Khan did not throw at you a 100-slide Powerpoint deck in 45'.

He really took the time to replicate the manual teaching process of writing on whiteboard. He improved upon in using colors. But basically had the same pace as a teacher writing on a whiteboard.

When professors are given a projector, the just throw together some slides and add their narration.

This is not very efficient. To learn you need to suffer. Or you need to watch the suffering.

offsetting in what horizon? I give you 100 in q4 2026 you give me 100 in q1 2027

They're already ahead of you; you have to consistently book revenue (accrual or cash basis) which means they both go at the same time (which would offset) or that real money is being exchanged. You can't accrue the 100 you're (supposedly) giving me now and THEN accrue the 100 I'm giving you next year.

You could but the other guy would have to book 100 of goodwill in the interim, matched by me booking goodwill later, and that brings it's own problems

Goodwill almost always raises concern with authorities and audits, so I'd imagine so sort of quid pro quo version is equivalent to loudly yelling to be audited!

It’s down already. The fund exceeded its capacity.

If money is what you are after the PhD offers one of the worst (effort / probability of becoming rich) ratio .

"Money" is subjective. Professors are paid a pittance (barely middle-class in some areas), are no longer generally respected, and are abused by their students and leadership at rates that have never been seen. There aren't many pros to academia.

Oh no, the workers have power!

My first thought. I was browsing comments to see if everyone from the US did their mandatory bootlicking and yes, they did. Of course they did.

People are weird. Their government is strongarming half the world at the moment and they do not pause and go "wait, does this mean that if we unionize we can threaten to wipe all the databases unless?"

Also we are past that. Now IQ started decreasing.

it's hard to separate IQ decreasing and return to mean with IQ stabilizing

in 20th century most of the world moved past famine and toxins - did any factor of similar scale happen in 21st century as well to start looking for opposite processes?

Generally that statistic refers to populations in isolation, not the entire world in aggregate.

It is fairly well agreed upon that American kids across the nation are currently testing lower than they were in 2010.

So is this effectively a way for aws customers to more easily access Claude code?

Claude itself was almost from the very beginning available in bedrock.

Might be for enterprise customers (with slow procurement departments) who already have AWS accounts but not Anthropic ones.

So then they could use Bedrock, no?

A lot of tools, community and Anthropic’s, have zero support for Bedrock.

Anthropic’s offerings for Bedrock lag behind their main platform by months, maybe up to a year or more.

Bedrock is both more expensive, less feature complete, and less reliable in terms of raw volume of 500 errors.

A lot of really good reasons:

1) Higher level code is easier for LLMs to review and iterate upon. The more the intent is clear from the code, the easier it is for humans and LLMs to work with.

2) LLMs get stuck or fail to solve a problem sometimes. It is preferable to have artifacts that humans can grok without the massive extra effort of parsing out assembly code.

3) Assembly code varies massively across targets. We want provable, deterministic transformation from the intent (specified in a higher level language) to the target assembly language. LLMs can't reliably output many artifacts for different platforms that behave the same.

4) Hopefully, we are still reviewing the code output by LLMs to some extent.

I'd add to that

1.5) Having a compiler in the loop that does things like enforcing type constraints (and in the case if Rust in particular, therefore memory safety guarantees) is really useful both for humans and LLMs.

In addition LLMs also make bugs, and debugging assembler is more difficult, wasting more tokens, thus more money.

A very big practical reason is also that assembler code would eat context like no other.

> 1) Higher level code is easier for LLMs to review and iterate upon. The more the intent is clear from the code, the easier it is for humans and LLMs to work with.

The counter-argument, and one that matches my experience is working at a lower level is actually beneficial for LLMs since they can see the whole picture and don’t have to guess at abstractions.

Feel free to post a project of yours where you gave a bunch of prompts to an LLM and it produced a working application written in assembly without you having to check for anything

Is this a serious question or are you just trolling?

I'll bite:

Programming languages are tools for thinking. It's not clear that assembly code has the right abstractions to encourage the kind of thinking that programming large systems requires. After all, human intelligence found assembly insufficient and went on to invent better languages for thinking, why should artificial intelligence, trained on human intelligence, be any different? Maybe AI in the future will have its own languages for thinking, but assembly is likely not that.

I get what you mean but I think if anything AI pairs extremely well with strongly typed languages that are at times cumbersome for humans, but decrease the latency at which AI can get feedback on its code. In my (very) limited experience Rust is an excellent target for AI codegen.

(I meant statically typed / high level of type safety here not strongly typed)

This is a Rust to CUDA converter so I guess it is for codes where the programmer wants it to function properly (Rust) and have good performance (CUDA).

It’s just a matter of different workflows for different users and application.

Because when this idiotic hypemachinery finally dies an agonising, painful death, some of us still want to work with computers

I mean, AI is not good at writing x86-64 assembly code. Last time I tried (with both Claude and ChatGPT), the AI failed to even create basic programs other than Hello World.

I believe that I have learned cumulatively double in the past 10 years from YouTube compared to what I learned in 6 years of middle and high school. And I don’t spend 8 hours per day on YouTube.

Plumbing, react, combinatorics, real analysis, python, c++, cad, micro and macro economics, reinforcement learning, to name just a few of the things I learned through YouTube.

We don’t give enough credit to what we take for granted today.