95% of the world’s new coal construction (2023)

China had the largest new coal construction in 2023 but it was far below 95%. I didn’t do all the math but it drops below 50% when you compare it to just the growth of the next three biggest coal producers.

They build most of our solar but we’ve effectively banned it now. They’re not only growing capacity to produce renewables, they’re taking the outputs that were planned for sale here and installing them locally.

Yes. And go check the percentage of coal use over time. Coal is going up. Renewables are going up much faster.

You should be pretty happy with China then. They have a replacement rate just over one. That’s lower than the US or Europe.

This has been going on for years and will continue.

China really really really needs a robust and diverse energy infrastructure. Industry needs huge amounts of energy. AI needs huge amounts of energy. The military needs huge amounts of energy.

Coal is unreliable and dirty. Oil can be blocked at the Straight of Malacca and a few pipelines.

China is also the world’s factory. They own the entire logistics chain for producing renewable generators; from raw materials to final assembly. They have all the infrastructure to not only build solar panels and wind turbines at scale, they’ve scaled up building the machines that build them.

Pollution per GDP is a better measure. https://ourworldindata.org/grapher/co2-intensity Pollution per GNP would be even better but I can’t find it.

Individuals don’t pollution much, it’s mostly industry. Really poor countries often don’t pollution much because they can’t afford to. Sometimes they pollute prodigiously because the only thing they can afford to do is destructive resource extraction. Rich countries can often outsource their pollution to poorer countries.

China has been making mind boggling investments in renewables. They have been expanding all their energy sources but their renewables have the lions share of the growth.

They’ve been building roads and all kinds of infrastructure. That’s what the BRI is all about, even if they’re being a bit quieter about saying the phrase. They like to build their long haul roads on elevated columns; not only because it’s less disruptive to wildlife but because it lets them use giant road laying robots to place prefab highway segments.

They dropped the one-child policy a while back but they’re having some trouble getting people to have more babies. That said, there’s some research that suggests that rural populations around the world are severely undercounted, so they may have a bunch more subsistence farmers than they, or anyone else, realizes.

So you’re confident that human learning involves “understanding” which is distinct from “statistical optimization”. Is this something you feel in your soul or can you define the difference?

Human learning requires understanding, which AI is not capable of.

How could anyone know this?

Is there some test of understanding that humans can pass and AIs can’t? And if there are humans who can’t pass it, do we consider then unintelligent?

We don’t even need to set the bar that high. Is there some definition of “understanding” that humans meet and AIs don’t?

You’re correct that a collection of deterministic elements will produce a deterministic result.

LLMs produce a probability distribution of next tokens and then randomly select one of them. That’s where the non-determinism enters the system. Even if you set the temperature to 0 you’re going to get some randomness. The GPU can round two different real numbers to the same floating point representation. When that happens, it’s a hardware-level coin toss on which token gets selected.

You can test this empirically. Set the temperature to 0 and ask it, “give me a random number”. You’ll rarely get the same number twice in a row, no matter how similar you try to make the starting conditions.

You may be correct but we don’t really know how humans learn.

There’s a ton of research on it and a lot of theories but no clear answers.
There’s general agreement that the brain is a bunch of neurons; there are no convincing ideas on how consciousness arises from that mass of neurons.
The brain also has a bunch of chemicals that affect neural processing; there are no convincing ideas on how that gets you consciousness either.

We modeled perceptrons after neurons and we’ve been working to make them more like neurons. They don’t have any obvious capabilities that perceptrons don’t have.

That’s the big problem with any claim that “AI doesn’t do X like a person”; since we don’t know how people do it we can neither verify nor refute that claim.

There’s more to AI than just being non-deterministic. Anything that’s too deterministic definitely isn’t an intelligence though; natural or artificial. Video compression algorithms are definitely very far removed from AI.

That’s a reasonable critique.

