The AI ArtDAO (Decentralized Autonomous Organization) - as outlined by @trentmc0 in https://www.slideshare.net/TrentMcConaghy/tokens-complex-systems-and-nature
The AI ArtDAO (Decentralized Autonomous Organization) - as outlined by @trentmc0 in https://www.slideshare.net/TrentMcConaghy/tokens-complex-systems-and-nature
"Introducing Computational Creativity" - a brilliant overview presentation by Tony Veale:
https://www.slideshare.net/kimveale/introducing-computational-creativity
Building character AI through machine learning: https://medium.com/mit-media-lab/building-character-ai-through-machine-learning-7a3159dc4940
Role of the audience in creative arts is underestimated. When an artist performs, they've audience in front/back of their mind. Who do ANNs think about when performing?
Good question posed by Harsh Hemani. #ML #Generative #Creativity #Ideas
Office Voodoo: a real-time editing engine for an algorithmic sitcom (2003):
http://alumni.media.mit.edu/~lew/research/voodoo/Voodoo-Siggraph-03-sketch.pdf
Office Voodoo is an interactive film installation using exclusively live action footage and running on a real-time, shot-based editing engine that fluidly assembles the film as it is being watched, while respecting the conventions of continuity editing. Each character in the film is represented by a physical voodoo doll. As viewers manipulate these dolls, they affect the emotions of the people on screen. They can also call the people in the film using their phones.
Bored with your video game? Artificial intelligence could create new levels on the fly
"The British Library Talk: Part 1" - by @BrianEnoMusic :
Eno's conceptual points on "infinite media" (enabled by generative systems) are spot on. Eno measures the length of his musical pieces in millions of years - jokingly saying that "the holocene will be followed by the enocene". #Generative #Music
A strategy how to make computational creativity system outputs much better:
Give bots the ability to spend real-world money.
- Generative Music Bot? Let it buy new samples.
- Generative Design Bot? Let it buy new fonts.
- Generative Journalist Bot? Let it hire mechanical turkers.
While this strategy might seems like a "hack" from a "pure", academic computational creativity perspective - it is precisely such hacks that will make generative system outputs become "good enough" across many creative disciplines in the near term.
The "Remix.Army" and "DadaBots: Socially Automated Dadaist Music Remix Bots" presentations by @dadabots from a few years ago are still super fun!
"So before we start thinking about how evil machines are, really this is just meta music - instead of playing the music, we are playing the musician" - @dadabots
Algorithm Automatically Spots 'Face Swaps' In Videos:
https://www.technologyreview.com/s/610784/this-algorithm-automatically-spots-face-swaps-in-videos/
The new technique could help identify forged videos as they are posted to the web. But the work also has sting in the tail. The same deep-learning technique that can spot face-swap videos can also be used to improve the quality of face swaps in the first place -- and that could make them harder to detect.
FaceForensics: A Large-scale Video Dataset for Forgery Detection in Human Faces:
https://arxiv.org/abs/1803.09179
The Screen Actors Guild-American Federation of Television and Radio Artists (SAG-AFTRA) says it's "fighting back" against the dangers posed by new face-swapping technologies:
https://deadline.com/2018/04/deepfaking-technology-sag-aftra-actors-union-fighting-back-1202371117/
A deepfake homage to Grace Kelly, with Grace Kelly's face replacing Nicole Kidman's:
Sheldon County is a procedurally generated podcast about life in a simulated American county that inhabits your listening device, with just a single listener: you. It is made by James Ryan (www.jamesryan.world). Podcast coming in 2019 but you can listen to some proof-of-concept episodes online already: #Generative #Music #AGB
DeepWriting: Making Digital Ink Editable via Deep Generative Modeling: #ML #Generative
Project: https://ait.ethz.ch/projects/2018/deepwriting/
Code: https://github.com/emreaksan/deepwriting
How Music Generated by Artificial Intelligence Is Reshaping -- Not Destroying -- The Industry: https://www.billboard.com/biz/articles/news/legal-and-management/8350905/how-music-generated-by-artificial-intelligence-is #Music #ML #Generative
ResGAN: Speech synthesis from MFCC sequences with generative adversarial networks:
Paper: https://arxiv.org/abs/1804.00920
Samples: https://users.aalto.fi/~ljuvela/mfcc_vocoding/
Code: https://github.com/ljuvela/ResGAN
Dream Lens: "Exploration and Visualization of Large-Scale Generative Design Datasets" - by Autodesk Research: https://www.autodeskresearch.com/publications/dreamlens
Generative design browsers are a key (missing) component for making generative processes go mainstream. The "Dream Lens" browser combines many relevant ideas, applicable to generative design tasks in other creative disciplines as well.
I find it comedic how professional digital creative tools (video, music, design, etc.) in 2018 are still being mainly promoted as "tools for hand-crafted, artisanal creation" - while we are living in an age where bot-nets are producing and recommending artefacts by the trillions.
While the quality of creative artefacts generated with machine assistance is debatable - it is quickly getting "good enough" across many fields. Combined with other unique benefits of the generative model, traditional digital creative tools are becoming obsolete fast.
The outlines of the new creative economy are clear: 90% of all content will be machine generated (with little to no human intervention) and extremely cheap to buy. The remaining 10% will be artisanally made by humans, which focus mainly on branding and storytelling. Technologically, you don't need "AI" or very sophisticated machine learning for most of this to happen - good old creative computation (at massive scale) does the trick.
Finally, a note to the people saying "no worries, AI + creativity is all about empowering and augmenting human creators, there won't be blood". I've worked in this industry for years & find such talk intellectually lazy and at times even deceptive. There will be blood.
I am not questioning that AI can and is empowering some creators and enable wonderful new forms of creation - it will. But to see what happens with such tech when at industrial scale, one can look at fake-news and spam generation bots. That's a more realistic view of whats coming to all creative industries, operating under our current global economic paradigm.
The generative creative economy won't be a utopia by any stretch, but likely a messy hell-hole for many creatives - not unlike the situation right now. We have to actively shape this future by fighting vested interests in the system. Yes, human jobs destroyed through growing automation capabilities can be replaced with entirely new jobs. But it won't happen magically - its a major generational effort which requires deep investment in education and culture. Human creators need to be recognised, not hidden by AI. Under the current klepto-capitalist global system, that seems highly unlikely.
If we want more positive outcomes, the "why" needs to be addressed properly, beyond our current semi-pornographic obsession with the "how".
"Is this concern real or just doom & gloom? Is there a concrete example?":
https://samim.io/p/2018-05-26-my-recent-piece-thoughts-on-the-generative-ai-creative/
Thoughts on Botnets and Creativity:
https://samim.io/p/2018-04-23-the-solution-to-exploding-fake-news-and-spam-botnets-ca/
Deep Painterly Harmonization - Paper: https://arxiv.org/abs/1804.03189
Code: https://github.com/luanfujun/deep-painterly-harmonization #ML #Generative #Art
KidPen: A Stroke-based Method for Kid-style Sketches Synthesis from Photos:
https://dl.acm.org/citation.cfm?id=3149434 (paywalled. force open access version: https://sci-hub.tw/https://dl.acm.org/citation.cfm?id=3149434) #ML #Generative
Two-Stream Convolutional Networks for Dynamic Texture Synthesis: #ML #Generative
Paper: https://arxiv.org/abs/1706.06982v4
Project: https://ryersonvisionlab.github.io/two-stream-projpage/
Code: https://github.com/ryersonvisionlab/two-stream-dyntex-synth
Video Based Reconstruction of 3D People Models: https://arxiv.org/abs/1803.04758