A generative adversarial network is a type of machine learning framework often used for creating images.
Thinking of GANs as a collage
Dada, Instagram, and the Future of AI/
- collages are Dada 1.0
- snapchat (filters, AR facemasks) / its clones are Dada 2.0
- many things are a collage today - video edits, hip hop samples, photoshops
- a collage is the consumer and the creator combined
- GANs are the ultimate form of this (generator is the creator, discriminator is the consumer) - Dada 3.0
- AI generated images are kind of like a super collage of thousands of images
- Neural nets are just linear combinations
AI and Democracy
“Education is a founding principle of democracy, because a democracy is only as good as its voters, just as any algorithm is only as good as its data.”
Like a collage or a democracy, AI depends heavily on its source material. A medical algorithm is more accurate than any one doctor only because it represents the combined expertise of a thousand doctors. It can’t go beyond that and learn something that no human has ever learned before.
“The data is the fabric, and the code is just the stitching.”
Your data isn’t just something trivial you can give away; your data is who you are. It takes the culmination of many years to grow the height of the person whose data shows up on the driver’s license. In fact, some people can provide extremely unusual data from being a living edge case, and therefore push AI forward. However, this is only possible if the platforms are built centered around humans and what value they can bring to the table, rather than what we currently have: human labor is hidden behind AI
Just like Dada 1.0, AI is not the end of art
- Allows humans to spend time on their passions
- Allows us to focus on “pushing the boundaries” (going beyond what AI can do) in ways that are uniquely human
- surrealism wasn’t the end. abstract art wasn’t the end. rock and roll wasn’t the end, and hip hop won’t be. these things didn’t destroy art; they opened it up
“Meanwhile, untold amounts of venture capital get spent on rooms full of developers who make systems that exist to hide human labor under a veneer of technological progress.” “…but sometimes it is gig workers making $.01 per job, or an unsuspecting person on social media or simply walking with their phone in their pocket.”