"Good enough" is not the goal. It's the rationalization for doing whatever benefits Trump and his supporters. This move is entirely about being able to give government money to Musk. Fuel efficiency and electrification are about protecting fossil fuel companies. Wages is about benefitting large companies.

Whatever Trump wants, his administration comes up with a rationalization for it. Often it's a ludicrous one. "Favoring superior technologies over inferior ones is unfair to the inferior ones!" It doesn't matter. No one needs to believe it. They just need to pretend they have a reason.

Suppose a company decided to pirate all the software they used instead of paying for it. If they tried to argue in court that they weren't violating the copyrights because they weren't distributing the software, only using it themselves, they would be ridiculed.

Copyright infringement does not require distribution, only copying.

They aren't even vaguely equivalent. One is commercial, the other is non-commercial. One competes in the market with them, the other doesn't. If the kid decides to give a concert, charge admission, and not pay royalties, that would not be fair use.

Fair use is based on four factors.

1. Is the copying for commercial purposes? Commercial use is less likely to be fair use than non-commercial use. The AI companies fail on this one.

2. How much are they copying? Copying just a small piece of a work is usually fair use, while copying a larger amount is less likely to be. The AI companies are copying tens of thousands of songs in their entirety. They fail on this one.

3. Does it compete in the marketplace with the original work? AI companies have signed contracts with lots of other companies to pay large amounts for the rights to train on content. There's a large market for training rights. Pirating the works instead of paying for them competes with the plaintifs. They fail on this one.

4. Is the use transformative or just plain copying? In this case, it's just plain copying. I'm not talking about the models they produce (there's an argument about whether they infringe the copyright) or the outputs of the models (which also arguably infringe the copyright). I'm just talking about the raw data used to train on. They copy it directly because they don't want to pay for it. They fail on this one.

Since they fail on every one of the four criteria, there just is no credible argument that it's fair use.

We need to distinguish between the theoretical effects that AI might have in the future, and the real effects it is actually having right now.

First the theoretical effects. On one hand the optimists promise to solve all the world's problems, from disease to poverty to climate change. On the other hand the pessimists warn it could destroy humanity. There are credible arguments on both sides and we should take both possibilities seriously, but also remember they're speculative and may not happen.

If a technology has a credible chance of solving the world's biggest problems, but also a credible chance of destroying humanity, should we really be rushing ahead to develop and deploy it as fast as possible? Or should we move more slowly and cautiously?

Next the real effects it's actually having right now. It's become clear that for most people, those are overwhelmingly negative. From the slop that's flooding the internet and drowning out legitimate content, to layoffs, to rising electricity prices, to skyrocketing prices for memory and storage. These things affect people in practical ways every day.

It's no wonder public opinion is increasingly negative about AI. That threatens the CEOs who've bet their companies on it. Of course they're trying to push a better narrative that serves their interests. And maybe the negative real world effects are invisible to them. It's not like they have to worry about getting layed off, or like it matters to them if their next phone costs 30% more than the last one.

That quote is an unintentionally ironic illustration of Jevons paradox. Apparently the goal of electrifying airplanes is not to reduce their impact on the planet. Instead his goal is to increase the amount of air travel, which would likely end up increasing the total energy used for it, not decreasing it.

Congress did not pass a law giving them authority over this. The Trump administration routinely claims to have powers that the law doesn't give them, knowing most people will never look it up and will believe their false claims.

Here, I looked it up for you. This is how the law defines a "swap", which is what the CFTC has authority over. It's pretty dense and includes several different types of contracts, but here's the one that's relevant to Kalshi (emphasis added by me):

The part I marked in bold is critical. That's what distinguishes financial instruments from gambling. It must depend on something of financial, economic, or commercial consequence. If you buy an insurance policy that pays you if your house burns down, that's a swap. Your house burning down has financial consequence to you. If you bet on who will win a football game, that's not a swap. The football game has no financial consequence to you. It's pure gambling.

The problem for Kalshi is that they offer both types of contracts and don't distinguish between them, but the law does distinguish. Some of their contracts fall under the CFTC's authority, but a lot of them are firmly outside it. Those contracts are subject to state gambling laws that they blatantly ignore.

