horsh1

Well, if we count claude as tui.

Like Shanghai

In April

If programming in the US to become unconditionally 10x more expensive, then the exodus from the US is about to begin.

Because these are completely different requirements.

Enclosure of XXI century

Instead of useless stock photos we now have useless AI generated photos.

The hikers move from tin canned tushonka to plastic pack like these ones https://yandex.ru/images/search?from=tabbar&text=%D0%BA%D1%8...

This article positions as something invented between the USA and the USSR... But, just looking into wikiepdia:

"One such solution was a method proposed in 1809 by the French pastry chef Nicolas François Appert, which involved long-cooking meat or vegetables (approximately 1-2 hours) and pasteurizing the finished product in a brine solution. Appert received a personal award for this invention from Napoleon."

So why would they bury a man with a book?

My typewriter has been successfully serviced 45 years after being produced.

You are describing a typical state of a wibecoded project.

The problem is that organizations providing infrastructure (such as message exchange, money exchange, physical entities exchange) are allowed by law to manipulate the stream, heavily advertise, provide credit etc all kind of scum. Depriving children from writing a message to parents and friends is nonsense. Exposing them to these for-profit organizations is questionable. But that is also questionable with the grown ups.

Comparing the same article in different languages sometimes gets very educational.

So what countries will be the beneficiaries of this process?

How many of them have mobile or internet connection?

Why in this story students use GPT and professors don't?

If you are an expert -- sit down and start working with GPT on your own. See what it can and what it can not do. See where it helps. See where it hands down lose.

Unless we intend to surrender everything about human symbolic manipulations (all math, all proving, all computations, all programming) to llm in the nearest future, we still need some formal representations for engineering.

The major part of tradidional NLP was about formal representations. We are still to see the efficient mining techniques to extract the formal representations and analyses back from LLM.

How would we solve the traditional NLP problems, such as, for example, formalization of law corpus of a given country with LLM?

As an approximation we can look at non-natural language processing, e.g. compiler technologies. How do we write an optimizing compiler on LLM technologies? How do we ensure stability, correctness and price?

In a sence, the traditional NLP field has just doubled, not died. In addition to humans as language capable entities, who can not really explain how they use the language, we now also have LLM as another kind of language capable entities. Who in fact also can not explain anything. The only benefit is that it is cheaper to ask LLM the same question a million of times that a human.

You have "Rows" repeated eight times in this short snippet. You have all the rest of your variables as two letter names. It feels low level, almost like Fortran.

Latin should be okay.

Writing any documents with Word is a sin.

50 years is a long time.

Between the existence of Austria-Hungary empire, when the Latin alphabet was very relevant for all the Balkans, and the end of WW2, when the Cyrillic became much more relevant for them is just 1945-1918=27 years.

Why Latin? At the moment of EU dissolution, there will be an easy choice of switching to Arabic vs staying with Cyrillic.

Yeah, a pulp fiction par exellence. Very good when you are 11 years old. Why on HN?

6. Go to whosampled.com and figure out the originals. Whom do they cover, whom do they sample.

As with the majority of the recent search startups: no hiragana, no hebrew, no cyrillic.

"File not found. Apologies are made". "The computer is being updated. Grief and sorrow."

... yet.

So with the current state of technologies we can answer all kinds of questions regarding a sentence, and an llm can even invent questions to all different words in the sentence. And we can translate between hundreds of languages live and dead. And we can even ask llm to produce a parse for a sentence.

But we just can not solve the question of mass translating sentences to their AST.

It amazes me. I really hope someone will step up and tell: sure it is possible as a cheap byproduct of the transformer technology, just do this and that.

For a systems language execution environment there is a huge list of crucial questions: what is a library, what is a package, how does linking works, how do you interact with different build systems, how you separately compile having only interface declarations, how to manage alternative libraries for the same interface, how to tweak compilation for multiple targets, how do you ffi?

The answers to all of the above will have an impact on your language features: how you import, how you declare foreign entities, what visibilities are possible, what data storage classes are theere, what is the module initialization order, what code can be executed during module initialization, what is your conditional compilation mechanisms?