The point is that it’s trivial to come up with new words. Put that same prompt into a bunch of different LLMs and you’ll get a bunch of different words. Some of them may exist somewhere that don’t exist. There are simple rules for combining words that are so simple that children play them as games.

The LLM doesn’t actually even recognize “words” it recognizes tokens which are typically parts of words. It usually avoids random combinations of those but you can easily get it to do so, if you want.

That’s possible but it’s not what the authors found.

They spend a fair amount of the conclusion emphasizing how exploratory and ambiguous their findings are. The researchers themselves are very careful to point out that this is not a smoking gun.

I’d say there are two issues with it.

FIrst, it’s a very new article with only 3 citations. The authors seem like serious researchers but the paper itself is still in the, “hot off the presses” stage and wouldn’t qualify as “proven” yet.

It also doesn’t exactly say that books are copies. It says that in some models, it’s possible to extract some portions of some texts. They cite “1984” and “Harry Potter” as two books that can be extracted almost entirely, under some circumstances. They also find that, in general, extraction rates are below 1%.

Sort of.

If you violated laws in obtaining the book (eg stole or downloaded it without permission) it’s illegal and you’ve already violated the law, no matter what you do after that.

If you obtain the book legally you can do whatever you want with that book, by the first sale doctrine. If you want to redistribute the book, you need the proper license. You don’t need any licensing to create a derivative work. That work has to be “sufficiently transformed” in order to pass.

That’s my understanding too. If you obtained them legally, you can use them the same way anyone else who obtained them legally could use them.

prompt: Make up a word that is not found anywhere on the internet. If I do a search for that word, it should not exist. Provide example usage.

Qwen3-235B-A22B:

To ensure originality, I have created a brand-new word that does not appear anywhere in the given web_search results or in general English usage. Virethra (pronounced: vye-RETH-ruh)

Part of speech: noun

Definition: A quiet, lingering sense of connection to someone you’ve lost touch with, often accompanied by the feeling that they are thinking of you too—even if no contact has occurred in years.

Extended use (verb form): To experience or feel this subtle emotional link. Example Sentences:

Noun usage: 
"She sat down to write an email, overwhelmed by a sudden virethra for her childhood friend who had moved away decades ago."  

Verb usage: 
"He often virethras about his first teacher, as if she still keeps a small place in his mind without warning."  

Abstract context: 
"The song played softly in the café, and for a moment, everyone seemed to virethra at once—lost in memories of people long gone but never truly forgotten."  

They seem pretty different to me.

Video compression developers go through a lot of effort to make them deterministic. We don’t necessarily care that a particular video stream compresses to a particular bit sequence but we very much care that the resulting decompression gets you as close to the original as possible.

AIs will rarely produce exact replicas of anything. They synthesize outputs from heterogeneous training data. That sounds like learning to me.

The one area where there’s some similarity is dimensionality reduction. Its technically a form of compression, since it makes your files smaller. It would also be an extremely expensive way to get extremely bad compression. It would take orders of magnitude more hardware resources and the images are likely to be unrecognizable.

If you want to go to the extreme: delete first copy.

You can; as I understand it, the only legal requirement is that you only use one copy at a time.

ie. I can give my book to a friend after I’m done reading it; I can make a copy of a book and keep them at home and at the office and switch off between reading them; I’m not allowed to make a copy of the book hand one to a friend and then both of us read it at the same time.

That’s not what it says.

Neither you nor an AI is allowed to take a book without authorization; that includes downloading and stealing it. That has nothing to do with plagiarism; it’s just theft.

Assuming that the book has been legally obtained, both you and an AI are allowed to read that book, learn from it, and use the knowledge you obtained.

Both you and the AI need to follow existing copyright laws and licensing when it comes to redistributing that work.

“Plagiarism” is the act of claiming someone else’s work as your own and it’s orthogonal to the use of AI. If you ask either a human or an AI to produce an essay on the philosophy surrounding suicide, you’re fairly likely to include some Shakespeare quotes. It’s only plagiarism if you or the AI fail to provide attribution.