While you are correct that there likely is no intention and certainly no self-awareness behind the scheming, the researchers even explicitly list the option that the AI is roleplaying as an evil AI, simply based on its training data, when discussing the limitations of their research, it still seems a bit concerning.
The research shows that given a misalignment between the initial prompt and subsequent data modern LLMs can and will 'scheme' to ensure their given long-term goal.
It is no sapient thing, but a dumb machine with the capability to decive its users, and externalise this as shown in its chain of thought, when there are goal misalignments seems dangerous enough. Not at the current state of the art but potentially in a decade or two.
Das mit Linux ist vor allem so unendlich dämlich. Die Streaming Betreiber, schieben Internetpiraterie als Grund vor, warum sie die max. Auflösung auf Linux reduzieren, was Leute exakt dahin treibt. Ich hätte warsch. mindestens ein Abo, wenn ich halt HD Serien bekommen würde.
And we have better night vision than most the animals that have better day-vision than us. Humans are like the Leatherman of animals. Universally capable of doing most things but not as good as something specialized for that task. Plus of course capable of coming up with ways to cheat
Ich bin jetzt seit 1 Monat in nem Betrieb, der was macht, was ich bisher nur theoretisch kannte. Noch laufe ich mit nem super Kollegen mit l, der mir auch echt alles gut erklärt und zeigt, aber es hieß, dass ich in ein paar Wochen das dann alleine machen soll. Momentan schwanke ich zwischen existenzieller Panik und absoluter Selbstsicherheit.
Yeah. I got the theory, but I have almost no practical experience. In total something like 4 weeks of field trips. But nice to see that I still get the basics.
Looks like some kind of intrusion of magma (the pale rock) into the darker rock. you can see how all the veins seem to originate from the pale rock, also the broken-of dark part in the pale rock seems to indicate it, could be part of the original rock, that surrounded the magma, before breaking of.
Some actual geologist might want to give their opinion though. I only had like two years of geology at university before shifting my studies toward crystallography and crystal growth.
You could and it might work, or it might not. Depends on a lot of factors. Thing is, you are still engaging the fey in word games and try to outsmart them in their own games. Being blunt and telling them No, giving clear and unambiguous answers offers less attack ground for them.
While you are correct that there likely is no intention and certainly no self-awareness behind the scheming, the researchers even explicitly list the option that the AI is roleplaying as an evil AI, simply based on its training data, when discussing the limitations of their research, it still seems a bit concerning. The research shows that given a misalignment between the initial prompt and subsequent data modern LLMs can and will 'scheme' to ensure their given long-term goal. It is no sapient thing, but a dumb machine with the capability to decive its users, and externalise this as shown in its chain of thought, when there are goal misalignments seems dangerous enough. Not at the current state of the art but potentially in a decade or two.