Notices by Tito Swineflu (tito_swineflu@sfba.social)

@inthehands that quote struck me as very out of character too. How can they steal something they deserve?

I have made a new superintelligence that outperforms all of the frontier models!

len([x for x in 'strawberry' if x == 'r'])

@kingrat It'd also be nice if the generic upgrade/update just worked. I don't have a modified mysql, it's just the basic install, but it breaks every time. It hangs on restarting mysql, and I just don't want to deal with trying to untangle that giant ball of yarn.

@kingrat thanks.

@kingrat How do I do that?
Every time this happens, I'm like, "I should figure out how to stop this from happening," but then I have to spend an hour or two remembering how to rebuild mysql with no root password and never get around to it.

God damn it. Every time ubuntu's auto-update touches mysql, it completely nukes it and I have to back out, delete it and re-install. It's infuriating.

@benroyce @nixCraft discovering that once you offload cognitive ability, you can't just get it back.

@kingrat @edgeworth a big electromagnet that stacks cars on top of the bus for later auction?

Man, the #49 is efficient!

@edgeworth now if only they'd add a snowplow attachment fit mission past cesar chavez to clear the cars...

@kingrat I like cod. The cod/potato/oil scramble thing is blech. I don't eat octopus, and I wish it wasn't on the menu.

I am concerned that the cuisine of Lisbon might be a little too centered on cod.

@kingrat this seems right. I just feel like GOOD training data for a general purpose LLM or even a coding LLM would be really expensive and a cost that wouldn't magically go away between models. You can use synthetic data, but at the end of the day, a human has to tell the LLM trainee that it's correct or not. So, like self driving cars, humans are always going to be in the loop, and that will also remain expensive, even if they're all in lagos. And, if they're all in Lagos, the training data will be missing oversight from humans who must be in the US or other expensive countries.

Someone explain this to me: getting and tagging training data for AI must be horrifically expensive, right? thousands of people sorting through the data and the outputs? Is it just cheap compared to the compute costs? No-one ever seems to talk about it when talking about AI economics. Having done some scraping and tagging in 2010, I imagine the scale of work doing that for AI must be huge. What am I missing?

@kingrat I think that's low. I lived on isis for years, and SOMA is bad. Really bad.

Any word on a pedestrian killed by a car on Brannan St. Today?

Holy shit. I only use chrome when I have to for work, and these assholes...

https://www.thatprivacyguy.com/blog/chrome-silent-nano-install/

@mhoye I love that the first line in "What needs to change" isn't, "We should not let non-deterministic programs have free range across our systems"

Hey @historyofpunkrock , remember I said my band was covering the Avengers? Here's a little bit of it.

Taiwan high speed rail business class. Pretty sweet!