Aransentin
p ≥ 0.05 zombie
No bio...
User ID: 123

Yeah, there's a lot of text on the internet. With a pretty cursory Bayesian analysis, even with a 99.9% accuracy you're looking at a thousand false positives if you are combing through a million posts. Without some other thing to narrow it down, it seems reasonable that it'll not be possible from writing patterns alone.
You need to cook/toast the pop tart.
I did! It made it even drier, if anything.
Sampling the candy and processed stuff was more for the experience, not to get something that I expected to be genuinely enjoyable. The US for sure had some very good food and drinks as well; the craft beer is especially good compared to the naïve opinion of the Americans as shitty-beer drinkers.
Same here, but presumably there's not nearly enough of us to significantly affect anything. I think it mostly targets children; they don't run ad blockers to the same extent as other groups, they often spend a large amount of time on youtube, and are more perhaps more easily influenced by advertising itself.
Price too high and people complain about prices
I wonder if that's part of the reason for the giant "service fees" and such that the ticket vendors charge – by claiming a large part of the price is from the vendor, they are effectively unloading the bad publicity on them instead of on the artist. It doesn't really matter that much for Ticketmaster if people hate them, but it does for a musician with a fanbase to think about.
My article really only covers generative models, like the recent Stable Diffusion. Controversial models like classifiers that try to evaluate how likely somebody is to commit a crime has entirely different considerations. Maybe I should have made that more clear.
Also I disagree that a "de-biased" crime model would discriminate against white men! Men commit a highly disproportionate amount of crime compared to women; any sort of adjustment you make has to adjust for that, adding a whole bunch of likelihood on women especially, probably more than the racial difference even.
Another reason why it intuitively feels worse than murder is that I could imagine myself (if the conditions were sufficiently extreme) perhaps killing another person with my own free will. Not so with rape; even though the act of murder itself is worse in my ethical calculus, rape categorically reveals the base nature of the perpetrator in a way murder doesn't.
I'd compare it with somebody who has their pet cat put down so they could cook and eat it. Morally not much worse than cooking some calamari, but it really says something about how messed up the person is.
If a English isn't your native language, are there any words that aren't in English that you miss?
As a Swedish speaker, glapp is one example. It means a loose connection somewhere in a circuit. E.g. if the sound intermediately cuts out in a livestream, you can say "the sound is glapping". A useful concept to be able to quickly express, but sadly lacking in English. Any others?
Regarding the evidence standards in the first point - how is this not constantly abused by police?
If a policeman commits a crime when looking for evidence he'll potentially be charged with misconduct (in addition to whatever else he did, like battery or burglary). A particularly egregious case from the top of my head was this one where the policeman was later convicted, fined, and fired.
the wholesale trade of fundamental liberty.
I think the idea here is that if the government ignores the constitution, then any law you have to address it can just be ignored as well and it's up to the citizens to fix it by whatever means are necessary. The important thing here is if this actually works, or in practice results in constant low-level constitutional violations that people ignore; personally I don't think so. I helps a bit that the Swedish constitution isn't as hairy as the American one – the exception being freedom of expression where boundaries can be unclear / debated, which is indeed the parts that do have legal systems in place to decide that.
(Another fun legal thing I forgot to mention is that the king (or regent) is immune from prosecution. This question comes up from time to time as the king pretty frequently gets caught speeding and potentially drunk driving, but the police always has to just let him go.)
If you somehow was able to give me positive dignity equal to the inverse of me flopping about in slow motion and failing to catch a ball in front of a million viewers, then I probably would pay 500 USD for that, and I suspect a lot of people would too.
(Actually not even $500 – it's multiplied with the estimate that the person would actually pay up without any sort of trick or gotcha, which would be pretty low in that situation if you weren't aware that the person is some sort of moderately famous youtuber.)
In Demolition Man (1993), there's a scene where they mention Taco Bell. In the international version, this is dubbed over to Pizza Hut, presumably because it's more recognizable outside of the US. You can still kinda tell that the audio had been changed though, as their mouths don't match.
I'm pretty sure I wouldn't have spent much time reading the Taco Bell wikipedia page if that hadn't been done.
With an account you can just use the API or the playground. That gives you access to the model without all the safety stuff trained into it, though you'll have to pay a modest sum for usage credits.
Using that you can create pretty much whatever problematic content you want.
Swedish and German does a thing where we end questions with "or" ("eller"/"oder"), causing people to write stuff like "Do you want a sandwich, or?" in English.
Much of media works on narratives. "Wealthy petrostate spends an obscene amount of money on prestige sporting event" is a nice bite-sized one.
If the journalist mentions that the official numbers might be made up, it adds complicated nuance to the story. How expensive is the World Cup really, if you can't trust the numbers? This leaves the reader uncertain and mildly confused instead of thinking they have learned something about the world. Easier for the article to just print whatever BS amount Qatar says and move on; the vast majority of journalists have tight deadlines and wouldn't have time to investigate it anyway.
Some effect similar to the euphemism treadmill, perhaps? No matter what method you use, that method will steadily lose favour due to it being associated with killing, so you keep inventing new ones to shed the old emotional baggage.
They cut the fiber optic cable between Svalbard and Norway
Is there any good evidence for Russian involvement in that? From my (admittedly cursory) search it seems it's just speculation.
"Coincidentally" there was this popular tweet ("Horror story where the same ominous figure recurs across Stable Diffusion samples regardless of the prompt"), shared by e.g. Yudkowsky three days before. Quite likely that the "Loab" author saw that and decided to spin up a hoax on it.
For the purposes of this comment, I will try to define good as "improving the quality of life for many people without decreasing the quality of life for another similarly sized group" an vice versa.
