Will FreeBSD adopt a No-AI policy or such?

This one must make you feel silly, given that facebook is one of the top 5 companies int he world by market capitalization. Wait, you wrote that after all that happened?!
Yep, Meta is also a leader in VR and LLMs. Basically *all* the buzzwords. You might say it profits from convincing plebs that these products are important.

Though looking at the stats, the Facebook product has declined, as has the social media space in general and VR is continually a non-starter due to over-monetisation attempts.
 
Though looking at the stats, the Facebook product has declined, as has the social media space in general and VR is continually a non-starter to to over-monetisation attempts.
Facebook recently posted that 10% or 20% of their income is from scam postings. The ONLY reason I ever got a Facebook account is some relatives of mine at a funeral told me they were all on there. Since then, I rarely see any postings on Facebook by anyone I know including those relatives but lots and lots of advertising, scams and soft porn.
 
LLM means Large Language Model and is hardly an invented marketing term.

If you say so.

---

Are you gentlemen saying that because you find facebook distasteful, that that means that "social network media" or whatever they call it was a gimmick of no consequence?

Because, as far as I can tell, it reorganized society globaly to its core.

---

Not much else to respond to here. Basically a whole lot of "nuh UH!"
 
While I'm here, some more thoughts:

There's definitely some cosmic justice to the fact that the class of people being wiped out of the labour market right now is the very same that so gleefully exclaimed "learn to code!"

Dear managerial class: eat cake.

And a question here for cracauer@: if you had done that C to Rust conversion by hand, how long would it have taken you? How long did it take the gimmick?
 
I have not written substantial Rust code by hand, so I wouldn't know.

Gemeni 3 Pro (High) spent quite a bit of time on the conversion. I think a minute or so?

That a minute or so is "quite a bit of time" in your mind for autogenerated software to produce fully functional code says really anything that needs to be said about what you tell yourself autogenerated software's capabilities are, and what you really estimate them as.

Let's say it might take your average Rust programmer an hour 20 minutes? Is that 100x?

Why not take you, a life long systems engineer withe professional design experience in several languages and not insignificant familiarity with Rust. What would be the multiplier there?

Curiosity.

---

And that's just the idiot "Larry your LLM companion" chatbot.
 
I was once known in my hometown as the programmer who could program as fast as he can type, but I really didn't write nontrivial Rust programs by hand yet. I just don't know whether it is any slower than -say- C.

One thing is for sure - if you want to learn Rust from scratch this thing would be of help. Especially when it comes to finding the incantations for various library calls. It also generated the cargo stuff.
 
Might be an interesting experiment to try, no? Specially know that you've already glanced at a computer solution.

I mean if it won't take more than 5 mins, why not? 15 mins? 30?

Way more. This is the C++/POSIX version:

One thing that Gemini handled well is that it knew about getopt(2) and translated it into neat Rust library use. I would have to look this up and learn the library.

But I don't see that we are in a race of producing code in minutes. Productivity in programming is not coding as fast as you can type.
 
Way more. This is the C++/POSIX version:

One thing that Gemini handled well is that it knew about getopt(2) and translated it into neat Rust library use. I would have to look this up and learn the library.

But I don't see that we are in a race of producing code in minutes. Productivity in programming is not coding as fast as you can type.

I don't know. A machine that can translate a C program to Rust in a minute where it would take an expert like you who knows how long including research sounds pretty productive to me.
 
I don't know. A machine that can translate a C program to Rust in a minute where it would take an expert like you God knows how long including research sounds pretty fucking productive to me.

Well, yes, but what happens afterwards? My attempts to change programs with the LLMs (as opposed to writing from scratch) were miserable.
 
Well, yes,

This is a start.

but what happens afterwards?

Maybe the program was designed to produce code from scratch instead of waste time on a dumb bag of meat that would take days to produce an original anyway. I don't know just making wild leaps here as to what the industry might consider "productive."

Your question is not a benchmark of efficency or power. We established that it can do in 1 minute something that would take you X. That's an objective benchmark. The other thing could have all sorts of reasons.

It's called "gutter talk" and we prefer to have a modicum of decorum here so we don't stoop to that level.

Oh yeah sorry let me go to the "enshitification" thread to get all this hoodlum speak out of my ears.
 
Sure, but how does that help a group being more productive?

Most software projects of reasonable complexity have all parts already re-written even before the first release. Initially hacking some code in is only a fraction of the work.
 
Sure, but how does that help a group being more productive?

Most software projects of reasonable complexity have all parts already re-written even before the first release. Initially hacking some code in is only a fraction of the work.

Maybe you are asking a Ferrari production plant to change a flat. Maybe there's a tire spot down the street that does those also 1000x humans. I don't know.
 
Here's a more relevant question: in a project big enough that it would not be trivial for you to look over every single line of computer generated code, imagining a situation in which it comples and runs, did you look over every line?

If you didn't, could certain judgement calls about how such a complex system should be built be designed into the machine? Would it then put those desicions into production without you even noticing they were made?

Real life questions.
 
Well, LLMs have limited context windows. They cannot make or change large codebase yet.

I'd say I would review the tests firsts and ask the thing to make more edge case tests. Whether I review the code line by line depends on who's signing my paycheck.
 
Back
Top