daniel

Public Key

npub1sg43lsu2prvlw2r5j9l7dpznf88f2ep53yhjy2xys6caxmvtsgzq4ykh67

NIP-05 Address

Profile Code

nprofile1qqsgy2clcw9q3k0h9p6fzllxs3f5nn54vs6gjtez9rzgdvwndk9cypqpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dsjjwfhd

Publishing to

relay.ditto.pub

relay.damus.io

nos.lol

Author Public Key

npub1sg43lsu2prvlw2r5j9l7dpznf88f2ep53yhjy2xys6caxmvtsgzq4ykh67

Show more details

Last Notes

2026-07-29 20:25:06 GMT

- reply

@nprofile…20kn Well, the way this AI training goes is that all of open source and all their forms have already been used at least once with disregard for any license in training in these models since at least the GPT-3 aera, so the models don’t see anything new. I have no doubt of that. The damage is done to humanity’s corpus of intellectual work and there is - for now - no turning back.
Pandora was right.

2026-07-29 20:21:31 GMT

- reply

@nprofile…20kn is there a formal voting process to approve such a policy, or best done through an issue or PR?

2026-07-29 19:23:10 GMT

- reply

Maybe a clear communication about the use of AI in the Friendica project would be useful:
Here are the direct URLs to the official generative AI contribution policies of some prominent open source projects as source of inspiration:
The Apache Software Foundation (ASF) ASF Generative Tooling Guidance
apache.org/legal/generative-to…
The Linux Foundation: Generative AI Policy
linuxfoundation.org/legal/gene…
The OpenInfra Foundation Policy for AI Generated Content
openinfra.org/legal/ai-policy/
Mozilla (Firefox) Firefox AI Coding Policy
firefox-source-docs.mozilla.or…

2026-07-29 16:37:14 GMT

- reply

@nprofile…20kn It rubs me, indeed, especially because of the black and white judgement. On one hand you say code will be reviewed on its own merits, but then you say "but not if the code was..." carelessly generated by someone pointing an AI tool at the codebase (my wording, you did not get into details) -- but you made the thing personal with "casts doubt on the contributor themselves".

2026-07-29 07:56:50 GMT

- reply

@nprofile…20kn So you are assuming that AI is not creating usable code?
I work with AI and we generate lots of code using AI — thus my experience is different and my opinion otherwise.
But with a codebase as old and battle tested as Friendica or other open source i can understand that accepting AI generated code is different than if you write all code with AI in the first place.
We all have limited time, and i am only trying to find out if i can help the Friendica project my way - with AI as main code generator - or if i will focus elsewhere.

2026-07-28 21:08:40 GMT

- reply

@nprofile…20kn
So a well executed PR for a small fix or bigger feature is ok if there is no additional ai generated bloated text?
I think there’s an opportunity here that can create good value, but i would hate to waste people’s time with things they don’t want to have.

2026-07-28 16:15:20 GMT

Friendica dev codex on using AI?
@Art4, thank you for approving the small bug fix PR of mine. It was generated by Jules.google.com
git.friendi.ca/friendica/frien…
What is the Friendica codex for this? If it's OK to use AI for these kind of bug fixes, security hole filling or small improvements, I would volunteer to keep my Jules running and make small PRs like that.