Tangential, but the term in economics you are touching here is a Kaldor–Hicks improvement I think. It's not Pareto-optimal, but total-wealth increasing, and could theoretically be converted to a Pareto-optimal situation with redistribution from the winners to the losers (assuming such redistribution does not have any externalities itself!).
you shouldn't trust ridiculous nonsense like a Kardashian sponsored toothbrush
I figure that can be perfectly valid evidence for the quality of the product! A company shelling out money for sponsorships signal that they believe in the product itself, which is important in situations where the consumer needs confidence that they won't drop support for it in the near future. For e.g. tech like game consoles this is especially valuable.
The Kardashians also have a personal brand to protect. If they sponsor a product I can be more confident it won't be so bad it'll damage their image; ceteris paribus this is certainly better than nothing.
How concerned would you be if your home address was visible online if somebody googled your name? What about your age, SSN, tax return (naturally including your exact income), company involvement, real estate ownership, and every court case transcript and and police report you appear in?
In Sweden, that's all public information. If you looked up my name you'd find my address pretty much immediately, which you could drop into google maps to get a view directly into my kitchen. Police reports is slightly more difficult; you'll find websites stating things like "Foo Bar is present in 2 court cases! Pay $10 to see them!", and you'd have to send an email yourself to the court/police if you don't want to pay for that.
In addition, if you happen to work for the government, then all your work letters, emails, and instant messages become public information as soon as the case it concerns is closed. Tangentially this means that it's very easy for a Swedish person to be a major pain in the ass for government agencies, as you can anonymously keep bulk-requesting random emails and and the employees have to do time-consuming archive digging.
I find it interesting how people here really don't care about it. Hell, I don't really care, mostly because it's been like this my entire life and nothing bad has come out of it (yet? Knock on wood).
Could you, say, send an API request to a bank from within your webpage, and then read the response and cookies from the host page? I'm thinking this would be blocked by both browser and site technology. This has to be what CORS is for, right? Not just to annoy me while I'm developing?
Yeah, it shouldn't matter, but if a site has e.g. an XSS vulnerability the attacker will need to be able to run some initial Javascript as the victim to kick it off. Sending an email so that they'll visit a specific page might be just that.
nothing bad can really happen to you just from following some random link.
Attackers can host exploit kits on the target site, which spray a bunch of exploits against your computer if you visit them. If your browser/plugins/extensions/OS isn't fully up to date this might very well successfully install malware. This was really relevant when I worked in IT security about 6 years ago but seems to have declined a lot recently; still, it shouldn't be dismissed out of hand quite yet.
I find using GPT-3 as an "unblocker" works quite well. Insert the last few paragraphs you've written, and let it complete the text. The result isn't always very good, but you frequently get decent ideas on how to structure the next section.
bias the algorithms ahead of time
While anti-bias efforts are easy to abuse, I don't think they are inherently bad. There really is a bunch of detritus in the datasets that causes poorer results, e.g:
-
Generate anything related to Norse mythology, and the models are bound to start spitting out Marvel-related content due to the large amounts of data concerning e.g. their Thor character.
-
Anything related to the "80s" will be infected by the faux cultural memory of glowing neon colours everywhere, popular from e.g. synthwave.
-
Generating a "medieval knight" will likely spit out somebody wearing renaissance-era armour or the like, since artists don't always care very much about historical accuracy.
This can be pretty annoying, and I wouldn't really mind somebody poking around in the model to enforce a more clear distinction between concepts and improving actual accuracy.
Hm, yeah, thinking about it that would create an enormous uproar for sure. Even if you spent a huge amount of time and resources to perfect the plan people wouldn't really know that; you'd have lots of people assuming anyone can now buy a drone on aliexpress and easily assassinate people without getting caught.
As for the viability itself, it's still kinda iffy:
- I think letting out a drone from a hotel room window might easily get noticed as well? They aren't exactly silent.
- The police is going to wonder where the drone came from, and for sure check out who rented the rooms in the nearby hotel.
- A switchblade-style kamikaze drone seems exceedingly hard to make yourself. The repurposed commercial drones used in e.g. Ukraine simply drop grenades, and take a lot of time to position. That works for soldiers hunkering down for a long time in trenches, not so great for a person walking to his car.
- Renting a hotel room takes time. You'd have to know the exact itinerary of the target to prepare all that in advance, and sit around waiting for a really long time for the moment to strike. The more time it takes, the more potential for mistakes and random events leading to your discovery.
The idea that GDP is fake is mostly cope (and I say that as an European), but there is a nugget of truth in it as a lot of things rich societies spend money on really is 'fake' in a way:
-
Positional goods. Things that only/mostly benefit you if you have better stuff than everybody else. This includes luxuries like fashion, but also much of higher education, and all ways people price out poor people to e.g. not having to live next to them.
-
Waste. The government spending a gazillion dollars on a 4-year environmental pre-study for some infrastructure project without any tangible result absolutely counts as GDP but doesn't really benefit society much.
-
Paying for results that other people get for free. If you live in a high-crime area and have to spend a bunch of money on replacing stolen goods, security, insurance, fixing vandalism etc. you are contributing to GDP even though somebody living in a low-crime place get that automatically.
This has probably always happened in all societies to some degree or other, but it's just more prevalent in the richer ones that can afford the slack.
You can use the "function calling" mode for that, using the API. It restricts the model to output JSON, so it doesn't get any opportunity to scold you about your questions.
I asked it to "Provide a list of (at least ten) races and their average IQ", and limited it to only return an array of objects with a "name" and "iq" field. The result was this.
More options
Context Copy link