Tmux has AI generated code?

Judging from what blackbird9 posted, it doesn't so much scare me as make me thing something I've said here before--that AI, or LLMs or whatever the accurate term for it is, is tempting--that is, if I have something like a .mailfilter file that isn't working, I used to take the time to troubleshoot it, testing, web searching, thinking, etc. Now it's just tempting to use claude (which *seems* with little researching to be the least unethical one) and say, here's my .mailfilter file, what's wrong with it. I suspect that most of us, regardless of good intentions, will start falling into that sooner or later. And, I'm not even sure it's bad. To give an example, in the early 2000's getting X working was often, checking various and sundry things, getting mode lines right, making sure drivers were right, etc. Now, we expect it to Just Work(TM). And I'd think most of us consider that an improvement. I'm not sure where one draws the line between convenience and self-indulgence.
 
does anyone know what portions of that commit were "assisted by ai"? I mean, could it be possible that ai 'assisted by' "changing the dates in the files" (i.e., something minor)?
 
if one puts aside the "assisted by or coauthored by AI" aspect, what is the primary problem?
To me, it comes back to "what license".
If I am writing code, I specifically put in the code "licensed under BSD 2 Clause" any code I personally write is under that license.
Is the code "assisted by or coauthored by AI" under that license because "I used AI to assist me" or is the AI code under a different AI-agent license?
 
mer said:
if one puts aside the "assisted by or coauthored by AI" aspect, what is the primary problem?
To me, it comes back to "what license".
If I am writing code, I specifically put in the code "licensed under BSD 2 Clause" any code I personally write is under that license.
Is the code "assisted by or coauthored by AI" under that license because "I used AI to assist me" or is the AI code under a different AI-agent license?
Click to expand...
Because systems that fall under the broad term of "AI" have been demonstrated to make verbatim copies of licensed code and when presented to the user they omit the license under which the code was made available.
 
  • Like
Reactions: mer
JohnK said:
does anyone know what portions of that commit were "assisted by ai"? I mean, could it be possible that ai 'assisted by' "changing the dates in the files" (i.e., something minor)?
Click to expand...
This is the article that was posted in the "valuable news"

OpenBSD and slopcode: raindrop to a torrent? – OSnews

www.osnews.com www.osnews.com
Quote
"Nearly six weeks later, and the question of whether “AI” generated code in tmux — not tool-assisted bug finding, not refactoring, actual LLM-generated slop with questionable license(1) — that was consequently merged into OpenBSD base, is considered acceptable by the lead devs, remains unanswered."

I didn't dig into it much further than reading that article.
 
scottro said:
it's just tempting to use claude
Click to expand...
The religious vocabulary seeps into this discussion because it is somewhat religious. AI is the new spawn of the devil as once was electricity. AI is a tool. Use it responsibly like any other tool. Or don't use it if you don't feel like it.
scottro said:
which *seems* with little researching to be the least unethical one
Click to expand...
That's just marketing. They're all big corporations, ergo untrustworthy and abusive.
 
JohnK said:
does anyone know what portions of that commit were "assisted by ai"? I mean, could it be possible that ai 'assisted by' "changing the dates in the files" (i.e., something minor)?
Click to expand...
How can you know? I can ask an AI to create a certain specialized function and then paste it in my code. Who is gonna know that those few lines of code were generated by an AI? I can also tell an AI to refactor some code in a certain way. Who is gonna know?

I suspect that AI's being used by many programmers who don't declare it because they use it to assist them in specific tasks. Do you have to declare that you use templates or autocompletion? I don't think so.
 
LibreQuest said:
I admire this Holwerda for earning such a prestigious reputation.

I also respect those who dream of realizing an actual AI. I am a sci Fi fan. 😄
Click to expand...
I used to like what he does on OS News. Then I made the mistake of following him on the Fediverse where he mostly shitposts and lots of times he clearly doesn't know what he's talking about.

He says that FreeBSD is fascist because there's a package for xlibre. I find the abuse of this word dangerous. In Venezuela we were called like that and a political dissident was planted a brand new copy of Mein Kampf with an obvious swastika in the cover for media effect.

He's the left-wing equivalent of Lunduke sowing division in FOSS and I find that unacceptable, so I had to unfollow him.

And then there are the "light" versions of him who say that AI is "fashy tech", which is idiotic, so I no longer dwell on the Fediverse.
 
rbranco said:
I used to like what he does on OS News. Then I made the mistake of following him on the Fediverse where he mostly shitposts and lots of times he clearly doesn't know what he's talking about.

He says that FreeBSD is fascist because there's a package for xlibre. I find the abuse of this word dangerous. In Venezuela we were called like that and a political dissident was planted a brand new Mein Kampf with an obvious swastika in the cover for media effect.

He's the left-wing equivalent of Lunduke sowing division in FOSS and I find that unacceptable, so I had to unfollow him.

And then there are the "light" versions of him saying that AI is "fashy tech", which is idiotic, so I no longer dwell in the Fediverse.
Click to expand...
Woah, well, I avoid political things all together. I certainly don't want anything to do with that. :/
 
LibreQuest said:
I also respect those who dream of realizing an actual AI. I am a sci Fi fan.
Click to expand...
Larry Niven had a story about an AI that, when powered on, rapidly learned everything that there was to be known. But before you could ask it any questions, it had realized that from now on it was only to be boredom, and it had killed itself.
 
scottro said:
I agree. I think this is a big part of it Firstly, I feel AI is an incorrect name, it's more of a REALLY big database.
Click to expand...
SirDice said:
An LLM is basically a fuzzy search engine on steroids.
Click to expand...
I'll quote my own work. I wrote this sometime last year.

Crumb 241: Embrace AI: In fearing or detesting AI, you’re becoming the grumpy, superstitious elderly person who thought electricity was the devil’s work. A tool is a tool is a tool.

Crumb 281: Largely Loquacious Motherfumblers: An LLM model is basically internet hyper-compressed into a single file plus a very clever method to search it and extract coherent fragments.

As a curiosity, yesterday I wrote Crumb 1736, which I'll now share for your pleasure.

Crumb 1736: The Drunkard: In a dive bar, a surly attractive young woman drinks alone at the counter. Another customer approaches her and asks, “Why so glum?” She looks at him, unfazed, and replies, “I used to be a good person.” “What happened?” inquires the interloper. The woman says, “I started punching people in the face and I’ve come to like it.” Now, the man asks, “Do you punch them without reason?” To which the woman replies, “No, I’m not a psychopath.” The one with the dangling urethra smiles and declares, “Well, then you aren’t a bad p…” And she punched him in the face.
 
You must log in or register to reply here.
Back
Top