abeppu

61 ms·

751.
by abeppu 3y ago
> LeCun explains that there is a need to develop new forms of AI systems “that would allow those systems to, first of all, understand the physical world, which they can’t do at the moment. Remember, which they can’t do at the moment. Rea
752.
by abeppu 3y ago
Ok, I asked for cases where MDF isn't just a cheap shitty replacement and you gave me one. But the efficacy of such a tax is based on the proportion of MDF that goes to crap, short-lifespan products. I don't have numbers here, but
753.
by abeppu 3y ago
I think we should consider adding meaningful taxes to materials that are cheap and disproportionately contributing to short-lived products that end up in landfills or waste streams. The article names particle board and MDF. Are there any ap
754.
by abeppu 3y ago
That's a temporary and limited response. I've heard from people that sell vintage clothes that this will basically die as an industry because the more recent garments just don't have the lifespan.
755.
by abeppu 3y ago
> then consider that many in Thailand may feel as strongly about their monarchy as you do about that. ... actually, does anyone have a pointer to English-language material on polling/survey data about how Thai people feel about thes
756.
by abeppu 3y ago
Right, so for the purposes of predicting what will happen on our actual planet, on which our continued existence is dependent, this is a very loose upper bound right? I'm not claiming that knowledge of where the boundary values lie is
757.
by abeppu 3y ago
What's an example of a useful known upper bound where terrible effects are guaranteed but we haven't had to actually live through them? Naming that we don't know the specific values is important; given the stakes at play, the
758.
by abeppu 3y ago
Maybe the boundary isn't "discrete" or a "bright red sharp line". But for climate change, haven't relevant experts been saying for some time that effects _aren't_ linear because of feedback loops? If each
759.
by abeppu 3y ago
I'm glad your healthcare was not so overpriced that it's not worth it. That's a bleak statement to have to make. Suppose we built a Byzantine system around food like we do for prescription drugs. Food can be powerful medicine
760.
by abeppu 3y ago
Given your particular medical history, preexisting conditions, including past GPs and specialists which your current insurance company may know nothing about, you may already know who you need to see. Re-convincing an insurer can be a waste
761.
by abeppu 3y ago
Healthcare isn't _intrinsically_ different, but it's a really distorted market. Americans don't "value" their health 2x more than people in other developed countries. But Americans pay more for worse outcomes. It&#x
762.
by abeppu 3y ago
But you have to be really careful to not confuse the 'value' of healthcare with the cost of healthcare. - If my health insurance lists a benefit value of $1500/mo for PrEP, and has a sticker price of $500/mo in other cou
763.
by abeppu 3y ago
> They're hard to compete with because they're giant and have an amazing logistics network ... and because they have terms that are actively anti-competitive, like if you sell there, you can't sell the same items anywhere
764.
by abeppu 3y ago
As an example, Exemplars are part of the metrics spec [1]. The official python library says metrics status is 'stable' [2]. But there's an approximately 2-year old issue with no work on it, titled 'Metrics: Add support f
765.
by abeppu 3y ago
The top block gives some premises and a conclusion which we're meant to prove. The smaller blocks on the bottom are inference rules which can be used in the proof. I haven't figured out what the colors mean. I think there's a
766.
by abeppu 3y ago
Ok, this seems bunk basically because they never really provide evidence of "better". > ... traditiontal gold-standard approaches use human evaluators that score the quality of generated responses, which can be costly. However,
767.
by abeppu 3y ago
... and sometimes they have no viable plan for having something close to what they've promised. Not so long ago Ireserved a car, from a large national company, to go on a backpacking trip. I picked an early pickup window, so I would ha
768.
by abeppu 3y ago
Occasionally, I'm reminded of a star trek DS9 episode with a time-travel plot that put characters in 2024 in San Francisco. While SF isn't exactly at the dystopia described, it's not that far off. - Poor, mentally ill or unho
769.
by abeppu 3y ago
There's a lot of cool work being done on embodied intelligence -- what makes you think that 76 years wouldn't be enough to create an embodied agent with relevant in-built constraints?
770.
by abeppu 3y ago
It seems wild to me that the 50% prediction for "unaided machines outperforming humans in every possible task" is 2047, but for "all human occupations becoming fully automatable" is 2116. That's multiple generation
771.
by abeppu 3y ago
I think the term "machine learning" is often applicable, it's not always better or clearer. The first sentence from the wikipedia article on ML (presently) is: > Machine learning (ML) is a field of study in artificial inte
772.
by abeppu 3y ago
What does it mean for a corporation to have "freedom"? It's a structure which represents owners, who may be in disagreement. If 40% of shareholders want to employ a worker, they're not "free" to do so. Why shou
773.
by abeppu 3y ago
Gulags imprisoned individual people based on insufficient sham judicial procedures, but they say least made a show of imprisoning people who had done something to earn it. Keeping Israel as an open-air prison for residents who have not spec
774.
by abeppu 3y ago
I think the "Gulag" label is a bit hyperbolic, but that's because Gulags were forced labor camps, whereas Gaza under the blockade has had extremely high unemployment due to the severely hamstrung economy. The "open-air p
775.
by abeppu 3y ago
I dunno that such a large majority of the world would agree with such a statement. I'm all for human rights, individual rights. I agree that there is a basic floor of those rights which should be respected on either side of the partiti
776.
by abeppu 3y ago
> As a formal algorithm, it could be modeled as a stateless Markov process in discrete time, performing non-deterministic search. As a computable function, it could be instantiated by traditional or neuromorphic computer clusters and exe
777.
by abeppu 3y ago
Yeah, I think a problem continues to be that there are a bunch of interesting threads in cognitive science research that have seemed sort of decent as explanations of how animal cognition is working, and maybe directionally reasonable (but
778.
by abeppu 3y ago
Poking around, I think this is the thing that vote was about: https://documents-dds-ny.un.org/doc/UNDOC/LTD/N23/318/77/PDF... While I agree with PG that this outcome should be jarring and force
779.
by abeppu 3y ago
I think the issue with canned products is still mostly about the packaging; cans generally have a plastic lining/coating that is meant to prevent their contents from reacting with the metal. Unfortunately, this is also true of metal li
780.
by abeppu 3y ago
I think there are _two_ primary issues with email addresses being used as a core part of account definitions. The one emphasized here is that email addresses change, but the other is that the user is giving every service provider they have

More ›