It wouldn't matter anyway as they announced the deadline as 17 February, which would mean midnight local time by default (and should mean midnight AoE, especially for an organisations purporting to be concerned with openness).
This this is another issue, and it's why I'm not holding my breath for the data and will be satisfied with rebranding to the more appropriate Open Weight moniker (caveat: license).
We’re finally starting to see some precision in language from the company, in that they consider that the Meta LLaMA AI models themselves are NOT in fact Open Source, rather Ope nWeight. Thank you for admitting on the public record what the Open Source community has been saying all along! The data is the source for AI and it’s not provided, though Harvard’s move is a step towards rectifying that.
"Nobody disagrees about the principles [behind the OSAID], where we see disagreement is implementation"
The principles are the four essential freedoms of free software, and the one job the OSI had in implementation was to work out what had to be open to protect them.
“Some of the tension we've seen is in people that haven't made that mind shift to the wider picture and are just trying to apply what they are very familiar with, that they're experts at, and that made it more difficult.”
Like many of the objectors, I’ve been in Open Source since before there was Open Source or the OSI, and am now doing a masters specialising in ML while coding an AI OS. This is condescending BS from talking heads who don’t even claim to have AI expertise.
Meanwhile their chosen “co-design” process is literally the “Do Your Own Research (DYOR)” of technical standards:
“We believe that everyone is an expert based on their own lived experience, and that we all have unique and brilliant contributions to bring to a design process.”
Karsten correctly called the consensus on that thread, including my own view as one of the folks named in the article, that accepting less-than-open "public" datasets like the dumps of the Internet made available by the Common Crawl Foundation may have been an acceptable compromise to cast a wider net in recognition of AI industry norms. I no longer believe that to be the case, accepting the Open Source Definition (OSD) author Bruce Perens' view that the data IS the source, on the basis that it is what you need to modify in order to freely change the output of the system.
The OSI's position that ANY data is acceptable has shifted the Overton window of Open Source, and categorising it into open, public, obtainable, and unshareable non-public data, only to then accept all of them, is a form of doublespeak appearing to maintain openness while accommodating even the most restrictive data. We don't negotiate with terrorists.
Indeed, there are two dimensions to "Open" AI systems which "can be freely used, modified, and shared by anyone for any purpose" (per The Open Definition): openness, which is already well-covered by the Open Source Definition, and completeness, which is covered implicitly — after all, AI systems are software — but which would need to be specified in approved frameworks that could be self-applied like the MOF (were a new Class 0 to be created requiring open data licenses rather than "any license or unlicensed" like Class I).
In other words, the Open Source Definition (OSD) covers openness but not completeness (at least not explicitly, which is arguably a bug the community may want to fix in a future version so it covers both). The MOF covers completeness but not openness. The OSI's proposed OSAID covers neither, so any vendor using it to open-wash closed systems as Open Source AI rightly deserves ridicule as it is patently ridiculous.
I don't necessarily agree and suggest the Open Source Definition could be extended to cover data in general (media, databases, and yes, models) with a single sentence, but the lowest risk option is to not touch something that has worked well for a quarter century.
Yes, and Open Source started out with a much smaller set of software that has since grown exponentially thanks to the meaningful Open Source Definition.
We risk giving AI the same opportunity to grow in an open direction, and by our own hand. Massive own goal.
Nobody's asking for exact reproducibility — if the source code produces the software and it's appropriately licensed then it's Open Source.
Similarly, if you run the scripts and it produces the model then it's Open Source that happens to be AI.
To quote Bruce Perens (definition author): the training data IS the source code. Not a perfect analogy but better than a recipe calling for unicorn horns (e.g., FB/IG social graphs) and other toxic candy (e.g., NYT articles that will get users sued).
"We believe that everyone is an expert based on their own lived experience, and that we all have unique and brilliant contributions to bring to a design process."
The result is a proposed standard that does not do its one job: protect the four freedoms that define Open Source (and Free Software in general).
"We believe that everyone is an expert based on their own lived experience, and that we all have unique and brilliant contributions to bring to a design process."
"If computers of the kind I have advocated become the computers of the future, then computing may someday be organized as a public utility just as the telephone system is a public utility... The computer utility could become the basis of a new and important industry."
— John McCarthy (speaking at the MIT Centennial in 1961).
It's been pointed out that McCarthy used the term "utility" rather than "cloud", but many of us would argue they are one and the same. Indeed Wikipedia defines it as:
"Cloud computing is the delivery of computing as a service rather than a product, whereby shared resources, software, and information are provided to computers and other devices as a utility (like the electricity grid) over a network (typically the Internet)."
In any case, can anyone think of anyone more deserving of the honour of the title "father of cloud computing" than McCarthy? I certainly can't.
Item 2: Adopt a process for formal review of previously approved licenses
Item 3: Remove “code of silence” from Board Member Agreement
Item 4: Directors should be allowed to use FOSS for Board activities