curious_straight_ca
No bio...
User ID: 1845
curious_straight_ca
1yr ago
What we will see is specialized AIs and AIs trained on specific data for specific tasks.
People have been predicting this for half a decade and it isn't what happened. Instead, we got LLMs trained on all kinds of text data that generalize across the different kinds, and now we're getting omnimodal models trained on text, images, video, etc. Why not just train your big model on all the specific data from different tasks together?
curious_straight_ca
1yr ago
Which brings me, ultimately, to my main point: what are you all using AI for?
For me, it's significantly better than google or stackoverflow on programming topics where I don't need to deeply understand what I'm doing, I just need to fix some misconfiguration or use some library or find the right function or whatever. And it's good at asking questions about long documents when I don't want to read them. And I don't use AI that much relative to other people.
LLMs are basically a context-approximation and text-generation tool, not an organic information generation system.
They can do many college-level math problems that are beyond the capacity of the average person! Sure, they're doing it with more 'memorization' and less generalization than the average person who can do them, but that's still a huge step up from what computers could do ten years ago. Why will AI stop improving?
curious_straight_ca
1yr ago
Reading this forum has been one of the biggest shocks I've experienced to my beliefs about the world ... people who have such crazy policy preferences to me not because they disagree about facts like the far-left people I meet, but because they honestly believe that people should be treated differently solely because of the race they were born as.
My impression is that very few mottizens are racialists, in that they think that eg black people should be treated differently from white people 'solely' because of race? Some people here don't really believe in HBD and are 'liberals' or 'progressives' on immigration. There are people here who want to treat people differently based on IQ and genes, but with genetic predisposition to IQ or other traits set equal would happily let in the few Nigerian or Iraqi neuroscientists if their merit was assessed accurately. There are people here who are more like normal American conservatives on the topic, and would judge people not on their race or IQ but their culture, their willingness to assimilate to American ideals and national identity, etc. I think all of those outnumber people like SecureSignals
x: 10 as stated, but with the caveat that differences within populations matter more than differences between populations - I'd rather work with 99th percentile white person than the average jewish person.
y: 7, rising to 9 at the tails, but the difference in eg income or achievement is somewhat less because there are other factors in between intelligence and outcomes.
z 1, if you take anything resembling a materialist or consequentialist approach to morality, even if it's also religious, smart people are capable of having more valuable experiences / contributions to society than less smart people
What evidence would change my mind:
If the next generation of the best writers, startup founders, scientists and mathematicians, programmers, et cetera had significantly higher black and hispanic representation, that would pretty quickly change my opinion. Or a change in my observation that, in general, any area with a high 'skill ceiling' in terms of intelligence seems to have asian/jewish overrepresentation and hispanic/black underrepresentation. The fact that that's currently not true is a bigger factor in my belief in HBD than any science or statistics right now. And the claim that various forms of racism cause this just doesn't add up, and here are a few more things that, if I was wrong, would change my opinion: Jews and asians with poor or average backgrounds are more likely to rise up into that cognitive elite than black people with rich backgrounds. And top people who are self-taught, top people in hobbies that require a lot of intelligence, or just the smartest people in internet communities I'm in, have just as much asian/jew overrepresentation and black/hispanic underrepresentation, even though you'd expect 'implicit bias' and 'systemic discrimination' to matter less for something entirely driven by personal interest. Gifted children who score high on tests show the same racial patterns. "Okay, but, uh, environmental racism, poor children have worse diets and this stunts their growth, lack of representation prevents them from achieving their dreams, epigenetics..." One of my favorite programming bloggers "grew[sic] up in a household that spent so little money that [he] was regularly lightheaded from hunger because I was actively starving", and yet has gained a significant following and is clearly very smart. He is, of course, asian. And leaving aside races, my personal experience also very strongly suggests that there's some 'general factor' of intelligence that some people just have a lot more of, that it's almost impossible to increase, and that is heritable. Every alternate explanation for differences in IQ or achievement between races just feels like a 'god of the gaps', and doesn't fit the whole picture nearly as well.
