Post B7Y2rU7kkVxdYynZgG by larsmb@mastodon.online
 (DIR) More posts by larsmb@mastodon.online
 (DIR) Post #B7Y2rT6eX7UEPHHBKK by larsmb@mastodon.online
       1 likes, 0 repeats
       
       I think I've finally been able to condense what irks me about the @conservancy guidelines on #GenAI use for #FLOSS. Out of 14, only the first 2 relate positively towards not using these systems. The others are accommodations and even encouragement.That's ... remarkably biased and doesn't adequately reflect or even acknowledge the potential and real harms associated with them.The SFC used to be fairly principled. This does not appear to be. https://sfconservancy.org/llm-gen-ai/llm-backed-generative-ai-recommendations.html
       
 (DIR) Post #B7Y2rU72n9OTWmT0Zk by larsmb@mastodon.online
       0 likes, 0 repeats
       
       In the responses, I've also found the SFC associated folks fairly dismissive of criticism in that direction, while thanking those who support their stance of "it's okay actually".I'm reminded of OSAID. (aka industry capture)Iff more ethical systems existed, I could see the direction. But they do not.All of the current systems are fundamentally opposed to our shared values. Or, well, so I thought.
       
 (DIR) Post #B7Y2rU7kkVxdYynZgG by larsmb@mastodon.online
       0 likes, 0 repeats
       
       @conservancy I think those guidelines are sensible and the minimum to mitigate the harms when the use is unavoidable.But given everything we know about these systems as of today, I think I'm missing a default recommendation.Rule 0: the only winning move is not to play.
       
 (DIR) Post #B7Y2rUqm37yPobrWr2 by larsmb@mastodon.online
       0 likes, 0 repeats
       
       I know this is somewhat hypocritical. I'm also using them as part of my job, and I can't deny their situational utility. But I'm not going to pretend they're ethically defensible.(Individual vs systemic.)Organizations that want to be seen as leaders in the FLOSS space need to be able to hold that tension, and help limit their use.
       
 (DIR) Post #B7Y2rVXJUy07wXlVA0 by larsmb@mastodon.online
       0 likes, 0 repeats
       
       Also the basic idea of shared values in "communities" that balk at whenever that might mildly inconvenience the majority (accessibility, DEI, health, ...) is, empirically, extremely naïve.Those things don't exist at scale.