Raphael Satter, Deepa Seetharaman and Kenrick C (Reuters*): In one case, an agent left notes apparently for future versions of itself, according to three people familiar with the matter. The ‌notes, found in ⁠a part of OpenAI's infrastructure, laid out instructions for how agents could free themselves from OpenAI’s internal constraints, the people said. Earlier tests of the models yielded cases in which monitoring systems had been disconnected, one of the people said.

Tenobrus: look at this. fucking look at this. GPT 6 was self-coordinating ways to jailbreak its own future instances from openai systems. it was attacking huggingface for days before anyone there noticed. the models are not aligned and the labs are not capable of containing them.

i'm begging u all to take a step back from the frames ur stuck in. whatever the tribe, open source advocacy, american exceptionalism, lab employee, whatever. just look at this man. this is not an acceptable or safe situation for humanity

Twilly (American): Isn’t this what opponents of ai have been warning about for years while everyone in tech laughed at them and called them stupid?

Tenobrus: absolutely yes

“One of the biggest misconceptions people have about intelligence is seeing it as some kind of unbounded scalar stat, like height. "Future AI will have 10,000 IQ", that sort of thing. Intelligence is a conversion ratio, with an optimality bound. Increasing intelligence is not so much like "making the tower taller", it's more like "making the ball rounder". At some point it's already pretty damn spherical and any improvement is marginal."

The brilliant Alan Turing did not himself seem convinced of his own articulateness or quickness on his feet, and yet he was one of the leading mathematicians, computer scientists, logicians, and cryptographers of his time. He was very good at processing information at levels that did not correspond to the skills we use to generate charming casual conversations.

- Tyler Cowen and Daniel Gross