Home
Artists
Search Recent Random
Posts
Search DMs Popular Hash Lookup Tags Random
Importer
Import FAQ
Register
Partychan Telegram ThePornDude
Home Artists Posts Import Register

Crossover episode: Is A.I. good or bad for art? (Patreon)

Published:
2022-09-26 20:38:53
Imported:
2023-11
Tags:

Downloads

Content

Mat joined our friends Emilie & Andrea at The Culture Journalist to cover a brief history of the current wave of AI art, why artist replacement narratives can veer on the absurd, and the work we are doing launching our new organization Spawning. 

Read and subscribe to the Culture Journalist: https://theculturejournalist.substack.com/ 

See if you feature in AI training models, and opt out or opt in: https://haveibeentrained.com/ 





Files

Comments

Victor Mustin

This discussion was great, every line of yours is gold Mat

Zane

Great convo. I'm curious for Spawning, has there been any sort of thought to the proof of ownership race? I.e. I don't own the rights to your work, but I go an opt-in every image you've produced as though it's mine. You later go and can revoke it, but if there's compensation involved down the line, it becomes much more difficult to ascertain who is the true owner. The person who registered first or someone else by some other criteria. It's like figuring out "true" proof of origin. I think part of that is in the dataset itself, and then human general detective work. The registering of a work itself it seems mainly a messy human problem. I think a few years back Vinay Gupta suggested insurance for identity it seems like this is the same problem. It seems like you'd want to evolve to a similar system for proof of ownership of the work, where people stake something ( in his case it was money, but could be reputation) to say that something is authentic, and if at any point it's challenged and proved not authentic they lose their stake. Seems like the system needs curators and detectives to find fraud and validate claims, and then incentives for people to do just that. It's been like 4-5 years on since that insurance talk/idea, and haven't really seen it floating around in implementation for these kinds of things. Proof of stake is kind of like that with capital ( don't validate, get slashed), but not really so much for curation and decision making (Truebit exists).

interdependence
>>93099564

you are totally right, verification is a minefield and I quite like your staking idea. At the moment we are playing with more simple verification processes, and we have a human check if the person saying they are x are actually x, but at scale it will require something more robust