They were tired of the reputational damage caused by scum selling fake chips that identified themselves as real FTDI usb-serial devices and issued a driver update that stopped working with fakes.

I'm not saying that as an insult that's just a literal fact.

It is extraordinarily expensive to train a police officer. Much of this expense is because of corruption. Most of the training that we give them these days is pseudo-military training that looks cool to idiots. Think gravy seals. But we still have to pay for it all and the city has to pay for it.

The upshot to that is doing the actual job is incredibly boring. It's mostly pulling people over and giving them tickets. Something like 70% of police officers will never even draw their sidearm.

However the job pays reasonably well given how bad the economy is and has a clear career path to retirement. Along with a lot of opportunities to make extra money on the side.

So the problem cities had was people would go through all the training get bored out of their mind at the job and quit and go do something else. So they started to do testing to detect people or too smart. People who would get bored.

I haven't done it myself but I have friends that have applied to be a police officer and been turned down and when they looked into it it literally came down to they were too smart and the algorithm said they would quit the job after 6 months to a year haven't gotten all that expensive training...

As a result we intentionally select dumb people to be police. As an added bonus if you need them to bust heads at a union meeting they're all on board...

Dumb people tend to be violent because when you're not all that bright you quickly run out of options. Like the old saying goes, War is the continuation of political policy by other means...

So you get a lot of police officers beating their wives. You get a lot of cops shooting people they shouldn't be shooting. That sort of thing.

And you get a lot of cops spying on their exes.

They're a well known and well understood solutions to this problem. Other countries have done them. But America generally isn't up for that

No. We do not know how the decision was made. We can only look at logs and guess when patterns are found

Please stop commenting on things you don't understand. This is a conversation about attribution graphs, not CoT.

I think we're pretty much on the same page with this. I think what we need to always remember about evolution/natural selection is that even the process itself is subject to natural selection. Adaptive behavior is itself a trait that can evolve like any other trait. Even to the degree that, while Lamarck wasn't exactly right, it actually really is the case that some environmental pressures will cause hereditary changes through generations through what I think we're still calling epigenetic factors. Usually these are where stress factors like prolonged food shortages can lead to, for example multiple generations of smaller individuals despite nutrition levels rebounding. Or, even in a single extant individual, prolonged exposure to environmental conditions can lead to long-term physiological adaptation such as lots of exposure to the cold reversing the behavior of blood flow to extremities from restricting blood flow to conserve heat while sacrificing the extremities to increasing blood flow to protect the extremities. Being able to change to adapt to an environment is usually an advantage, so it makes sense that organisms would not only evolve, but that evolving to evolve would make sense from the "selfish gene" perspective.

I mean, Milton Friedman died 20 years ago at the age of 94 so...

The problem with a plan for taxation "every time money changes hands" is that if you take it too literally, two people handing a $100 bill back and forth with a 20% tax can end up owing more than $99.99 after 42 passes. Or, if you're rounding up to whole dollars at 50 cents, the whole $100 after 26 passes. You can say that example is absurd and that you didn't mean it literally, but the point is that all kinds of absurdities like that can, indeed happen. The problem is that any scheme like taxation can run into something like the shoreline problem. If you're unfamiliar, that's where the length of a shoreline has a minimum value, measured in a straight line from point to point, but if you actually follow the contours, the length tends towards infinity the smaller your measurement increments are. With small enough increments, you're measuring the contour of every grain of sand down to the atomic level and below. Obviously in the real world there are limits like brownian motion, the level of microscopy, the time involved, eventually the Planck length, etc. but even in the real world, the shoreline length can vary by orders of magnitude. So, to actually compare shorelines by length, you have to be using the same standards to measure each and recognize that in come cases, the comparison will change a lot if you change those base standards.

The same applies to taxation. On its face, systems like what you propose seem simple, and therefore fair, but in fact that model is only simple when reality is very simple to match. The reality is that many perfectly legitimate ways of doing business would lead to ridiculously high taxation using the model you propose. going back to the shoreline example, your model works perfectly if all landmasses are perfect Platonic polygons and you're measuring just one edge. As soon as they turn into real world objects though, your system fails and you end up needing all kinds of rules to try to reach some semblance of reason and fairness.

All that said, the rules end up with massive numbers of loopholes, either accidental or intentional and are also simply chaotic and messy based on their nature and the fact that they can't account for all situations. They also definitely take advantage of the little guy. However, the proper response is vigilance by people who understand the ins and outs of the problem, not simplistic rules. For every problem, there is a solution that is simple, obvious, and wrong.

