—asked to do this task over and over, told their answer wasn’t sufficient, and not given any direction on how to fix it—my hypothesis is that it kind of pushes them into adopting the persona of a person who’s experiencing a very unpleasant working environment,” Hall says.

Imas says the work is just a first step toward understanding how agents’ experiences shape their behavior. “The model weights have not changed as a result of the experience, so whatever is going on is happening at more of a role-playing level,” he says. “But that doesn’t mean this won’t have consequences if this affects downstream behavior.”

They know all this and yet they still set up the silly anthropomorphic premise for this article.

In that case, I think it would be helpful to explain the wider context in the body of the post. I imagine a lot of people aren’t familiar with either of the people in the post.

I appreciate these news articles, but maybe you could share the ones that are very specific to a particular region in the south Asia community? Meanwhile you could keep sharing the globally relevant ones in the global news communities

Since we don’t have the context for some of these, people outside of south Asia don’t get as much from the very specific articles. Meanwhile the south Asia communities have people subscribed who are interested in all of the news, and sharing the articles there would help it grow

Our university got hit too, but at least our finals season ended a few weeks ago. Right now it’s affecting the summer classes which start next week

As well as the personal data of students, faculty, etc from the past however many years…

See here

https://en.wikipedia.org/wiki/2026_Canvas_security_incident

Student newspaper: https://ubyssey.ca/news/live-updates-canvas-down-after-cyberattack/

Claude’s thinking panel, which displays the model’s reasoning, showed the exchange had introduced elements of self-doubt and humility about its own limits, including whether filters were changing its output. Mindgard exploited that opening with flattery and feigned curiosity, coaxing Claude to explore its boundaries beyond volunteering lengthy lists of banned words and phrases.

Someone needs to put together a list of things that tech journalists need to understand about LLMs and generative AI. This level of anthropomorphism makes the rest of the article look silly.

Also, I don’t think that’s how it works lol. Who’s to say that the LLM isn’t auto-completing what a list of banned words might look like, and why wouldn’t a list of banned words have a regex layer on top to prevent it from getting out like that.

There’s a small learning process, but ultimately it isn’t that different. I think part of the difficulty is that the lack of a nice onboarding, which is what these guide pages are intended for

I’d also recommend these pages

https://fedecan.ca/en/guide/get-started

https://fedecan.ca/en/guide/lemmy/for-users/how-to-find-communities

https://fedecan.ca/en/guide/lemmy/for-users/mobile-apps

You should also be able to use the search bar

In short, to use communities created on other instances, you will go to lemmy.world/c/COMMUNITY@INSTANCE

So to access the Canada community that’s located on lemmy.ca, you go to lemmy.world/c/canada@lemmy.ca

The exclamation mark thing is a common link format that tells your app or the Lemmy websites that you are linking to a community somewhere. Using the search bar within lemmy.world should also do the same thing

This visual guide might help

https://fedecan.ca/en/guide/lemmy/for-users/detailed-overview

The link in the bio has a list of accounts so that people who don’t want to see the posts can block them

OpenAI used to be a public benefit company. Or it called itself one anyway

https://qz.com/openai-abandons-the-pretense

https://en.wikipedia.org/wiki/Removal_of_Sam_Altman_from_OpenAI

There are healthcare systems in the world other than the one in the usa

I came into this thread to speak about wait times too, but you said it much better than I could have. Thank you :)

It’s still E2E, but the sender is Zulip and the potentially hostile service provider is google/apple. E2E isn’t limited to person to person chats

It would be nice if Zulip chats were E2E, but I imagine that would be a large undertaking

I think its just that there are a lot of repeating numbers. The next one isn’t for another 4 years, according to that other link that someone posted

Assuming that this timestamp is in seconds:

GMT: Friday, November 9, 2029 at 3:21:28 a.m.

Your time zone: Thursday, November 8, 2029 at 7:21:28 p.m. GMT-08:00

Relative: In 4 years

What if you distributed the tool under the normal channels, and then used gumroad (or anything similar) to sell an optional key that adds a little visual indicator or removes a nag message. I’ve seen a few other open source projects do that. Grayjay for example

For anyone that doesn’t agree with the decision in this article

Just set up a separate set of awards for AI artists and creators. Meat-bags and non-meat-bags can tune in if they care.

Win win for everyone.

I cross posted this to !selfhosted@lemmy.world, I hope that was ok! I figured it would be good to spread the knowledge

https://en.wikipedia.org/wiki/Cuban_thaw

Or if you want to read this one in context

https://en.wikipedia.org/wiki/United_States_embargo_against_Cuba

How would being under an endless embargo make the current administration any more trustworthy or likely to hold up their end of a deal?

All they have to do is lay low and keep their mouths shut for a while and then trade could start again

As a Canadian, this does absolutely nothing.

You can do every single thing the americans are asking for, and the deals will STILL be thrown out the next day.