While I am at it - the stuff about "changes that are not subject to copyright law" is meaningless - copyright applies to content and not to changes and I doubt there is any content on GDAL that is NOT subject to copyright law.
I would humbly suggest that the question is more about whether "corrections" such as spelling and grammar (and possible syntax and style in code) are "generative". I would suggest that the following might be better "the use of AI to make corrections such as fixing spelling, grammar, syntax and style mistakes may be acceptable" I have left the "may" in there but I would be happier with it "is acceptable" - "may" is a worrying word. On Tue, 8 Oct 2024 at 10:31, Paul Harwood <rune...@gmail.com> wrote: > IANAL - however I have looked at a lot of these things in my time. > > The phrase "use of generative AI tools to contribute copyrighted material > is prohibited" is ambiguous (and actually nugatory): > > - for clarity - it should say that it is prohibited to use such tools to > contribute *to the project* otherwise you are making a general > statement about the law. > > - the contribution of (unattributed) copyright code is (presumable) > forbidden regardless of the tools and anyway - that provides a loophole > that anyone can say "I was not contributing copyright code - I had a > reasonable lack of awareness of the status of the code" which is anyway the > grey area you are addressing. It should probably say that the use of > generative AI tools is prohibited without reference to rights of the case. > > - The user has a great deal of uncertainty about whether they are using > GAI - when using VS it makes a lot of code suggestions. I do not know > whether they are created by GAI. > > Therefore - I might suggest something more like : > > "the intentional use of generative AI tools to contribute to the GDAL > project is prohibited" > > Paul > > On Tue, 8 Oct 2024 at 08:42, Laurențiu Nicola via gdal-dev < > gdal-dev@lists.osgeo.org> wrote: > >> Hi, >> >> > At the moment I am using copilot as a cut-and-paste on steroids and it >> > has improved my efficiency, to me it is no different than copying and >> > pasting from another file in the GDAL source code or going to SO to >> > search how to write a complex recursive lambda implementation in C++11, >> > cut the example and paste in GDAL. Note that in both cases the >> > copied/pasted or generated code will need careful review and >> adaptations. >> >> I think it's worth mentioning that code taken from SO is licensed under >> CC-BY-SA, so it needs at least proper attribution. Looking at the GDAL >> codebase, I counted about 21 stackoverflow.com URLs, the vast majority >> of those added by Even, according to git blame. So either Even is almost >> the only GDAL contributor that uses Stack Overflow, or this might already >> be a problem. >> >> Not to mention that a link in the source code might not be sufficient for >> attribution, especially if you're shipping binaries etc. >> >> Laurentiu >> _______________________________________________ >> gdal-dev mailing list >> gdal-dev@lists.osgeo.org >> https://lists.osgeo.org/mailman/listinfo/gdal-dev >> >
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/gdal-dev