This makes more sense if it's laid out in something like timeline order and with some more clarity on the parties involved.

WB makes the movie for $70 million, but shelves for whatever reason and treat the $70 million as a loss. On that loss, they get a $30 million tax writeoff.

Movie sits in "vault" for a while.

Ketchup, a distinct entity, not a WB shell company, buys the movie and distribution rights entirely from WB for $50 million. This forces WB to give back the $30 million tax writeoff, meaning that for WB it's minus $70 million, plus $30 million, plus $50 million, minus $30 million. So they end up with $20 million more than they had with the tax writeoff, but also ended up $20 million in the hole. Ketchup meanwhile, has a clean slate with a complete movie they got for $50 million.

Ketchup then takes their $50 million dollar movie and spends an additional $10 million on it for marketing, making their total cost $60 million. To make that up, they only have to make $60 million dollars, not $120 million. However, they don't own the movie theaters. Essentially, the cost to them for having the movie shown in the theater is half the box office price. So, in theaters, every dollar of ticket revenue is fifty cents for them. So, movie theaters need to make $120 million for Ketchup to get back their $60 million. However, to complicate things, Ketchup owns the complete distribution rights, so they can also make money on other media, streaming, disks, television, etc. not to mention that they also sold some of the foreign distribution rights for $20 million up front. Then there's the complicated question of merchandising. Ketchup owns distribution rights, but I don't know what their merchandising deal is. WB still owns the character copyrights and trademarks outside of the fixed forms for the movie, while Ketchup clearly has promotion rights. In theory they have a profit sharing agreement for merchandise based specifically on the movie, but it might get complicated where merchandising and promotion collide. Like T-Shirts, or happy meals, etc. While the lawyers probably hammered things out in minute detail, I still would not be surprised if we see a lawsuit over this at some point down the road. In any case, the extra demand for merchandise from WB that the movie will create will probably also make back WB's $20 million loss as well.

In any case, at its current pace, it is pretty much guaranteed to make $120 million at the box office, making it at least break even in theaters, then it will receive additional profit in other media.

That does lead to the question of whether WB would have made a profit since it cost them $70 million to make it. So if they had then spent say $14 million on marketing it for a total of $84 million, they would have need $168 million in box office sales to reach break even. So there's the question of whether they could have turned a profit. It looks like the movie is going to hit right at around $120 million at the box office, but with the extra they would have spent on promotion, it's possible it would have also gone further. Then of course there's other media. Ultimately, in the long run, it probably would have at least gotten close to break even. That might lead some to say that WB was right to not release it. Except that meant only taking a $30 million dollar writeoff, when it still would have brought in much more than that. On the other hand, when they consider competition at the box office, they are also competing against themselves, so they probably also consider whether they think they will get more of a pie slice (if they think of it as a zero sum game) if people go to see another movie and its one of theirs. So, their reasons may have been complex.

Ideally of course, for comparative rating, the ratings themselves will be meta-rated for reviewer bias. For example the guy in the summary who intentionally chooses the answers he thinks will hurt the model most. Then of course there's the issue of audience selection bias and self-selection. The employees who work doing this probably come from a various demographic subsets, but probably trending towards mostly male and towards the terminally online. Then there's also the question of whether audience preference for results is actually desirable, which depends on your goals for the AI. Consider, for example, if your goal is education. In an educational setting, material progresses and becomes more challenging, but if you go by what the class prefers, the lessons will trend towards the lowest common denominator and you'll never get to higher levels.
I'm sure theorists working at these AI companies have thought about all of this of course. It doesn't necessarily mean they're currently doing it though.

So, the reality is that the world "ran out of training data" for the most part years ago, and the models have gotten exponentially better relative to a number of parameters. Claude 3.7 Sonnet was released 1 1/2 years ago, and today it benchmarks about the same as Qwen 3.6 Sonnet 9B, a model two orders of magnitude smaller than it, and which is itself two generations out of date. And a large chunk of this is done with synthetic data - aka, data created by other models.

It's simply a myth that "data created by models consumed by other models makes them worse". In practice, it's used to make them vastly better. Models aren't collagers, they're reasoners. Learning the results of reasoning, the results of trial and error, etc helps build a stronger base for more advanced reasoning. Also, our training algorithms, while reaching a denser knowledge compression than human brains, are less efficient learners than human brains (per unit data), so they need to see the same sort of data from "many different angles", to substitute for our process of "mulling over" new information.

