As the Open Source AI definition is coming to an end, we would like to gather reactions regarding its compatibility with the Open Definition. Do you think it is compatible? Do you think OKFN should endorse it? Deadline for comments will be October 22nd. Here the details of the current version: The Open Source AI Definition – 1.0-RC1 – Open Source Initiative
3 Likes
As a developer, I can feel the pressure to make use of AI techniques and models being incredibly strong right now, and good guidance on the Open Source aspects would be extremely helpful - in other words, I think the OSI’s initiative is timely and well meaning. It is however, worth pausing to question their co-design process and transparency of motivations - as @tante here playing advocatus diaboli in a recent post:
As a (over the hill, based in a priviledged European country) technologist I am of course concerned about the urgency and hype with which this topic is being promoted, potentially overshadowing a cautious and well considered approach. During the OSI town hall today, Stefano Maffulli challenged us to remember all 10 points of the Open Definition (compared to their lighterweight one). We need to respond to societal needs and the needs of advocacy groups for these documents to continue to be meaningful, as recent update efforts have again purposed to do:
My personal opinion aside, I am trying to monitor conversations in our local community, engage folks at hackathons and keep internal channels on this topic alive (#admt). I am also collecting various bits and pieces, threads and projects - in particular looking for datasets related to ethical / fairness issues in AI here:
2 Likes