Julia AngwinAI is leading to a tragedy of the commons: It is overgrazing the commons, scraping up everything it can, and is also being used to pollute the commons with untrustworthy content.
Prof. Emily M. Bender(she/her)@Julia You make really good points --- and we need a better metaphor than the apocryphal "tragedy of the commons". On its eugenicist origins, see:
https://blogs.scientificamerican.com/voices/the-tragedy-of-the-tragedy-of-the-commons/
@Julia And for some work on pollution of the ecosystem that rhymes with your points, see:
Shah, Chirag and Emily M. Bender. Under review. Envisioning Information Access Systems: What Makes for Good Tools and a Healthy Web?
poritzj@emilymbender @Julia
@datamyna and I have a talk for the @creativecommons #CCSummit2023 which also echos these points -- and, of course, references your important work!
The slides and prerecorded video are at https://poritz.net/j/share/WGAIDDOS.
I'm particularly proud of our analysis of the failure of generative AInt to match CC's idea of #BetterSharing, and an argument that, while copyright is not the best or only tool to fight the ills of AInt, it is almost certainly violated in all current AInt
1/3
@datamyna and I have a talk for the @creativecommons #CCSummit2023 which also echos these points -- and, of course, references your important work!
The slides and prerecorded video are at https://poritz.net/j/share/WGAIDDOS.
I'm particularly proud of our analysis of the failure of generative AInt to match CC's idea of #BetterSharing, and an argument that, while copyright is not the best or only tool to fight the ills of AInt, it is almost certainly violated in all current AInt
1/3
@emilymbender @Julia @datamyna @creativecommons
Part of the copyright analysis is a new (maybe?) way to think about AInt: there are millions of parameters in modern LLMs. These are just a highly compressed form of the terabytes of training data: after you've processed many images of faces, the additional bytes needed to include one more are many fewer than the raw data content of that new image ... i.e.: it's a compression scheme! And a compressed copy is a copy, so violates copyright!
2/3
Part of the copyright analysis is a new (maybe?) way to think about AInt: there are millions of parameters in modern LLMs. These are just a highly compressed form of the terabytes of training data: after you've processed many images of faces, the additional bytes needed to include one more are many fewer than the raw data content of that new image ... i.e.: it's a compression scheme! And a compressed copy is a copy, so violates copyright!
2/3
@emilymbender @Julia @datamyna @creativecommons
Which is also why they cannot be transparent on their training datasets: copyright violation!
Even if they use CC-licensed training data, they owe attributions; if they use public domain or CC0 data, in many jurisdictions there is still a requirement of acknowledgement. To be transparent is to admit they are in massive copyright violation -- the statutory damages alone will be huge! John Grisham and George RR Martin's lawyers should use this!
3/3
Which is also why they cannot be transparent on their training datasets: copyright violation!
Even if they use CC-licensed training data, they owe attributions; if they use public domain or CC0 data, in many jurisdictions there is still a requirement of acknowledgement. To be transparent is to admit they are in massive copyright violation -- the statutory damages alone will be huge! John Grisham and George RR Martin's lawyers should use this!
3/3
@emilymbender OMG - thank you for pointing out. Good grief! But can we steal the idea back from him? It is such a good metaphor.
@Julia I think we probably need a new one --- because ordinary people working with a public good treat it as a public good. The tragedy (as you point out in your piece) has to do with profit-incentive driven big tech & billionaires...
@emilymbender I’m sure you have read Elinor Ostrom’s work on this - and how public goods are better maintained by communities of trust than by profit or legal motives.
It gives me hope, but is hard to figure out how it applies to such a global public good with so many diverse communities.
I ❤️ Placemaking@Julia @emilymbender A recent NYTimes Magazine piece that talks about "tragedy of the commons" as "widely debunked" (and Ostrom cited, who was new to me) --- also I feel like it's not fair; NYTimes editors should do a better job at...reading the NYTimes themselves.
https://www.nytimes.com/2023/07/26/magazine/right-to-roam-england.html
https://www.nytimes.com/2023/07/26/magazine/right-to-roam-england.html
Elias@Julia @emilymbender what those big tech companies are doing is disgusting, but I'm thinking what needs to happen is not so complicated after all: whoever makes such a system must be required to declare exactly what "training" data they use as input and if even a single one of those inputs requires attribution, like Wikipedia, then the system is obviously illegal to use.
I do realize that this is very far from what is happening now. But still. In principle the issue could be handled like that?
@eliasr @emilymbender transparency about training data would be a good start!
@Julia Yeah, I'm reminded of @alex and Tina Park's essay "Against Scale":
https://alex-hanna.com/static/pdf/Against_Scale.pdf
... which I think suggests that the question is, how do we build interconnected communities of trust, rather than trying for one big one.
@emilymbender @alex amen and is of course why we are here on mastodon trying to do that!
Dario@alex @emilymbender @Julia If I may also add to this by saying, scale and growth to what end?
The adage of ‘die a hero or live to become the villain’ feels all to true when there is no ceiling.
Alexandre Oliva (moved to @we already have. the "tragedy of the commons" meme, as well as its scientific underpinning in game theory, behavior analysis and economics, as well as the known solutions for the abstract problem, bear no relationship whatsoever with those eugenicist origins; those are just noise that has been discarded over time. rejecting the perfect-fitting memetaphor ;-) because of its author reeks of cancel culture, and is no more than an ad hominem fallacy. rejecting it because of the noise it was first published with is just as fallacious. rejecting it because some of the resource overusers attempt to guilt us into using less so that they can use even more would be missing both the point of the memetaphor and the known solution: a credible commitment between the parties to avoid overuse. throwing away the understanding and the known solution for a recurring behavioral pattern that's a perfect fit for pollution and for overuse of natural resources just because the term can be traced back to someone we disapprove of is about as fallacious (logically unsound) as demanding roses to be renamed, or claiming they stink, because someone highly reproachable happened to like them. me, I prefer shakespeare: they would smell just as sweet. the game theory models and solutions would still be the same, we'd just be using a less recognizable name for them, and that would hamper rather than promote knowledge and action that we need so urgently
John M. GambleBut it's not a good metaphor – it was debunked pretty thoroughly.
This not to say that the "commons" can't be despoiled, but it is not an inevitability – it takes intent.
I think the reason the concept has persisted is that it's a genuinely poetical name. It resonates like a good song lyric would.
Now at I still remember learning the tragedy of the commons in my first year Environmental Studies course at UWaterloo in 1997.
This article was a huge eye opener. Thanks for posting.
Óscar Morales Vivó@emilymbender @Julia it turns out the tragedy of the commons is actually the tragedy of extractive capitalism.
John Faithfull🌍🇪🇺🏴🧡 ✊🏼✊🏿
David Mitchell 
It is quite simply the theft of the commons, an enclosure of the commons, that old old story of the powerful and influential harvesting value created by others.