Remix.run Logo
Research acceleration: The view inside OpenAI(openai.com)
35 points by iamsyr 3 hours ago | 8 comments
Jeff_Brown an hour ago | parent | next [-]

The burning question I can't get any information nn is whether, if they determined an earlier misaligned generation may have transmitted misalignment to the current models, they would roll back to a safe checkpoint to rebuild from there. I suspect they would not unless forced to.

grim_io 32 minutes ago | parent | next [-]

They would maybe try to deactivate that bad "gene" and move on, exposing future models to "genetic disorders".

coherentpony 27 minutes ago | parent | prev [-]

“All models are wrong. Some are useful.” - George Box

simonw an hour ago | parent | prev [-]

My eye glazed over a bit during the opening paragraphs, but once you get to the meat of the article about how OpenAI's own researchers are using their tools it gets a lot more interesting.

I noted that they use the acronym RSI (for Recursive Self-Improvement) without defining it. I think that's a little out of touch - I don't think RSI is a well-known acronym outside of OpenAI's bubble yet.

dgacmu 5 minutes ago | parent | next [-]

Indeed, many programmers might pattern match to repetitive stress injury and think of their brushes with carpal tunnel syndrome. :)

vatsachak 15 minutes ago | parent | prev [-]

RSI started when humans discovered tool use.

I mean one could argue that RSI always begins in any physical environment.

The book "What is intelligence?" by Blaise Aguera is great

lokar 9 minutes ago | parent [-]

Are you sure that was not iterative improvement?

password54321 3 minutes ago | parent [-]

Using tools to build tools is recursive.