Even if we cordon that all off, start from an uninformative prior, and just look at the science and data, HBD still wins. Arguments for the validity of IQ and individual heritability of IQ are very strong, and the arguments for genetic IQ differences between races are pretty strong. There are sophisticated arguments against all of that, eg gusevlab.org/projects/hsq/, but if I dive into any individual technical argument usually either the HBD side wins or both sides have bad arguments. If that changed - if I changed my mind on the strength of the scientific evidence, or factual claims by hereditarians that I currently believe in ended up false - I'd become significantly less certain, but it'd move me less than the above observations changing.
curious_straight_ca
1yr ago
The kind of woman who would be 'impure' a few hundred years ago is one who went against the explicit desires of her family and culture. The kind of woman who is 'impure' today is one who does what the culture guides you to do. The supposedly deontological choice is just selecting for something very different today. Nothing is absolute, and not being pure is, if it's a negative signal, necessarily a weaker one today, and not one worth trading off against everything else. Just like a culture with deep traditions about planting and harvest times need to modify those traditions when they move to a new climate.
curious_straight_ca
1yr ago
It provides a lot more visual impact than just classified documents with markings.
This is true, and probably isn't ideal, but it is not a huge issue. Maybe prosecutors shouldn't do things for 'visual impact', but they do, and at any rate the conduct of Trump and his lawyers at various legal proceedings has been 100x worse.
It gives the impression that it would be obvious to anyone who casually looked in the box that it had classified documents. This is important because "knowingly" is an element of some of the charges.
Pretty sure the documents themselves have clear classification markings on them?
It effectively substitutes the FBI's CLAIM that the documents were classified for the actual evidence of classification.
To the public, maybe? The claim is true, though, and it's not evidence to the courts, though. If those documents weren't actually classified when trump was President, Trump's lawyers would be all over that.
Since the classification markings on the pre-printed cover sheets didn't have to match those on the documents, it provided the impression that the documents had perhaps a higher classification level than they did.
Do you actually think this made a difference in anyone's reaction to this case? And, again, the courts are considering the actual documents and their classification levels. Here is an article going over the actual documents and their classification levels and contents. I do not think the cover pages were materially misleading given that.
curious_straight_ca
1yr ago
Anyone can make a toplevel post, there've always been a few of these.
This is just politics as pro wrestling, meaningless simulacra level 3 stuff. He's posting galaxy-brained dumb justifications for things his audience believed anyway because it gets him likes. Threats of assassination are not a meaningful incentive in US politics right now, if anything it's lower than ever when you compare the rate of political violence a mere 40-80 years ago to now. If they were, the stakes would be a lot higher than 'have a VP that's slightly worse than trump'. Assassinating a sitting President Trump would both be a huge escalation against 'process' and 'democracy', which the deep state loves, and extremely stupid because it'd backfire horribly in terms of support for Trump, people love martyrs.
Sticking with the theme, let's look at the quote tweets. From one of the DR people, Peachy Keenan: A very experienced criminal psychiatrist who once testified in the Bobby Kennedy assassination case told me TODAY that he is convinced with absolute certainty that assassination is the only path left for the Democrats. Chilling.. Chilling, yes. Candace Owens: Was literally saying this 48 hours ago. Trump should obviously choose Vivek or someone like @RepThomasMassie for this reason. He chooses Tim Scott or Nikki Haley, and we are in trouble. Needs to be someone the establishment cannot control. Very reasonable.
curious_straight_ca
1yr ago
Remember how many times progressives on social media were wrong about Russia, and about Trump's legal woes in general? I think you're doing the same thing in reverse here. What the government's alleged to have done is very minor, but a lot of the words look like the words you'd use in a major situation, so it's blown up into a big deal.
curious_straight_ca
1yr ago
The DOJ’s clever wordsmithing, however, did not accurately describe the origin of the cover sheets. In what must be considered not only an act of doctoring evidence but willfully misleading the American people into believing the former president is a criminal and threat to national security, agents involved in the raid attached the cover sheets to at least seven files to stage the photo.