(Yes, it is possible to set up contrived scenarios where, say, an small image model is fed only its own outputs on loop, little bits of knowledge slowly being lost each go-round, in a situation equivalent to leaving a person alone with their thoughts in a dark room for ten thousand years - but even a tiny percent of new fresh data added to the mix prevents this degradation.)

And as for the article itself, they made it sound like they're talking about, say, programmers banned from using AI at OpenAI, but it's nothing of the sort. These are data labelers. In the old days, they used to be far more common, and in wide use in all types of model creation. That's no longer the case; they exist for special cases. For LLMs, this is much more limited:

* Subject matter experts: people with rare professional-tier knowledge. Often used to validate model outputs, where nobody else could (for example, OpenAI hires mathematicians to validate their models' proofs)

* Chain of thought / logic auditing. Increasingly important now that models are showing increasing signs of poor alignment. You can automate this a lot, but you really still do want a human in the loop *somewhere*, in case your auditors get compromised.

* Side by side comparative rating: Which model's output do you like more, A or B?

* Evaluating reported outputs where users reported that they thought the response they received was bad, and if there's actually anything wrong, copyediting the output for training.

* Adversarial prompt generation / jailbreaking and evaluation. Again, you *can* have models do this (and companies often do), but you don't want to just rely on them.

* Trying to set the bounds on whether given queries should be refused or not (for example, "How do explosive reactions happen in chemistry?")

Basically, a switch from "click work" to "knowledge work". This is no longer the era of "Write a poem about cats" or "Explain how to solve this algebra problem" to build up a training dataset. You're getting paid to think, not to repeat a rote task.

Other types of labelers aren't as far along. Multimodal data is less advanced than text, so you'll still for example have people labeling things in videos, transcribing heavily-accented audio, grading text-to-video consistency, things of that nature. Probably the least advanced field is robotics, so there's still an awful lot of manual evaluation and correction in that.

But anyway, if you're hired to do any of the above, it's because they specifically want you to do that. Having an AI model do the above (beyond the listed caveats) entirely defeats the purpose.

No. A black box is something you don't understand how it works. We can understand any and every decision that they make. We simply have to look.

I guess the only positive thing I can say about them is: despite how tough the models were to work with then and get good content out of them, these people were generally obsessive over their waifus and porn, so at least they weren't making like 8-fingered monstrosities. You could tell that they spent many hours zoomed in, upscale-regenerating and collaging and photoshopping over every imperfection.

hasn't really taken over human actresses despite that industry usually being in forefront of all kinds of tech, I wouldn't bet on this one either.

I dunno. This isn't exactly my space, and I haven't even been involved much in image generation at all in years, but in the old days at least, the Stable Diffusion Reddit and the model sites like Civitai were just *flooded* with stuff from horny guys, like 80-90% of the content. It was really annoying. Due to anti-porn restrictions, the Stable Diffusion Reddit mostly got flooded with waifus or similar, but Civitai was mostly porn models, or at least general models finetuned to allow porn. People were even widely using waifus and porn in their tutorials and user guides for image generation tools, even though the the tools were just for general image generation purposes. Oh, and YHVH protect you if you dared complain to any of the waifu or porn posters about any of this - you'd get a flood of "WHAT, ARE YOU A PRUDE???", "HAVEN'T YOU EVER SEEN A WOMAN BEFORE???" , etc and get modded to oblivion. And perhaps the worst part of it was just how damned derivative it all was. There was zero creativity over any of it, zero diversity of artistic style, zero attempt to do something aesthetically new, just pure hormone-driven churn, generally either straightforward-photographic style, anime style, or a Midjourney-ish digital art style.

So I rather have to disagree with this statement. Porn users are a major share of AI image generation, at least with run-it-yourself models (as most commercial models ban it).

I mean, you're not wrong ;)

It's like the whole "metaverse" thing of trying to reinvent shopping in a 3d gamelike environment. Even in video games it's common to not have to actually hunt for and pick items off the shelves in 3d, instead just choosing from a dialog. Just because something sounds shiny on paper doesn't actually mean it's good in the real world.

You don't need to see the person on the other side to chat with them. You don't even need to hear them in most cases - text is just fine, and actually helps prevent misunderstandings.

Jake Sully.