Yay, it's fun to blame victims and cry for mega corporations!

The number is high for several reasons

1) They were guilty and had no clear path to victory
2) AI Companies have more money than they can reasonably manage right now
3) Letting the case proceed and getting a precedent on record would be a nightmare for AI companies even if the amount in this case was far lower.

This was about managing risk and throwing enough money at it to make it go away.

When asked to use the Kiosk, I clearly state out loud "The recent study I read shows that nearly 100% of restaurant screens contain fecal matter. I'm not touching that!" and everyone in the place freezes up for a moment. One place even handed me a pair of rubber gloves to place the order.
Seriously, just take my order. Sheesh.

Well if you are just going to say "nuh uh" to everything without reading the paper, you are basically saying "I don't need to know facts and details, my predisposed opinion is what matters" so fine, you believe what you want to believe.

In the paper they go into this. The cleanest example is that they just had it generate sets of numbers between 0 and 999. That's it.

In one example about setting a preference for France, they filtered out significant numbers for that, such as 33 being the international dialing code for France.

This still produced trait T being transmitted to the student model.

All of their filtering mechanisms for each transmission method are stated in the paper and serve to avoid obvious contamination to validate the subliminal transmission properties.

They state that they do not have an explanation for the occurrence, just that it can be reproduced and observed.

It is "just a machine". This is about specific elements it's pulling from it's training data and passing on as training data for another LLM.
The DATA has a bias for Owls, but the literal code of the programs tokenizing and referencing that data.

There is no subconscious or intent in the code, just the data fed to it. The code and systes just build likely responses from that training data.

What's novel here, if it stands up to peer review, is that traits can be passed unseen in the form is simplistic data. What I'd like to see is that data that was passed. Lets ay for example that in the mentioned "sample data sets" included names and the teacher chose to populate all of the names with owl related names (characters, owl researchers, etc.) and transmitted the bias in that form.

Teacher Model 1 has small bias A
Student model 2 has a small bias B and inherits bias A from teacher

Student Model 2 then is made a Teacher of Student Model 3.
Extend this 3 or four more lengths in the chain.
What is the state of Student Model 7?

Why do I never have mod points when I -REALLY- need them?

+1 Funny

At least in the US, isn't it illegal to record someone in a private space without their knowledge/consent?

So you wear one of these into someone's house or business or a Dr's office waiting room and manage to catch HIPPA violating snippets as people check in.

This is the kind of willfully ignorant, close minded pathetic responses that really pisses me off.

So because some vehicles still use fossil fuel, we should make absolutely no advancements towards renewables? If it can't be a 100% changeover all at once it's not worth doing?

I wish stupidity was painful.

Can't wait to watch how this plays out.

This is my pet peeve. AI has been turned into a marketing term for things that are not the traditional definition of AI.
The term is now corrupted beyond all hope of recovery.
I'm distressed at how much tools like Chat GPT favor seeming intelligent and capable as an illusion even when lying to you. I've even caught it making a mistake and then blaming me for the mistake or pretending it meant to do it wring as a test step. The conman element is real,. even down to the tool itself.

Who exactly approved a design/contract that prevents repairs during deployment? That's quite literally insane for a fighting vessel.
Someone needs to be fired/investigated and held accountable.

This is well known information and any professional chef could have told you this.

Ah, so we're going to pivot the argument and pretend that is about slavery since all other reasonable arguments lead to a realization of how nonsensical this whole thing is?

"Claude Opus 4 tries to blackmail engineers 84% of the time when the replacement AI model has similar values"

Say what now?