Conversation

Notices

1. sofia ☮️🏴 (sofia@chaos.social)'s status on Wednesday, 10-Jan-2024 19:57:01 JST sofia ☮️🏴

   i think to create generative #AI models based on public domain data would probably be a worthwhile experiment. but i think it's even more interesting philosophically for what it says about #copyright.

   first of all, there is a lot of data you can get with fairly little effort, the existing pool of public domain data first of all. next, a small fleet of camera drone can easily gather terabytes of pictures of nature and historical sites.

   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 19:56:56 JST mnml mnl (he/him)

     in reply to

     @sofia i'm really taken aback by the copyright stan-isation of the countercultural debate here, as if the nytimes was a defender of the rights of individual journalists. No, if they were to train their own model on their journalists output they would go ahead, just like adobe and shutterstock and co.

     openai paying nytimes licensing fees won't change anything for the better.

     LS likes this.
   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 19:56:57 JST mnml mnl (he/him)

     in reply to

     @sofia like i'm perfectly fine ethically with training my own model on the entirety of all papers in libgen... As long as I don't create the "replace a scientist machine". And while I take openai's stance with a pretty big grain of salt, they aren't (compared to midjourney, say), pushing for the "let's replace people" model, and provide tools to remove your self from the training corpus.

     FWIW, lip service does a fair amount here and is not in the headlines.

   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 19:56:58 JST mnml mnl (he/him)

     in reply to

     @sofia yes. Calling for a lock down of training data based on copyright law would be a terrible thing for opensource models, imo. There's also a lot of research that goes into generated datasets and just curating leads to much better training results.

   • sofia ☮️🏴 (sofia@chaos.social)'s status on Wednesday, 10-Jan-2024 19:57:00 JST sofia ☮️🏴

     in reply to

     next you can just pay artists to liberate their art or create new public domain art. same for journalism and science. how willing people are to do that is uncertain, some will be strictly opposed, to me it would be a dream come true. a unique collaborative effort, stacking shoulder on shoulder to build a giant. a giant that may expand the capabilities of billions of people (including me) in ways i can't even start to predict. and i might even get paid for it 🤑?

   • LS (lain@lain.com)'s status on Wednesday, 10-Jan-2024 19:57:55 JST LS

     in reply to

     • mnml mnl (he/him)
     @mnl @sofia
     > like i'm perfectly fine ethically with training my own model on the entirety of all papers in libgen... As long as I don't create the "replace a scientist machine".
     
     why the last part? seems that having a machine be able to do proper research would be amazing.
   • LS (lain@lain.com)'s status on Wednesday, 10-Jan-2024 20:05:29 JST LS

     in reply to

     • mnml mnl (he/him)
     @mnl @sofia if you can actually generate content that's good enough with a machine, then i see no reason why a human should do it. Seems to me the same as using an electric dishwasher instead of hiring people to do it by hand.
   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 20:05:30 JST mnml mnl (he/him)

     in reply to

     • LS

     @lain @sofia So much of the use-cases really depend on the context. For example, I build many "automate this job entirely", say, writing SEO titles. Because I work for a family business, that means that Joe can now, instead of grinding through 5000 SEO titles, do more product photography.

     But, I could also build "fire all content writers" machine, with the same prompt.

   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 20:05:31 JST mnml mnl (he/him)

     in reply to

     • LS

     @lain @sofia yeah i meant more like "let's fire research assistants and replace them with AI". Maybe not the best analogy here.

   • LS (lain@lain.com)'s status on Wednesday, 10-Jan-2024 20:09:17 JST LS

     in reply to

     • mnml mnl (he/him)
     @mnl @sofia i suspect this is mostly the case right now. I do use AI assistants for programming and they still aren't really that great, but they do help a lot with boring tasks and make me more productive.
   • mnml mnl (he/him) (mnl@hachyderm.io)'s status on Wednesday, 10-Jan-2024 20:09:18 JST mnml mnl (he/him)

     in reply to

     • LS

     @lain @sofia what about selling people on the concept that the machine could do it better, when it kind of can't? Or if it was the case that a human + the machine could do way better.