Re: Provisions to the contribution guidelines for using LLM generated code

From: Alexander Leidinger <Alexander_at_Leidinger.net>
Date: Fri, 31 Jan 2025 22:37:42 UTC
Am 2025-01-31 22:21, schrieb Sulev-Madis Silber:

> so unless we get actual agi, we can't use this. and if this is agi, it 
> might understand gpl vs bsd too. and either obey  or not obey it. then 
> we could treat it like human, either accept or deny the code. tho, one 
> would wonder, if we actually manage to beat human brain, would the 
> legal issues in fbsd project be the biggest issue?

Be careful about the wishful thinking about AGI... keep in mind that 
what we have right now is just "glorified pattern matching", just 
machine learning. This has nothing to do with intelligence (or the lack 
of it) in terms of what you use this term for with humans. Any LLM or 
whatever is matching a pattern (your input) against some abstract data 
pool and spits out a result which matches with this pattern the closest 
(the actual tech is more complex than that, but in the end it is just 
"glorified pattern matching"). Impressive, yes, but still, this is not 
what you talk about if you say some kind of human is intelligent or not.

So yes, at some point in time there is probably someone who takes a 
classification of the license of the code into account for the stuff we 
call AI, and a human being can act on this (I'm sure some lawyers will 
make a lot of money until we reach a point where it is easy to handle 
the licensing part... and then it doesn't end, as there is still the 
question about patents), but so far humanity has not produced any 
software which can be called intelligent like a human, or understand 
something like a human. We are very far from that (one of the very far 
points is, that we are not really able to define what intelligence is, 
it's not the number of neurons, and don't compare bio neurons with 
computer neurons, they use the same words, but one computer neuron is 
able to do only a fraction of what a bio neuron is able to do).

Bye,
Alexander.

-- 
http://www.Leidinger.net Alexander@Leidinger.net: PGP 0x8F31830F9F2772BF
http://www.FreeBSD.org    netchild@FreeBSD.org  : PGP 0x8F31830F9F2772BF