Conversation
Post your reply
Just posting this here for no reason.
Quote
Rob Bensinger โน๏ธ
@robbensinger
It seems to me that we're at "it's time to panic about AI".
I gather that a lot of researchers are worried that we may be about to die, on a timescale of months, but this still isn't quite in the Overton window so we're still soft-pedaling.
(Where "panic", of course, means
Also, from the same post:
'We thought it was good-until we got our hands on an early internal snapshot of Claude Mythos Preview. We asked it to rewrite the whole thing, and with no further instructions, it produced a version that consumed 20x less memory and was 4x faster'
There have been a bunch of posts of 'blahblahblah' offhand TERRIFYING GRAPH 'blahblahblah' lately...
it's a sigmoid it's a sigmoid it's a sigmoid (rocking back and forth slowly)
Wow, Sep 7-9 looks like a step change.. Overall it seems to follow the trend seen in "Research acceleration: The view inside OpenAI" where data ends Aug 15
The size of the weekend drop might be a good proxy for the fraction of coding that still has nonzero human input. A large fraction of this work is being done by agents, not just in the sense of agents being prompted to do it, but also agents in more autonomous setups.
Also not the point but wild that the slope is so large that it persistently shows up as a Monday < Friday effect
Guess something weird happened around July/August, huh?
Quote
Lawrence Jones
@lawrjones
Enjoying digging into the data and seeing the massive impact AI adoption has had with our engineering team.
Kinda crazy to see how this suddenly jumps.
In a nice way, I don't think see another change like this coming soon. Think we're levelling off.
ya, its incredibly shocking and it seems like there's another step function change in capabilities in Sep internally available model?..
thank god they included the โengineers are no less valuableโ paragraph
link, for anyone else looking for it
