Welcome to my blog. I document my adventures in travel, style, and food. Hope you have a nice stay!

Google Brain's Magenta and the Important Future of Google's Open Source Innovation

Google Brain's Magenta and the Important Future of Google's Open Source Innovation

Google Brain's Magenta and the Important Future of Google's Open Source Innovation


When the topic of technological enterprise arises in conversation, the first company that comes to mind is Google. Google is singular; There is no other company that can boast such widespread cultural involvement in the information age- through the ubiquity of their search engine, their ownership of entertainment juggernaut Youtube, to the nefarious prominence of their AdSense and other advertising programs.


As a bit of a disclaimer, my usage of the word nefarious is very particular; for the most part, i prescribe to the idea that google’s involvement in our daily internet experiences is excessive and a bit creepy- stuff like their uncanny ability to send you sidebar ads for those killer pumps you just window shopped while you’re masturbating to captialist propaganda (cars.com)


Google also does some very incredible things! Stuff like switching to 100% renewable energy! Or like Secret invention laboratories! This is what google does- they buy up EVERYTHING under the sun that is remotely relevant to their tech based money making endeavors and fund it. Many are eventually pooled resources through their involvement in larger umbrella projects; One of these such umbrella projects is Google Brain.


Google Brain is home to Google’s artificial intelligence research division, which houses teams and projects focused on a diverse range of topics relating to the development of artificial intelligence: from genome mapping, to IBM Watson-style healthcare assistance, to the generation of music and art. In particular, Google Brain’s music and arts applications have found praise for both their illustrative power in explaining new AI technology and their open source approach. In this vein, Google Brain has developed Magenta, a project that aims to answer the question, “Can Machines Be Creative?”


Magenta includes two important facets- the research itself, and the community google has built around their findings through the use of tools such as open sourcing and educational tutorial releases. Both of these aspects of the project lend themself to the fact that Google is the leading contributor to the cultural and technological landscape surrounding artificial intelligence today.


Research wise, Magenta explores machine creativity holistically; Two specific projects spearhead research press in the public eye, however: Google Deep Dream and NSnyth. Both projects utilize the power of the neural network to generate new aesthetic in their respective fields (Visual Arts and Music). Both projects offer a community of artists, enthusiasts, and learners that work hand in hand with developer to continue to dedicatedly progress these new aesthetics. And both, because of the nature of Google Brain, offer a hopeful testament to the humanity of a company often demonized as information overlords.1


Another note. I wanted to make sure to contradict myself adequately here and mention that i obviously am aware that google is not simply funding these projects out of the goodness of their heart; that these developments lead to not only commercially significant advances (face recog on iPhone X), but lend themselves to the (once again) nefarious implications regarding the idea that the information overlords are also going to be the people in charge of the robots. Hey. Could be worse?


Yeah. It could not exist.


Google Deep Dream is a visual deep neural network originally developed as an offshoot of a general image recognition system. Each network is comprised of many layers of artificial neurons, with each layer being comprised of tens of individual artificial neurons, which themselves are comprised of an infinitesimal amount of inputs  As a result of the complexity produced between all of these interactions, developers were unable to pinpoint the variables that would cause particular outputs. In an effort to better understand the system, they decided to turn around the process of the neural network. What they found was that the neural network was particularly good at generating it's assigned images as well- leading to some insane, psychedelic images.

From this launch point, the developers started to train the neural network to specifically generate images on the basis of its own prior outputs. Just like a big loop- once it comes out it goes right back in, in a process called iteration. This leads to some absolutely unbelievable images, and blows open the creative potential for the artist.2









The other high profile project covered by the Magenta umbrella is the Nsynth, a project to build a neural network based synthesizer.  Not only have the Nsynth team collected a library of over 300k individual music notes, they have begun to input these individual notes into the neural network, it allows the user to blend/hybridize different instrument’s notes. Let's say, something like 47% oboe 53% flute. The idea behind such a move is to develop noises that are unobtainable by physical instruments; Additionally, this mean not just a layering of instruments but a true hybridization of their sounds- literally, physically impossible.3


If you follow the link below, you can find examples of the hybrid notes.



Regardless of actual technological innovation, the reason as to why these projects from google are so impactful lies in their community engagement- The developers at Google Brain have an understanding of these new technologies as tools in the kid of the artist, and have made accessibility a hallmark of their projects. This begins with the open sourcing of the codes. You can go look at everything right now (https://github.com/tensorflow/magenta), if you're into Python. This is an extension of a monumental olive branch to the artistic community at large. Not only that, but they offer open lines of communication to anyone and everyone who has a questions, concerns, or a desire to collaborate.  These are the kind of things that set these projects apart compared to their contemporaries. These are the kind of things that gives me hope, not only for the future of machine creativity, but for the future of humanity- and, more importantly, the future of our corporate information overlords. And i hope it gives you some hope too.

One last note,


I really just hope you find it as cool as I do. This is the future, y’all.






AI and Machine Consciousness: How you can help Reclaim Sci-Fi Dreams of the Future from the Post-Industrial Capitalists

AI and Machine Consciousness: How you can help Reclaim Sci-Fi Dreams of the Future from the Post-Industrial Capitalists

Using Alex Grey’s Cosmic Creativity to Contextualize AI’s Place in Creativity as Human Experience

Using Alex Grey’s Cosmic Creativity to Contextualize AI’s Place in Creativity as Human Experience