AI has become immensely better at image generation in recent years. It's now at the point where amazing images can be created readily by those with no artistic background. Tools like midjourney and stable diffusion allow a simple text prompt to be turned into infinite visualizations. These tools have had an adverse impact on artists who make a living selling the images they create. Painting and drawing is one of the first disciplines to become partially automated by AI.
Many artists are understandably upset about these new tools. Part of their case is that AI models are trained on most of the images on the internet, which includes all kinds of copyrighted work used without permission. Companies that train AI models are pushing to have the law modified to make this legal (which has already been passed in Japan). It's not currently illegal, since it's a scenario that has not arisen previously and has no precedent.
Many people that appreciate art do so because they form a connection with the artwork or the artist. That becomes harder when the artist is not a discrete individual but rather a continuous melding of many. We try to address what creativity means now and why people prefer human art. We also outline some ways in which society could act to protect the interests of artists, by fully supporting them or supporting career transitions as necessary. Although artists are the ones swept up by this change right now, automation will come to supplement and then supplant jobs all across the economy, so we need to start figuring out this problem now.
Antonio Cabrero @antoniocabrero
/ @antoniocabrero
Why Artists are Fed Up with AI Art. (SamDoesArts)
• Why Artists are Fed Up...
Kris Kashtanova @icreatelife
/ icreatelife
Create Worlds and Characters with AI with CreativeMornings and Adobe
/ 1753239271245234342
Japan Declares AI Training Data Fair Game and ‘Will Not Enforce Copyright’
petapixel.com/2023/06/05/japa...
#ai #art #creativity
0:00 Intro
0:42 Contents
0:49 Part 1: Economic agency
1:14 Examples of image generation AI systems
1:43 Contentious topic, be nice
2:07 Why are some artists angry about AI?
2:57 Also an economic issue
3:19 Copyright preserves economic benefit
3:55 Is it acceptable to train on all images on the internet?
4:32 Copyright law can change overnight (e.g. Japan)
5:24 What makes AI different from artists and past tech?
6:05 AI can remember everything
6:27 AI is 1000x faster than a human artist
6:57 Not fundamentally different from a human
7:27 Comparison to music sharing websites (Napster, Grooveshark)
8:30 Societal norms can be changed
9:03 In the case of art, AI companies have the economic backing
9:44 Example: Japan legalized training
10:14 Part 2: Creativity and the human condition
10:46 AI is a continuous entity, not discrete
11:29 Can AI be creative? At the high level?
11:57 Example: my thumbnails via midjourney
12:34 Example: tool is assisting human expression
13:22 AI will start to understand the "why"
13:34 Style evolves over time
13:56 Part 3: What society could do
14:14 Status quo not good (e.g. with disabilities)
14:43 "Democratizing art"
14:51 Case study: music separates copyright
16:03 Case study: books and novels
16:46 What should society prioritize?
17:10 Skill and hard work not economically valuable
17:47 AI unemployment insurance
18:29 But how do you pay for it?
18:59 References including workshop
19:38 Conclusion
20:25 Connection is difficult with an AI system
21:21 Outro