This is a tendentious presentation imo. Politico presents this as:
Smith’s team revealed in the filing that FBI agents carried printed “classified cover sheets” during the Aug. 8, 2022, search of Trump’s Mar-a-Lago estate and used them to replace any classified documents they discovered in cardboard Bankers Boxes that littered the former president’s residence.
“The investigative team used classified cover sheets for that purpose, until the FBI ran out because there were so many classified documents, at which point the team began using blank sheets with handwritten notes indicating the classification level of the document(s) seized,” the prosecutors wrote.
“Any handwritten sheets that currently remain in the boxes do not represent additional classified documents — they were just not removed when the classified cover sheets with the index code were added,” Smith’s team wrote. “In many but not all instances, the FBI was able to determine which document with classification markings corresponded to a particular placeholder sheet.”
I think it's reasonable to put cover sheets on the classified documents, given they are classified. The documents would have already had classification markings, so I don't see how this is "willfully misleading" the public "into believing the former president is a criminal and threat to national security".
It turns out that when the government alleged that Trump had classified documents he was not supposed to have, the government itself did not accurately know which documents Trump had, or which documents Trump was even supposed to have. Actually, worse than that, it turns out they fabricated some or all of the accusations
"Some or all", here, seems unjustified - I don't think anyone (other than perhaps Trump on Twitter) is claiming the accusations are all fake - that's a much stronger claim than "the documents aren't in the same order that they were when we scanned them". Your sources imply this is like "tampering" with evidence, and it may (not sure) be a procedural issue, but things like "adding cover sheets" and "reordering documents" don't undermine the claim that Trump committed a crime.
curious_straight_ca
1yr ago
This again? If Biden wanted to cut down on illegal immigration, he could do it now, without any additional Congressional authority
... Okay? Ben's not arguing Biden is blameless here, just that Trump is blameworthy. Yes, Biden and the Dems aren't doing this out of sincere care for immigration, they're trading a better chance at getting elected for a concession to the public's policy priorities. Trump should, by the values of his own voters, take the deal and reduce his chance of winning because this would hugely reduce illegal immigration.
curious_straight_ca
1yr ago
I suppose it is stupid that they don't just post the email's contents. Like when there's an article about something happening in a lawsuit without a link to a pdf.
curious_straight_ca
1yr ago
I don't think there's anything wrong with an email as a source, if you email MIT's comms office and they respond with a clear claim? They could be wrong, or lying, but they could lie to the NYT too.
curious_straight_ca
1yr ago
These are important barriers on a timescale of a few years, but on the scale of decades, the march of biotech and basic research will overcome imo.
curious_straight_ca
1yr ago
Hmmm I spend time around both, more around wokes / progressives and I really don't see it.
When I think about the thoughts that motivate someone who leans progressive, I think things like George Floyd, the idea of someone not being able to pay off medical debt and foregoing care as a result, black kids who can't get good jobs because of racism, imagining a kid who died in a school shooting because we don't have gun control, someone who's mocked for being gay, etc. And also the strong social taboos, and internally confusing the social taboo with justifiably taking offense at words that harm people. I'm having trouble of thinking of an interaction where it felt like people were really, genuinely, afraid that history would judge them. They feel much more afraid that their current social group will judge them.
curious_straight_ca
1yr ago
My take:
Look I love rolling in shit as much of the next guy, but I think it has to be interesting. Yeah, critical theory gay pedo, I remember watching that video four years ago or whatever, but you can't just say the line for the thousandth time, try to do something new with it.
If jannies had the balls to moderate on QUALITY, and just say "yeah you're irredeemably r-slurred have fun on .win", jannies would be seen as sigma monarch chads and not whining virgins.
