Training data access

[fontana]

@fontana

if there is any possibility that the OSI will completely ignore what “open source” historically meant at least at a high level

To be clear, I think it would be legitimate for the OSI to carefully undertake a radical reconception of the Open Source Definition to reflect contemporary views on the topic from various interest groups, without deciding that open source has lost all meaning. I’ve given at least a few minutes of thought in the past to whether the OSD could or should be revised to embrace some of the licensing models championed by the ethical source movement. But that’s not what’s going on here, clearly - the OSI is not touching the OSD at all and is signalling that for non-AI-system software the OSD is (for the time being) here to stay. Hence the questions a few people have raised here, about what the connection is between the OSD and the OSAID.

[stefano]

@fontana

Suppose a majority of the people you have classified as “stakeholders”, or maybe even just the most vocal or influential ones, decide “open source AI” means, for example, that licenses of components of an AI system can prohibit distribution.

The board would never approve such a definition. Also at this point, it should be quite clear that none of the drafts have even remotely touched on limiting the distribution of AI systems. This is not a likely scenario.

@fontana

I note that the OSAID drafts use language that is based on the Free Software Definition and do not refer to the OSD

Think of it this way: The OSD came after the FSD, so you can see that the short definition of What is Open Source AI is answered in a way similar to how you answer the question what is Open Source (software), by reciting the 4 freedoms. This document follows that path.

@fontana

who is a stakeholder?

We have identified 6 categories of stakeholders. You can find their description on page 29 of the last town hall presentation.

And yes, you’re a stakeholder, and you could fit in at least 4 of those categories with your corporate and personal role

This thread is now very far from its original questions… Confirming that every time we start talking about data in abstract terms, we go down a rabbit hole. Let’s close it here and move to a more productive approach (since draft 0.0.6 is coming out this week) and there are more interesting debates we should be having.

[Aspie96]

@stefano

Uhm … The very first draft (v. 0.0.1, unreleased publicly) used the term machine learning but nobody in the private working group liked it. There was strong push back to rename it Open Source AI.

I think the question was about “open source”, more than “AI” vs “ML”, if the standard ends up being different from what has been called “open source” in software so far.

@stefano

We noticed three years ago that we’re lacking a shared understanding of the AI space

Do we truly have a shared understanding outside of the AI space? Because many people use the phrase “open source” differently than OSI. I’m not sure AI is different in any significant way in this regard.

@stefano

If there was a strong push to use another term, I’d be very happy but we have to deal with reality. Open Source AI is the term nobody likes that everybody is already using, including legislators in the EU. We simply have to deal with it.

Legislators, including in my own country (Italy) trust OSI to define open source licenses and use “free software” (“software libero” in Italian) and “open source” (either “open source” or “a sorgente aperto” in Italian) interchangeably.
Precisely because of this, I think equivocation would be a problem. Governments appear to be assuming that OSI won’t change what’s meant by “open source” (else why even rely on OSI to determine which licenses relate to policies about FLOSS?).
If a standard for a kind of software is different than that of open source (such as in the ways @fontana pointed out), it really deserves a different name.

@fontana

Does that mean the OSI abandons its historical principles and embraces a definition of “Open Source AI” that legitimizes restrictions on distribution?

In my view that would rightfully undermine both the trust and the authority that OSI gets.

@fontana

To be clear, I think it would be legitimate for the OSI to carefully undertake a radical reconception of the Open Source Definition to reflect contemporary views on the topic from various interest groups, without deciding that open source has lost all meaning

This is where I strongly disagree with @fontana.
I don’t think it would be legitimate. I think it would be a betrayal to those who have supported OSI and the open source movement, in any way (not just economical) and entrusted it with the authority it has.
There should be a way for a legislator or any member of the common mob to say that they support a specific standard of licensing, based on a specific set of values.
Modifying the OSD to allow for “ethical licenses” would be deeply unethical because OSI has been acting as the way for people to make statements of that kind and have them be meaningful and stable.

[stefano]

@Aspie96

Legislators, including in my own country (Italy) trust OSI to define open source licenses and use “free software” (“software libero” in Italian) and “open source” (either “open source” or “a sorgente aperto” in Italian) interchangeably.

you answered your question: There is such a shared understanding (despite the occasional chatter on hacker news and twitter) of what Open Source means that politicians refer to the OSI-maintained definition to write their rules and regulations.

and let me be clear: NO, OSI board will never approve changing the meaning of Open Source.

Let’s stop this conversation on what-if scenarios that are extremely unlikely to materialize, it’s not useful.

[fontana]

@Aspie96

This is where I strongly disagree with @fontana.

I mean, I disagree with myself too.

[lumin]

I emphasized again about the dataset issue upon the release of OSAID 0.0.6 A new draft of the Open Source AI Definition: v.0.0.6 is available for comments - #10 by lumin