Hacker News new | ask | show | jobs
by raincole 123 days ago
Since the birth of ChatGPT, people have been talking about if one day LLMs will be trained to write bytecode or even machine code directly, making future code incomprehensible for humans.

It'd be funny if it ends up being just C++35.
4 comments
taminka 123 days ago
this is silly, we already have an algorithm for generating very efficient assembly/machine code from source code, this is like saying maybe one day llms will be able to replace sin() or an os kernel (vaguely remember someone prominent claiming this absurdity), like yes, maybe it could, but it will be super slow and inefficient, we already know a very (most?) efficient algorithm, what are we doing?
link
ainiriand 123 days ago
I would love to see the first LLM shot in the foot with C++. I mean it is not like human devs don't do that right?
link
ksherlock 123 days ago
Maybe that's why AI images have all those extra fingers and toes!
link
FuckButtons 123 days ago
I mean, you can see llms shooting themselves in the foot with c++ right now, just ask them to write it.
link
ainiriand 121 days ago
Crowdstrike level shot in the foot.
link
pjmlp 123 days ago
It is hardly any different from,

C Source code => Tradicional UNIX C compiler => ASM => object file

Now everyone is doing

AI tooling => C Source code => Tradicional UNIX C compiler => ASM => object file

For all pratical purposes, just like using a language like Nim, the workflow exposed to user can hide the middle steps.

Then there is the other take, if you start using agents that can be configured to do tool calling, it is hardly any different from low code applications, doing REST/GraphQL/gRPC calls orchestrated via flow charts, which is exactly what iPaSS tooling are offering nowadays, like Workato, Boomi,...

link
bestouff 122 days ago
> Now everyone is doing > > AI tooling => C Source code => Tradicional UNIX C compiler => ASM => object file

Certainly not. I'm using AI to write Rust, the compiler is way better at telling the AI it goofed completely, and the resulting code (once the AI managed to get rid of errors/warnings) has a much better chance of actually working well.

link
pjmlp 122 days ago
The use of C was only an example, and I can bet that AI can also goof Rust code that goes through the compiler if that is your argument.

Unless you now tell me that you drive your AI generation code with full coverage unit tests manually written by you.

link
cbm-vic-20 123 days ago
Or, like the Shakespeare programming language.

https://shakespearelang.com/

link