Since you’re writing a book you may also want to consider the impact of opaque data-less “open source” models in terms of creating a “chain of custody” from your work back to the content oligarchs: Disney, NYT, etc. but with models Rajiva’s on Common Crawl, everyone on the internet.

That today’s #opensource proponents like the OSI are actively participating in what effectively amounts to contributory infringement of copyrights boggles the mind!

@florenciocano

@sj @florenciocano because that's not how copyright works unless you'd also be perfectly fine with being a perpetual debt peon to the #Contentmafia...

• Regardless of the fact that most wannabe-"#AI"* is #WastefulComputing at best...

I’m afraid @senficon is mistaken (as you can see from the many comments pointing this out) and his posts (eg https://opensource.org/blog/copyright-law-makes-a-case-for-requiring-data-information-rather-than-open-datasets-for-open-source-ai) are being used by the bandits at the OSI to make the case for #OSAID based on wishful thinking.

These models are behaving more like photocopiers, which clearly do not magically erase copyrights.

For example, consider the GPL-protected output of Linux as a Model (LaaM) https://github.com/alea-institute/laam

@kkarhan @florenciocano