Can I ask what kind of work, and what models? Because I don't, I'm a senior embedded engineer working with frontier models, and I haven't seen a major mistake in at least few months now, since earlier in the year.
replyIt will sometimes get things wrong because it made non-viable assumptions, but that's on me and prompt accuracy.
What I see is: it makes tests against desired features or bugs or mistakes, and iterates until it gets it correct. I steer as it goes.
(I hate this, by the way, zero job enjoyment anymore, but I don't know what else to do.)
Yes but we see errors in human output too.
replyA human can be fired, or put on a PIP, and is otherwise responsible for their output.
replyAn algorithm isn’t. I still dunno how they’re gonna square that circle.
How many humans do you need to keep around as scapegoats concurrently? I figure one with a CEO title should be enough.
replyThey will square it by pointing out, correctly, that AI is getting better while humans are not.
replyThe state of the art doesn't matter. Only the first couple of time derivatives matter.
Sure but humans aren’t being billed as “superintelligence”
replyExactly. It will now be your job to catch those errors. Or else.
replyCode review, qa, staging, product review, dog fooding… what do you mean “now”? It’s always been all of our jobs to catch all of our errors. We’re just including the AI’s errors now hoping that the overlap of everyone’s errors is sufficiently small.
reply