吉恩·科根（Gene Kogan）谈信用分配的艺术。

作者:糖果
发布时间:2024-04-18 11:08:21 (1天前)
来源:https://foundation.app/blog

"Machine learning is the science of credit assignment." —Jürgen Schmidhuber
“机器学习是学分分配的科学。” -于尔根·施密德胡伯（JürgenSchmidhuber）
Working with AI, I'm often the last step in a long chain of events that encompasses literally billions of people.
与AI一起工作，通常是包括数十亿人的一长串事件中的最后一步。
From the countless content creators, mechanical turks, and innocent bystanders who make up a dataset, to the many builders who laid the groundwork for the technology we all use today, to the thousands of people maintaining the open-source software I rely on, to the handful of researchers and artists specializing in some obscure new puzzle that's fascinating me right now, there's no shortage of people who affect the outcome of my art.
从构成数据集的无数内容创建者，机械特克斯和无辜的旁观者，到为我们今天使用的技术奠定基础的众多构建者，再到成千上万的人维护着我所依赖的开源软件，少数研究人员和艺术家专门研究一些令我着迷的晦涩难解的新难题，现在不乏影响我艺术成果的人。
This forces me to think about who or what the creator of an artwork is. If it isn't all of them, how can it be me? With no pen, nor paintbrush, but a few clicks of my mouse, I summon a process inside a magic box that smashes terabytes of raw data together into one surprising image, sound, or text for me at that moment. That process began long before I left my mark on it. It's hard for me to take all the credit.
What excites me about crypto is its ability to capture the full provenance of that process. With NFTs inverting the ownership model of the internet from platforms to users, content is immutably linked to its creators. So when a machine learning model makes use of that content, it can assign credit to everyone who deserves it.
Indeed, crypto turns credit assignment itself into an art. As DAOs become more common, a new set of licensing models, co-ownership structures, and royalty schemes become feasible. They can align the incentives of ever-increasing numbers of people in sophisticated and novel ways. As they branch out into more creative enterprises, the notion of credit assignment will become increasingly important.
Just beyond the horizon are ArtDAOs—or DAOs that create art. When I first learned about the ArtDAO idea, I was inspired. To me, an ArtDAO is an artistic vehicle for the “hive mind.” It’s not a replacement for or alternative to a human artist, but rather a fundamentally different kind of artist altogether. Just as a distinct “voice” emerges from a choir of singers, the ArtDAO’s creativity flows from the collective imagination of its constituents.
Every journey begins with a step, and for me, that's adopting this technology for my personal work. With Foundation, I'm releasing the first in a series of short looping videos I've made over the last few years, which use neural style transfer to generate abstract textures that resemble famous artworks, patterns, maps, and various other subjects. They are notable for using a novel technique I developed for rendering them as perfect loops. 
I believe this technology is only just beginning to germinate, and I’m excited about where this journey will take me next. As happy as I am about sharing this work with collectors, selling my art is only stopping to smell the roses along the way. It was the ArtDAO idea that got me interested in crypto to begin with, and it’s the challenge that I intend to pursue next.

这迫使我考虑艺术品的创作者是谁。如果不是所有人，怎么可能是我？不用笔，也不用画笔，只需点击几下鼠标，我就在一个魔术盒中召唤了一个过程，该过程当时将TB级的原始数据粉碎成一个令人惊奇的图像，声音或文本。这个过程在我留下深刻印象之前就已经开始了。我很难获得所有荣誉。
令我兴奋的是，加密技术能够捕获该过程的全部出处。随着NFT将互联网的所有权模式从平台转变为用户，内容始终与创建者联系在一起。因此，当机器学习模型利用该内容时，它可以将功劳分配给所有应得的人。
确实，加密将信用分配本身变成了一门艺术。随着DAO的普及，一套新的许可模型，共同所有权结构和使用费计划变得可行。它们可以以新颖新颖的方式调整不断增长的人数的动机。随着它们扩展到更具创造力的企业，信用分配的概念将变得越来越重要。
ArtDAO（或创造艺术的DAO）就在眼前。当我第一次听说有关ArtDAO的想法，我的灵感。对我而言，ArtDAO是“蜂巢般的头脑”的艺术载体。它不是人类艺术家的替代品或替代品，而是完全不同的艺术家。就像歌手合唱团中出现了独特的“声音”一样，ArtDAO的创造力来自于其成员的集体想象力。
每一个旅程都是从一个步骤开始的，对我来说，就是将这种技术用于我的个人工作。随着基金会，我放开了第一个在一系列的短循环录像我已经在过去的几年中取得，它使用神经风格转移产生类似于著名的艺术品，图案，地图，以及其他各种问题进行抽象纹理。它们因使用我开发的将它们呈现为完美循环的新颖技术而著称。 
我相信这项技术才刚刚开始发芽，而我对这一旅程将带给我下一步的发展感到很兴奋。我很高兴与收藏家分享这份作品，而卖掉我的艺术品只是在途中停下来闻到玫瑰花的香气。正是ArtDAO的想法使我开始对加密感兴趣，这是我接下来要追求的挑战。

1 条回复

动动手指，沙发就是你的了！

登录后才能参与评论