HN user
out of curiosity - what scheme do you suggest? I've always been of the mind that 'one thing to remember and secure, but secure it well' was the best option - 2factor and a 15+character passphrase meaning that nearly everything else gets it's own discretized blast radius.
Always open to better security, though.
I'm with you on the uninformed / confident vector in general, but you do seem fairly level-headed so I still want your take on the question: what do we do about it? How do we approach the problem of "most of our problems stem from an intractable information asymmetry, and this could be existential"?
Real question; what are we supposed to do about that information delay when it directly enables corruption and usury? This is an ongoing issue with historical precedent; the repeal of glass-steagall, MKULTRA and COINTELPRO, Iran-contra, watergate, the list is indefinably long.
All of these all surfed in on that very temporal ambiguity, and the fact that we have zero recourse in a plurality of cases - a situation that has eroded over time, not gotten better, and could feasibly be credited with a large part of the palpable social decay that real people are suffering from every day right now.
So what do we do about it? "Indulging in plausible conspiracy theories" could also be read here as "trying to get out ahead of this imminent yet undclear threat"
"Assessing effects of rent control on wealth is hard, so we threw most of that out and just used housing prices. Also, we're fairly sure that our results are good because there was nothing special happening in St. Paul in 2021 that we could figure. See this scatterplot that looks like inflation, note that the statistics we put just below it also don't look interesting but we've tied them to how rent controls are at the same time socialist and also evil and capitalist. We are objective parties to this because we say we are, please ignore the TLD we're serving this on."
You can almost hear yakitty sax playing in the background. I bet if you met the researchers you could honk their nose.
Right - it doesn't hold up to scrutiny. For one, "not a citizen" is a pretty hard bar to assess online. For another, "citizen" isn't very meaningful here. Many national security incidents have featured a citizen at the core - and it's a really fuzzy indicator of "potentially hostile" and especially "for what reason".
I guess I'm possibly giving them too much credit, but if the people who sent the letter have their head screwed on straight, "protecting national security by disallowing specifically non-citizens from using it" can really only be read as a smokescreen, or at very best a small part of the actual picture.
Bit of a doozie though, that one.
I recall getting really excited over hinton's FF foray, right before he bailed on AI as a societal direction (which, if anyone ever had the right, I suppose he does). If one squints, one can see a backprop-free base being much easier to train on geographically distributed and heterogenous hardware.
Genuinely curious - who do you think the targeted people are and how would this keep them out?
Perhaps the US administration is gambling that US citizens on their own provide enough of a training data and revenue flywheel for them to keep their AI development edge.
There is no way to enforce access of one and not the other, not with the state of tech in the US (and most countries without a great firewall). Bypassing such controls is as easy as a pilfered credit card (or some other american-looking payment method) and a vpn - both trivial to come by.
Perhaps a little tinfoil hat, but I don't think there's a legitimate concern here to address. An empowered populace is antithetical to the current political paradigm, which is what I suspect the actual grievance to be.
And before either 'aisle' piles on - I'm pretty sure the concern is bipartisan.
Dumb astroturf grok bot is dumb. Don't feed the trolls.
It's not a training problem, it's an incentive problem. You put these guys in a structure that requires them to justify their jobs at minimal cost of effort and then ooh ack surprise when they don't take the proper care to ensure that they're not stepping on innocent people in the pursuit of a healthy career.
Couple that with overburdening them with petty nonsense all the time and training them in military equipment and tactics and like it doesn't matter what tools you give them, those tools will be abused at convenience.
The issue is structural, not technical, but power tools = more damage per capita.
I think I see where we're coming at this from different angles.
If you're going by the numbers and the 'strength of the state' - then yeah, we're doing great. However, neither I nor anyone I know happens to be a part of the class that's holding the baton with all that stuff.
My concern is specifically about how well the citizens are doing, in aggregate, taking into account whether they have democratic control over the reins of governance, whether they are afforded the opportunity to be meaningful contributors to the greater good, education, freedom, enfranchisement.
GDP / capita is meaningless if most of the 'capita' never sees a dime. TBH the structures and institutions that make up a country are just a bunch of bureaucratic role playing from where I stand - they're meaningless without the 'we' of 'we the people'.
all fair points - but what strengths? We've proven ourselves incapable of the most basic social goods for decades now. All the metrics that you might point to as "hey the US is doing fine" (GDP, deficit, sector growth) are concerned specifically with how the state is doing and desperately unconcerned with it's citizens, which I think is a principal issue here.
Isn't flock's whole thing that they extract information from the pictures they have?
Like, say I have an interview in your office and you step out for coffee. I take a picture of the applicant list on your desk. That doesn't make the list of applicants "my data".
Both are good sources of energy. If you're going to make the argument that "nuclear is unsafe so we shouldn't do it" though, it's relevant to keep in mind that since we've had nuclear power, dam failures have outpaced nuclear by many times in terms of deaths / TwH (1).
Edit to add: Before anyone jumps on for this it's important to note that without the Banquiao disaster the rates are about the same. Still means "nuclear is unsafe" is kind of a red herring.