This place as a whole doesn't want any of that toxic stuff anyway. I'd say run back to rdrama, but rdrama comments in 2024 are so dull I sympathize with coming here. Try twitter, idk.
curious_straight_ca
1yr ago
Not the kind of comment we're looking for here tbh
curious_straight_ca
1yr ago
I also vaguely remember something like that happening? It might (or might not) have been this. https://time.com/collection-post/6140206/cultivated-meat-passes-the-taste-test/
Again keep in mind that I'm comparing this to shitty meat. I don't even like shitty meat.
curious_straight_ca
1yr ago
I'm not overselling the relevance of critical theory to the western academic tradition as a whole, the critical theorists really do have hundreds of thousands of citations, and one of them is in the top 10 of all cited academics in any field (habermas), also you'd probably include Foucault who is the #1 most cited. I'm not claiming he has a lot of mainstream cultural relevance, but given the crossover between the 'elite' and academia it is not weird for him to be interested in critical theory.
curious_straight_ca
1yr ago
What do you mean "you can't, though"? I am really quite confident that I could get lab-grown meat that passed a blind test for something like tens of thousands of dollars per pound if I for whatever reason really wanted to. It's not that difficult of an engineering problem, we know how to create the relevant tastes and textures, the problem is getting costs down to what nature's gotten very good at over a billion years.
curious_straight_ca
1yr ago
I think sometimes the learned behavior of trying to make arguments that make outgroup look dumb overrides your attempts to actually understand the world around you. This is one of those cases. Historically religious people actually believe in their religions, as physical facts about reality, as much as they do anything else. People have mental breakdowns about heaven and hell! Whereas being on the 'right side of history' is, in its entirety, a rhetorical device to refer to social pressures or empathy for oppressed people who exist today. Nobody who says that is actually imagining dozens of people looking back on them from a century after and being disappointed. They are not at all comparable.
curious_straight_ca
1yr ago
The way you say it you'd think Critical Theory was an esoteric subversive cult and not a very popular and influential historical school of thought, whose notable figures rank among the most cited individuals in all of published research everywhere.
OP reads like it's criticizing the substance of therapy, as opposed to the standards of the self-regulating body. It might be a problem, but it doesn't fit into OP's narrative claim well. Anyway this is uninformed speculation, but therapists self-regulating to that extent seems much harder than medical doctors. Medicine exists in an institutional context, there are clear guidelines, obvious standards of harm, clear records of symptoms and treatments, and institutions that can keep those records and have processes for reviewing potential misconduct. Whereas a therapist individually meets with their clients one on one, interactions are private, and any investigation of misconduct would finely depend on the facts of personal relationships. Who would even start the process in this case, when the client and therapist are both happy?
</speculation>, Looking at official procedures reddit threads, it looks like therapists sometimes do lose their licenses, but the attitude in the thread seems to be that the process isn't great and many who lost their licenses for good reasons seem to earn them back, and the offenses are generally significantly worse than providing bad advice, mostly crimes or having sex with clients.
curious_straight_ca
1yr ago
I mean, everything is a norm. The command structure of a military at wartime is a norm. The security of your home is a norm. Adherence to contracts is a norm. That police arrest you for breaking the laws passed by the legislature, as opposed to the laws the police chief makes up, is a norm. Yet, we don't expect any of those norms to fall apart.
Yes, any of those norms could change. But they'd need good reasons to. Corporations are very familiar with what happens when they blatantly break federal regulations - it goes poorly - and they maintain close relationships with regulators. Texas just doesn't have much leverage, as one state of many, and the biggest and most economically productive states are blue anyway. Yes, sufficient disruption could break this norm - if the regulators started demanding war communism, your hypotheticals would stop being hypothetical. But they aren't, and nobody involved has enough dissident energy to do anything.
No, google will send me to a site like wolframalpha which can solve it because the techniques for solving these problems have manually been programmed in for each type of problem. LLMs learned how to solve them by, more or less, being trained on a lot of text that included a lot of math problems. The latter is clearly a lot more like how humans learn than the former.
More options
Context Copy link