@davidnjoku -- I think people have a fundamental misunderstanding of #AI generated material.
They think it is an easy way of making new art, but it can only be trained on current art.
Everything created by AI is a derivative.
People who rely on AI instead of real creative people will end up making a constantly reiterated derivative of itself.
How fucking sad is that?
You also never address the theft of art for these generated images. How can the industry move forward without addressing that?
Conversation
Notices
-
Embed this notice
Church of Jeff (jeffowski@mastodon.world)'s status on Thursday, 07-Mar-2024 17:58:22 JST Church of Jeff -
Embed this notice
David Njoku (davidnjoku@mastodon.world)'s status on Thursday, 07-Mar-2024 17:58:23 JST David Njoku (2 of 2)
Yes, AI must improve. But I suspect that users need to too. Maybe we need to recognise that expecting a single text field to satisfy all our needs - from diagnosing Alzheimer's to generating videos of purple puppies - is asking too much. It'll never do anything perfectly.
Maybe I need to explicitly ask for "White" Nazis. But I should be able to ask for "Black" doctors and get results.
-
Embed this notice
David Njoku (davidnjoku@mastodon.world)'s status on Thursday, 07-Mar-2024 17:58:24 JST David Njoku I've been trying to put together my thoughts on "#woke" #AI and how we're blind to the bias we live with, but keenly aware of any changes to it.
AI has always been biased; when I posted months ago about Dall-E refusing to generate images of Black doctors no one seemed that shocked. But when #Gemini overcorrects when trying to deal with the problem there's global outrage?
(1 of 2)
-
Embed this notice