the prison to chatting with opus 6 to prison recidivism rate

do the nonsense words themselves reflect on anything internal here>? do is the idea that nonsense words are conveying meaning by being conventionally nonsense. why this nonsense words instead of the many other possible nonsense words

would entertain me for weeks until i figure out how to break it's safety restrictions(adult me figured it out in 10 mins)

I feel like if i was a six year old, figuring out how nano banana practically work would be the fun project

yeah i remember gpt4 being released and this puppy-like behavioral tendency "oh can i write some code, oh can i derive some equations for this" no I am too lazy i like my thought experiments to stay thought experiments stop creating work for me

I think Yud's framework was specifically so he could avoid hinging his predictions on a specific form of the technology(which is actually to his credit) and this is arguable(but things like deceptive behavior and reward hacking can be said to be "predictions" for him) adversarial robustness too

Hanson and Yud fought about that ... I wouldn't say the scenarios are not fleshed out? Yeah I mean the problem with yud is that his arguments are overly general and are hard to counter but that's not an argument they can't happen or represent small parts of the probability space

It feels unfair to say black swan events can not be tested at all! And that the evidential standards of normal events should be applied to events that are by their nature unprecedented? The point of the problem is that this is hard, hard to solve using our normal means of technological control

They know more but they still have shit judgment and taste

Get a body you nonembodied freaks but yeah I mean gpt4 was basically the point that smart people got mogged outside of very specific areas of expertise

I think I'm better at metaphysics than any of these models

I am actually an insurance agent

I'm pasting slop in response to this

This confuses me. It's just like intelligence why do people like being smart because other people are dumber? That's the least amazing part of being smart. Does being hot not work in the same way?