A software always runs on a computer.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum things get difficult when "intended behaviour" is only probabilistic.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum yes - if you write software, not so much if you grow it…
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum thus the rush to integrate it into critical systems so that the financial and/or mortal cost of shutting it down will be used to justify its owners continued crimes against society.
There is a lot of infrastructure that we could theoretically choose to turn off whose loss would be catastrophic.
-
@ploum We want it to have full access to the internet, but not exactly full and we can't be bothered to whitelist it. Also, we won't even monitor it, just leave it running and go home.
@t_var_s @ploum it's the same when you have a memory leak in a process, leave it running for a while, and the said process will "eat" the available memory and crash the whole system. who would've thought! what should we do! if it eats enough memory, we all run "out of memory" and will be "killed". the solution of course is to keep "feeding" that process with memory, otherwise it might go hungry and start eating people.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum I imagine that viruses can get out of control and spread to more computers than intended
but I'm perfectly happy to think of AI as just another kind of malware
-
@ploum The issue with (or purpose of?) calling it rouge software is that it muddles the responsibility of it’s creators.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum I partially disagree on the last part. Because shutting down a computer is only easy if it's one system in you physical reach and you're not depending on anything running on this system.
-
@ploum I partially disagree on the last part. Because shutting down a computer is only easy if it's one system in you physical reach and you're not depending on anything running on this system.
@feyter a good argument for never having anything depend on an AI-controlled system!
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum saying that an AI agent is going rogue is similar to saying that the main character in a James Bond movies its his laser watch.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum Actually, there is such a thing as Rogue software. It's a video game from the early 80s developed for unix-like systems. Many games are "like" it these days.
-
A software always runs on a computer. A software always do what its creator wants it to do.
If the software doesn’t do what its creator wanted, it is a mistake from the coder, a bug.
If the software runs without the consent of the computer’s owner, it is a malware/virus.
Stopping a software is as easy as shutting down the computer on which it is running.
Terminator and Mission Impossible are hollywood fiction, there’s no such thing as "rogue software".
@ploum I can't do more than agree. Far too many softwares and even websites can be classed as malware these days. Facebook, Instagram, Spotify, TikTok, any Google software, Windows XP SP 3 and newer (especially Windows 11), and more are the biggest malwares of them all.
-
@ploum We want it to have full access to the internet, but not exactly full and we can't be bothered to whitelist it. Also, we won't even monitor it, just leave it running and go home.
-
@ploum yes - if you write software, not so much if you grow it…
-
@ploum saying that an AI agent is going rogue is similar to saying that the main character in a James Bond movies its his laser watch.
-
J jwcph@helvede.net shared this topic
