What's the difference between a Tweet and a prompt? (The minimum Copyrighted text is 11 expertly crafted words I think, so the above would not even fall under that category.)
@alexjc
-
ML Data Use and User Privacy Consent on Social Platforms
By
–
By now (already last decade) ML use is not future and anyone still on the platform is explicitly consenting to continued use of the data. Copyright is not a good ground for primarily mediocre thoughtless Tweets, but privacy may be…
-
Rightsholders monetization rights on tech platforms
By
–
The rightsholders signed them away via TOS. Time for them to move to different platforms if they don't want to be monetized twice.
-
Web Scrapers: Essential for Keeping the Internet Open
By
–
Scrapers are the cause because without them, the web would remain open.
-
Twitter Closes Dataset Access Against Commercial Scrapers
By
–
I'm not here to convince you because you made up your mind. My goal here is to make great points for smarter people than you reading, and use humour to have fun in the process! Twitter is closing up because their "dataset" is valuable and they don't want commercial scrapers.
-
Fair Use in AI: Legal risks and educational approaches
By
–
My own educational perspective has been instead "it's possible to do Fair Use legally, but it's by far the riskiest approach and there are the fewest benefits."
-
Expert positioning and government policy subtlety in AI debate
By
–
I don't think it's beneficial for experts to take such positions, nor for lawfirms to secure additional business, because the debate has moved beyond that. Even governments are issuing policy statements that have more subtlety to them:
-
Rhetorical Tactics in AI Model Theory Debates
By
–
When arguing against this perspective, make sure you don't mention the idea that a model is a database, or that a model does compression, or you'll fall off the edge of this carefully crafted argument.
-
Fair Use for ML Training: Legal Risks and Uncertainties
By
–
[🅳] conclusion I think it's irresponsible to make definitive statements like "Fair Use for ML is probably legal" because it's misrepresenting the situation. – it's early on in the process, lawsuits underway, risky
– depends on the use, depends on the model, depends on the data -
Defining Expression: Open Questions on Copyright and AI Generative Models
By
–
Beyond that, there are so many open questions. What is an expression? At what point do combinations of statistics no longer represent expressions? At what point does Copyright disappear? This paper doesn't address any of them, yet presents many of its conclusions as facts.