Here is the actual data on state government debt. California's per-capita debt is $2833, which is very close to the national average of $2637. States cover a huge range from Idaho with only $356 to Alaska with $9418. California is right in the middle of the pack.

Comparing that to the US federal government's debt ($114,136 per person!) is absurd. Even Alaska's debt is tiny in comparison.

This is absurd on so many levels.

To start with, it's not like the lawsuit has anything to do with what state they're based in. The California AG happens to be leading it, but it's a coalition of 12 states. It's not as if they could escape the suit by moving to another state. This is purely a pressure tactic, nothing more.

Actually moving would be incredibly expensive. Build entirely new studios. Relocate all their employees. Lose a whole lot of employees who don't want to relocate.

And anywhere else they moved to would be a worse place. They're located in Los Angeles because that's the heart of the film industry. That's where the workers are. That's where all the other companies that support the industry are. Trying to operate in Georgia or Tennessee would be much harder for them.

Even if they somehow decided it made sense to move, that's not a decision you make quickly or lightly. It would take long and careful planning to choose a new location and prepare for the move. But here he is, saying they're going to start moving in less than two months, and he hasn't even decided what state to move to. Is anyone really supposed to take this seriously?

If in spite of everything he really tries to go through with this, expect a shareholder lawsuit and calls for the board to fire him. But he'd have to be crazy to really do it.

Wikipedia has a pretty good discussion of this. If you burn them at sufficiently high temperature, you break down almost all the toxins into simpler molecules that are either non-toxic or easy to neutralize with standard processing. An exception is heavy metals, but those are also easy to filter out and capture.

1 in 3 the alert is wrong. 1 in 1,000,000 the officer acknowledges the possibility that it might be wrong.

Also, 58% of all statistics are just made up, including both the 1 in 1,000,000 statistic he made up and the 58% statistic I just made up.

To burn hydrocarbons and change the climate. Obviously!

You mean like old fashioned code that just does something without having an LLM to hallucinate the result? That is so 2022.

He kind of answered his own question. Kids move between schools all the time. People move between companies all the time. Either you dictate that everyone in the world must use the same environment, or else people will sometimes have to change environments.

Sure there's a cost to switching. You shouldn't do it without a good reason. You also shouldn't be afraid to do it when there is a good reason. You're trading a short term cost for a long term benefit.

There also is value in knowing multiple platforms. I didn't like it when I was at a company that had standardized on Windows for everything, but there have been lots of times since then when it was useful to have learned Windows. There's educational value in occasionally being forced to learn a new platform.

The parallel to the Luddites is striking. They were labor activists who campaigned for a living wage and an end to child labor. Contrary to what some people think, they didn't object to technology. That was just a lie spread about them by their opponents. But they thought technology should be used to benefit everyone, not to enrich a few people at the expense of the many.

The situation today is remarkably similar. Few people object to AI on principle. But a lot of people object to the way it's being used to enrich a few at the expense of the many.

You're confusing mainstream nutrition advice with fad diets. The Mediterranean diet was proposed in 1975, and by the 1980s it was already widely embraced by mainstream nutritionists as the recommended way to eat. In the 40 years since then, that has never changed. The evidence for it has only gotten stronger with time.

Meanwhile, countless fad diets have come along, each based on some fringe theory and unsupported by solid evidence. And every time that happens, the mainstream nutritionists say, "Please ignore them. Eat a Mediterranean diet." And people like you say, "What's wrong with these scientists? Why do they keep recommending new diets that are different from the old ones?"

So it's turtles all the way down?

Just get another AI to review the code generated by the first one. Because if you don't trust the first model to understand the requirements and implement them correctly, surely you can still trust it to understand the requirements and determine whether another model implemented them correctly! If you want to understand how the code works, just ask the AI to explain it to you. No need to study the code to determine how well the explanation matches what's actually implemented. And while you're at it, no need to write any tests yourself. Just get the AI to write your tests for you. And you don't need to study them to make sure they're testing the right things. Just get another AI to check them for you. Then you know everything is correct!