@RadishHarmers @kangminlee Genuine question: why do you troll in ways that you know will increase misunderstanding and backlash? I get that you find it funny but do you think it’s a good contribution to an already heated discourse?
English
shaky sage
305 posts
@kangminlee Dear everyone: Do you want to take this at face value or do you want to respect your own intelligence? You are letting a grifter (who does not respect your intelligence one bit, who thinks you are easy marks for an "I'm the MAGA Asian!" novelty influencer career) mislead you.
English
Hindus: "Stop calling us invaders, that's racist"
Also Hindus:
English
Yes we should train more American doctors but until then, there is a huge labor shortage in the healthcare industry which costs American patients health and dollars.
These things are not zero sum - having more skilled labor grows our economy, improves services for Americans, and creates more jobs.
English
@c0rtic4l @mattyglesias @kausmickey Journalists? We need to import journalists? No.
We know how to train American doctors and we can make it cheaper to do so. And yes the government should take our own people's side, not mercenaries who will work for less because its better than living in India.
English
@jo64yr @mattyglesias @kausmickey The idea that we should have worse quality, less quantity, and more expensive doctors or journalists or contractors just to protect American jobs in the short term is both very bad for the Americans who rely on these services and is long term bad for our economy.
English
@jo64yr @mattyglesias @kausmickey If someone else can do the same job equally well for less pay, then yes they should do it! In the long term, this pushes developed societies toward upskilling, which becomes a competitive advantage.
English
English
@mattyglesias @kausmickey Yes I know who you are it was a bit tongue in cheeky. But we're graduating people from college and importing competition for jobs. Even colleges want to hire H1Bs! Why can't they hire their own graduates?
English
Having a higher supply of doctors is a good thing for a society. Having a higher supply of construction workers is also a very good thing! Very bad for our society that it became politically toxic in the last 10 years to say this out loud.
English
A very common view on immigration among “smart” people is that “high skill = good, low skill = bad”. Unfortunately:
(1) this is not the view of the Trump admin which has reduced both kinds, and (2) most econ research shows that both kinds are very good!!!!
William B. Fuckley@opinonhaver
A million people have rightly dunked on this guy, & I don’t care, I’m going to do it too, bc these people should have their catastrophic and massively consequential failures in judgement shoved in their faces forever. Sometimes a dog doesn’t learn unless you rub it’s face in it.
English
What does it even mean to slow down? AI is advancing so fast because there are a lot of fruitful research directions that are addressable by the tools we currently have. We could ban training bigger models or building new datacenters but does this stop AI progress? Doubtful imo. And even if so, is there a way to do this without equally slowing down AI safety research? Even more doubtful
Elizabeth Barnes@BethMayBarnes
(4) IMO, any “reasonable” civilization would clearly be taking things much more slowly and carefully with AI. The benefits of getting upsides of advanced AI a little faster are small compared to the risks of getting it irrecoverably wrong, and we could lower these risks by going slower
English
@JaneAFlegal @KelseyTuoc Ok but now you’re saying that allowing philanthropy at this scale is bad policy, which is a very different claim from it bypassing democracy.
English
@KelseyTuoc “Legal” and “good policy” are not the same thing? Using democratic legitimacy to defend a tax structure that lets billionaires accumulate the influence that shapes democratic outcomes is strange.
English
Oof. These are bad reasons to bypass democracy. Philanthropy at this scale is governance without legitimacy or accountability. It determines agendas, resources, legibility of problems, but w no elections, no oversight. And I worry it won’t complement government, but undermine it.
Megan McArdle@asymmetricinfo
I would say yes, for the same reason I prefer markets to central planning. Yes, rich person philanthropy will be idiosyncratic and many will disagree with individual priorities but it avoids the information problems, bureaucratic sclerosis and incentive issues that plague democratic policymaking.
English
I just hope communities use this moment of leverage to make real demands like making AI companies pay for upgrading the power grid, or distributing data center tax revenue for useful social services, rather than just slowing data center construction or AI progress.
English
imo the vast majority of opposition to data centers is based on misinformation and general anti-tech sentiment. But I do think people are justified in using their leverage, in this one moment while they have it, to demand fair recompense.
English
Every conversation about AI consciousness must start and end with:
1. we have no idea what consciousness is
2. we have no idea how to go about studying consciousness
English
Not going to be convinced that we’ve reached AGI until a model can do a git rebase without blowing up the codebase
English
Of course everyone thinks it’s wrong. The real question is how much of a cost premium are people willing to pay for cage free pigs or in-ovo sexed eggs.
Bryan Walsh@bryanrwalsh
Americans think grinding up male chicks and keeping pigs in tiny crates is wrong. They also think eating meat is as morally neutral as using IVF. Both of those things are true simultaneously. @KennyTorrella for @voxdotcom on the "meat paradox" — and why it's so hard to crack
English
shaky sage retweetledi
Listening to Tristan Harris talk about AI is a reminder that you can’t just port over all your priors about social media to AI.
English
shaky sage retweetledi
@khoomeik To be clear though, Sutton’s critique of LLMs comes from the exact opposite place as Gary Marcus. GM wants to put more knowledge and symbolic reps in models. Sutton dislikes language training bc it’s a way to build in knowledge rather than let models learn from the world.
English
i need you to understand that for sf ai bros this is like when kanye turned nazi
Richard Sutton@RichardSSutton
@GaryMarcus @ylecun @demishassabis You were never alone, Gary, though you were the first to bite the bullet, to fight the good fight, and to make the argument well, again and again, for the limitations of LLMs. I salute you for this good service!
English
Ads are fucking terrible and have ruined most websites - 100% agree. question is would you still use most sites if you had to pay for them? IMO, email: yes, maps: yes, social media: probably not.
is that a better world? absolutely yes.
rat king 🐀@MikeIsaac
people are flocking to AI apps because the open internet was killed by the modern digital advertising business model most web pages are a garbage fire of terrible ads by pure necessity of survival because all the $$$ goes to Goog, Meta and Amzn
English
Key distinction between next-token prediction in language & vision: text is human-generated and directly reflects cognition. Video/images are capturing the real world. Learning directly from observing the real world is harder than learning to emulate an already-intelligent being
Sergey Levine@svlevine
I always found it puzzling how language models learn so much from next-token prediction, while video models learn so little from next frame prediction. Maybe it's because LLMs are actually brain scanners in disguise. Idle musings in my new blog post: sergeylevine.substack.com/p/language-mod…
English