Thoughts on AI Contributions in Open Source

Thu, 07 May 2026 00:00:00 +0000

With recent grumbling about permitting AI contributions to FreeCAD, threats of core maintainers departing under this policy and my general desire to have absolutely NOTHING to do with AI, I figured it was time to start taking part in the weekly FreeCAD developers meeting. Last Sunday (26/06/02) was my first one and while I mostly kept to myself (some snark was required) I have some observations.

The Core Argument Against a no-AI policy.

Firstly I think its important to show Yorik’s proposal for a no-AI policy, so far I’ve yet to see anything as articulate as this from the pro-AI camp.

So the argument goes, for some reason, a no-AI policy is will be more difficult to enforce than a ‘some-AI’ policy. At first, I cannot fathom how this is possible, it is proposed that ‘because people are going to do it anyhow’ the no-AI policy will be more difficult. This still doesn’t make any sense to me. I likened this to “people commit murder, therefor we should relax murder laws”. Of course the (intended) absurdity of this was jumped on in an effort to not acknowledge the underlying point.

But as I think about it more, I believe I have the answer. It’s because with a watered down policy, they got what they want, an opening to use AI and they have ZERO intent to actually do any enforcement. Enforcement is someone else’s responsibility.

Bike Polo

(didn’t expect it to go here, did ya?)

I used to play bike polo, a lot. I spend the better part of 5 years traveling the world playing. I played so much I was part of the core organizing team for the 2008 World Championship. Good days, I had a lot of fun and met a lot of great people, many of whom I am still friends with to this day.

What’s this to do with AI policy?

In and around 2008 there was a lot of debate about contact. The no contact group had a list of concerns, people getting hurt, equipment getting broken and most importantly, it made referring increasingly difficult. The pro-contact people basically came with, “I wanna hit people”. The 2008 worlds was the last to have a no contact rule set and in the year following contact was ratified in the increasingly sizeable rule book.

Fast forward 20 years and from what I understand (my partner still plays), the rules have evolved to be basically no contact, but now with a half dozen pages dedicated to describing what no contact means.

The 2008 rules where on a single page.

Foss on oit blarg!

Thoughts on AI Contributions in Open Source

The Core Argument Against a no-AI policy.

Bike Polo