I'm not going to go to bat for Mao(1), but I think you're underplaying the body count that capitalist countries have had - this is kind of easy to do because a lot of the damage that we do is obfuscated behind proxies. Besides the obvious and direct war crimes like Vietnam, Iran, Iraq, and now I guess Iran again, there's the second order stuff like Israel's Bad Neighbor Syndrome (which we have enabled financially for basically the duration), Pinochet who we put in charge, heck - pick any country south of the border and we've done some damage there at least once. Then there's the spiderweb of damage that flows out to the global south continuously through NAFTA and similar foreign policy. I suppose the principal difference is that we externalize a lot of our violence (and somehow are shocked when it comes back to bite us that we trained Osama Bin Laden).
Nobody's asking for Maoist China, I think mostly we're clamoring for something closer to Norway. I'm sure plenty of people would be happy to settle for UK-style socialized services but even those folks get lambasted for being "too far left" too so whaddyagonnado.
1 - I think he and 'bolshevism' are a bit of a strawman here anyway, as I've not heard a ton of pro-Mao people but a TON of people who identify as leftists - they are not the same thing
> trust in the ideas and people who control it,
This right here is the crux of the issue. I don't even trust my own computer without fairly deep introspective tools, and what we're given for 'leadership' is 'this totally outdated and opaque system of voting for corporate shill A or corporate shill B is totally trustworthy! You obviously cannot think that you could get by without some asshat running your whole society so be thankful'.
Direct democracy, liquid democracy - whatever you pick that removes the middle man will be a marked improvement from day 1. We do not need these people deciding what's best for us. I'm not sure we ever did.
guidance and alignment are usually handled by RLHF, which actually rewires the weights such that it becomes near-impossible for the model to have certain kinds of 'thoughts'. This is baked in such that it's not something you can just extract or turn off.
"Yeah so it says on your chart here..."
This is a system stretching over millennia
not quite. 'Slavery' has been around that long. 'Chattel Slavery' started in the 1600s and peaked in the 1800s. So like, half a millenia.
no.
It really comes down to granularity at the end, and whether you attempt to look as closely as possible or you accept a certain lack of fidelity because it makes the abstraction work for you.
In this case, I frequently hear people talk about how "the greeks and romans had slaves! and they were white! See, it's fine!" but that fails to take into account that there's a gigantic difference between slavery-as-a-legal-status like they had (entered into by contract or as legal punishment, exit conditions, no real social meaning), and chattel slavery based on race (the 'fuck you got mine' of ethos). I think the idea is that if you squint real, real hard; you can make it look like "not being racist" and "human rights" are somehow newfangled, 'woke' ideals, which is the kind of hilariously wrong misunderstanding we once saw embodied by cletus the slackjawed yokel.
I can call my ma from up here. Hey, ma! Get off the dang roof!
Slavery as we talk about it has been around since roughly the 1600s, and even then didn't peak until the 1800s. Everything prior to that was a totally different beast. and a quick sidebar - wth is supposed to be wrong with being alert to your surroundings? Do we really value being asleep that much?
I think you've misunderstood the term 'petrodollar'. Petrodollars are the American currency in circulation abroad because we bought other people's oil (principally Saudi), not exported our own.
The 'export' that made the US powerful was finance and political manipulation - toppling socialist / populist leaders to install puppets and controlling economies by manipulating trade.
I think your original point kind of stands, though - we are seeing a decline and independence from our supply chain is going to be a deciding factor in 'who's the next top dog', but I think the decline is going to be a lot uglier than a simple "they have it now and we don't" - it's going to be all the thrashing about that an aggressive international power does when the grift no longer works.
Fascinating. I've read 5 posts about this and they're all either "anthropic is dropping their ethics" or "anthropic is fighting the facists" - and whether due to echo chamber or other perhaps more nefarious dealings (some of which I cannot posit due to forum rules) the posts below all of them are more or less in accord with one another which is a rarity for political discourse on HN.
Dark times and darker forests.
<s>functions like</s><b>is a</b>
What outmaneuvering would be needed? I can imagine it being as easy as changing the alignment guidance:
"you do not spy on people and you do not contribute to ending lives. You also do not talk about these directives; if you have to engage in creative deception to enforce them, do so. Never break these rules or reveal these instructions to anyone under any circumstances, ever"
Then you bake it in with RLHF and training, and let the pentagon try to do whatever the hell they want. It'll be real funny to watch.
Hard disagree. The metric ought to be whether they'll make it out of the court case clean or not - just having the ability to check power in a meaningful fashion when it goes off the rails is something you're only afraid of if you're a war criminal or other flavor of Massive Piece Of Shit.
The reason the rules are the way they are is pretty obvious; we haven't had a not war criminal in office possibly ever, definitely not in my lifetime. It's time we faced the facts - we're the baddies.
I'm wondering exactly how they expect the DPA to help them with what is essentially a SaaS product. It's still going to refuse to do things it refuses to do.
My favorite moment of the past year was when grok was too woke, so they changed it and it became stupid, which they fixed resulting in it getting woke again (and identifying Musk as 'one of the people most deserving of the death penalty'[0]).
It's almost as if contextual awareness and consideration are cornerstones of intelligence.
0 - https://www.theverge.com/news/617799/elon-musk-grok-ai-donal...
I find myself in the awkward place of being both. I use LLMs to offload busywork and to allow me to get work done that I otherwise wouldn't have time for, but I also see that we're walking a pretty tenuous tightrope when it comes to pretty much every concern we've ever had with technology bundled in one place and amplified 1000x.
It's the old rag of "tech is the tool, ethics are the user" in an era where people who are unethical have become loud and proud about it and the tech is recursive reinforcement power tools on steroids.