Re: Question about: Debian puts artificial intelligence to a vote; developers will decide how to accept AI-generated or assisted code

This e-mail is written in Thunderbird, with spell and grammar checking. 
As a non-native English speaker it helps to write a text that can be 
read and understood my more people that without.

AI can help to write better code and find bugs, it also can create bugs 
and attacks.

A knife in the hands of a surgeon saves lives, in the hands of a killer 
it ends lives. Should we ban knives?

Was it better in the past, before AI? Remember Heartbleed in 2014. 
OpenSSL had this bug since 2012 and was not properly written. Not by AI 
but by humans. There are certainly more examples like that.

But we all must be aware that AI is not perfect. It helped me once in a 
while, but most of the time is produced rubbish, lied, cheated and tried 
to be humble. This week I read that somebody has put the novel Moby Dick 
in a respected AI checker. Surprise, surprise 44% was written by AI.

If Debian decides to use AI to write better code and put it in 
production after human checking and human testing, I will be happy as 
long as it is marked as generated by AI. I guess that a lot of 
non-Debian packages have already AI written code.

Op 21/08/2026 om 22:01 schreef Stefan Monnier:
>> Behind every line of AI-generated code stands a human—one who crafts the
>> prompt, sets the intent, and bears responsibility. The machine does not
>> initiate; it responds. And yes, I, too, have used AI to express myself more
>> clearly in English, so that my meaning might be understood.
> This is a very narrow view of the problem.  For many of us the
> discussion around LLMs has to go far beyond whether particular projects'
> code can be trusted to work reliably or not, or even whether chatbots'
> output are valuable.
>
> For example, a friend of mine just discovered that he has been helping
> financially OpenAI, Anthropic, and several other such companies for the
> last few years by paying for the electricity used by his home webserver
> replying dutifully to the avalanche of requests from those companies.
>
> Lots of sysadmin nowadays have to spend a significant part of their time
> fighting this constant DDoS attack on most/all web services.
>
> And it's just one of the many impacts of LLMs.
>
>
> === Stefan