Generative AI has disrupted openness and the commons sparking considerable discussion and re-evaluation of our collective work. In an effort to contribute to that conversation I’ve written a blog post AI, Creators and the Commons summarizing some of what got discussed at this pre-summit workshop. It reveals the many questions and issues that have surfaced and initial ideas for reconciling them and finding a way forward. I hope it is of interest to all of you and welcome discussion here.
The workshop and summit further motivated me to be proactive about shaping policy and regulation pertaining to AI. In hopes of motivating others to similarly be proactive I’ve written a separate Policy Recommendations for Canada’s AI Act blog post describing the briefing note of policy recommendations I submitted regarding Canada’s draft AI Act. As always feedback and discussion welcome.
Thanks once again, Paul, for sharing these important posts of the long view of AI and the Commons. I have to say it says something about the act of blogging that your first post created the opportunity for you to be part of Canada’s AI policy recommendation.
We hardly had sufficient time at the conference in Edmonton last month along with @cdlh to coordinate what we might be able to do through OE Global to coordinate some kind of working/interest group. Maybe there’s still a lot of grasping at what we can effectively do, or avoid the overwhelm of how broad the shifting landscape is. I’m overwhelmed!
However, the summaries in your new piece suggest that looking to copyright and licenses as a means to navigate and move forward look questionable. Yet it seems the top worry I hear in conversations and webinars with educators is an expectation or wanting for the rules of these familiar vehicles to be honored. It sounds like opt-out of AI mechanisms will be hard to do and then one never can know for sure if/how it is done.
It even questions about what the future “Commons” looks like. I wonder too what happens as AI starts including in its training content that includes AI outputs.
For me, and I hardly speak for anyone else, I can see in the near term:
Suggestions for now for best way to advise, attribute use if AI generated text/images/media.
Develop ideas for how these can (or if possible) be folded into licensed content (comprehensive attribution statements?)
Share, collect case examples of the effective/ethical ways AI is being used to create open content
Collaboratively develop guidelines so we are wheel reinventing, every institutution must be doing this at fast speeds now
I might want as well to make sure we are harnesssing the considerations of fairuse/dealing in other parts of the world, especially where not established.
And let us know what can help feed your work with the organizations you are actively working with. I’ve been taking in the writings and outputs from OpenFuture which you have been referencing, and would like us to somehow be generating some deep thought into what the Future of Open really can look like, it is already changing as I write, and we need to look how to shape it before it shapes us.
@cogdog , thanks for the interest and encouragement.
I’d love to see some kind of AI Special Interest Group (SIG) form here in OEGConnect. A SIG could do demos of AI, identify AI topics of interest to our global open education community, and even discuss and make decisions around how we could collectively influence AI. @cdlh is one of the most knowledgeable about AI and open so I agree it would be wonderful to get his guidance and involvement in such a SIG.
I like your list of things to explore in the near term. There are so many topics and questions it can be a bit overwhelming. A SIG might make it feel less so.
Thanks for identifying all the AI sessions from the OEGlobal conference. I by no means saw them all so feel that I missed out on some understanding that has already been shared. I’ll spend some time digging into the sessions I missed and glean what I can. Obviously those who presented may also have interest in an AI SIG.
And finally I really like your desire to see us generating some deep thought into what the future of open really can look like. As you note it is changing in real time and I think the global open education community is ideally positioned to actively shape it. This is the part that I’m most focused on.
Particularly interested in your thoughts “Creative Commons tools have reduced relevance in the context of generative AI. The way Creative Commons licenses involve attribution, giving back to the community, and creating a commons has been disrupted by AI. The original idea around Creative Commons was to give creators choice. How does Creative Commons support creator choice in the context of AI?”
And also the notion that CC / OER sites could effectively develop a [technical] blocker to being scraped by AI tech companies. So the OER is still accessible to those that need it most, but the creator is protected, the source/reference points is exposed and remains trusted.
Thanks Patrina. Creative Commons is absolutely carrying out their mission in this new territory.
Our experiences and intuitions built on decades of how we created and used digital content are being tested here compounded by how generated content is even generated, the systems being opaque.
The latest from CC is interesting, but also still in flux
noting the key points in the response:
AI training generally constitutes fair use
Copyright should protect AI outputs with significant human creative input
The substantial standard similarity should apply to Infringement by AI outputs
Creators should be able to express their preferences
Copyright cannot solve everything related to generative AI
While its uncomfortable for many of us, and I myself question the decision to train on content without consideration of permissions, I do rather agree that AI training is fair use-- it’s not making or distributing copies. Is it infringement for search engines to index content for search? It seems not really fair, but the arguments CC makes I can understand.
The second point is going grow vague of course with how one defines “significant human creative input” hinging on the lack of copyright protection for purely machine generated content, e.g. like purely random generators. When we use generative AI,. there is a human interplay in the prompting, refining, and perhaps post editing. This will be interesting to play out. I’d like us to be able to comfortably acribe reuse content to works that are not just spit out.
Likely in the grey zone is “substantial standard similarity”!
I appreciate the idea for “Creators should be able to express their preferences” be it exclusion (NOAIBOTS.txt?) like you suggest or the preferences/signals concept. But also, does that conflict with the first point? If training is fair use, do our preferences matter? (I know that’s harsh) Much to be worked out and even more hinges on willingness / or regulatons to have the AI owners comply. And then it goes to the complexity-- how will we ever know that they did? The call for transparency is urgent.
Mostly the last point is why this is challenging. Copyright is being stretched here.
Thanks CC for being out there in front of the machines!
I agree with @cogdog about the way Creative Commons is carrying out their mission in this new territory. I believe they have a significant role to play and part of that is representing all of us in the AI space! No small task.
I’m also eager to see how we can help CC and other organizations find the way forward.
Like Alan, and CC itself, I think creators should have a say in how AI works. The terms of the Writers Guild contract seem highly pertinent:
AI can’t write or rewrite literary material, and AI-generated material will not be considered source material under the minimum basic agreement (MBA), meaning that AI-generated material can’t be used to undermine a writer’s credit or separated rights.
A writer can choose to use AI when performing writing services, if the company consents and provided that the writer follows applicable company policies, but the company can’t require the writer to use AI software (e.g., ChatGPT) when performing writing services.
The Company must disclose to the writer if any materials given to the writer have been generated by AI or incorporate AI-generated material.
The WGA reserves the right to assert that exploitation of writers’ material to train AI is prohibited by MBA or other law.
Eryk Salvaggio breaks the areas of agreement into four ways the writers have a say; 1. Authority, 2. Agency, 3. Disclosure, 4. Consent.
I find it especially interesting to note that this is a labour agreement not a copyright reform and keep wondering what such an agreement might look like in education and from the lens of Creative Commons.
These 4 area are a useful lens to consider in education though as you suggest there is not a parallel labour agreement, does this call for some kind of policy agreement, some signal of alignment for educational works?
Authority and Agency support the use of AI as an educators choice, not mandated, and left to the creator as expert in subject matter. Disclosure seems doable with guidelines for how to attribute, cite.
There still seems looking at thing as an AI or not designation, making me wonder what happens as intermixing happens (AI enabled? Ugh). But maybe this matters not if Agency is established.
Consent is what we want but questioning how one ever knows if it is followed?
Is developing a corresponding series of voluntary agreements worth pursuing?