canjobear

People at Microsoft Research certainly do. But the bet is that, with scale, these limitations won't matter in practice. For example Transformers can't recognize well-nested brackets (like {[()]}) to infinite depth; the depth that Transformers can recognize is limited by how many self-attention layers they have. But in practice, you rarely need much depth.

Check out Sean Carroll's recent books if you want an introduction to these ideas that conveys the actual mathematical content and not just a word jumble.

Interestingly Chinese did have things like this at the time the writing system formed, and they simply weren't written. You can see remnants of the old morphological system today in characters that alternate in tone.

For example there used to be a suffix -s that turned a verb into a noun. During the Han Dynasty this turned into -h, and then later turned into the falling/departing/fourth tone. Hence 教 jiāo is the verb "to teach", from Old Chinese *kraw, and 教 jiào is the noun "teaching" from Old Chinese *kraws.

Also -们 as a pluralizer on pronouns and animate nouns.

What the heck is this accent of German?

There were various attempts to develop an organic phonetic writing system for Chinese, like hiragana for Japanese, for example Bopomofo (still used in Taiwan) and General Chinese (https://en.wikipedia.org/wiki/General_Chinese). The Simplified characters that you see on the mainland today were originally part of a multi-phase scheme to eventually replace characters altogether, but the second phase (https://en.wikipedia.org/wiki/Second_round_of_simplified_Chi...) was bungled so badly that it didn't continue. In practice Pinyin is the standard phonetic writing now and is used when people can't remember a character.

There are sporadic attempts at making things more dynamic, like the Neural Turing Machine. It doesn’t seem to buy much actual power.

It’s more than a coincidental resemblance. Programming languages were designed using constructs from linguistics (context-free grammars, parsers) that were originally developed to describe natural languages.

Neural network representation spaces seem to converge, regardless of architecture: https://arxiv.org/abs/2405.07987

It would make sense for the human mental latent spaces to also converge. The reason is that the latent space exists to model the environment, which is largely shared among humans.

Are children more traumatized now than in the past? Seems like the past was harder on children.

> Aren't all animals sentient?

How far are you willing to go with this? Is a roundworm sentient? An amoeba? A bacterium?

If the brain can get any extra fitness from exploiting quantum effects, it will. But the reason people are interested in this stuff seems to be that they think there's a connection between the quantum nature of the brain and consciousness. I don't see how the argument is any more than this:

1. Quantum effects seem mysterious.

2. Consciousness seems mysterious.

3. Therefore, quantum effects cause/are consciousness.

...which seems pretty weak.

The topic paper is part of a big project called Project Ceti that's aiming to do just that.

On the theoretical side for why it may be possible, see https://arxiv.org/pdf/2211.11081 https://arxiv.org/abs/2303.10931

Wittgenstein didn't present any evidence.

> EVerything more or less evolved into a sort of tron-bike shape for most of the runs I tried.

I ran it in the background for a very high mutation rate for a long time and it managed to come up with something very different---a little wheel attached to a big wheel, which bounces around and goes over all the obstacles.

The multi-sound characters are common in usage. For example 长 can be cháng meaning "long" or zhǎng meaning "to grow". 行 is xíng "to walk" or háng with no real single coherent meaning, appearing in compounds like 银行 yínháng "bank" and 行业 hángyè "profession". All of these are very common usages. In context they are essentially never ambiguous, but if you are going through character-by-character it's not going to make sense.

Chinese also has furigana-like characters that can be written next to characters, bopomofo or zhuyin fuhao https://en.wikipedia.org/wiki/Bopomofo, although only taught and used in Taiwan as far as I know.

> Since the sounds per character are 1-1 and the semantics are very clean you can get a lot out of a text by looking characters up in the Unihan database.

You probably got a lot of wrong meanings this way. The characters aren't 1-1 mapped to sounds and their semantics are profoundly context-dependent.

They aren't underpaid relative to other developed countries: https://data.oecd.org/teachers/teachers-salaries.htm

(1910)

My experience is that the quality of answers is mixed, but uniformly vastly better than Yahoo Answers.

How about just one parameter! https://pubs.aip.org/aip/adv/article/8/9/095118/22887/One-pa...

This is a live hypothesis for the Yeniseian languages, the Siberian languages that are most likely to be related to North American languages. The Yeniseian languages are very unlike other nearby languages. So it’s possible they represent a back-migration from America into Siberia.

Stephen King has always been bitter that Kubrick made a movie that was better than his book, while inverting the moral of the book.

Do you have a citation? I didn’t know the idea went back that far.

> What does "strategy" mean here?

Any function that outputs bets.

Why is this a problem? No information is lost when characters (or graphics) are encoded in binary.

The whole point of Kolmogorov complexity is that description lengths under different Turing-complete description languages (such as UTM and lambda calculus) are only different up to a constant that depends on the languages and not on the thing being described.

I always thought he passed the test there, and the guys that just mindlessly shot failed.