Computers now have a better way to depict our mental condition. Given any thought that comes to mind, the next thought might connect along any of a large array of other connected thoughts. Traditional linear media like text and film can only capture a single subsequent connection. Nonlinear computer media can offer an array of connections but until recently the relationship between thoughts had to be very explicitly stated using things like if statements.

Now machine learning networks not only allow those if statements to be replaced by much more subtle connections using millions of weights between nodes, but those weights are also automatically set, not by explicit and articulated rules, but by example. This allows computers to capture what we implicitly know which is vaster and more interesting than what we explicitly know. This is probably a big deal.

You can pretty easily create one of these machine learning networks. In this session we will talk about making GAN machine learning models using a program called Runway. For instance instead of a one dimensional film of your face you can make a 512 dimensional space of your faceness.

Then we will look at how to use p5 to create “vectors” to travel through those 512 dimensions to explore what is called the latent space of versions of your face that never existed but very convincingly look like they could have (https://thispersondoesnotexist.com/).

All great and interesting so far. But what decides where is your next move in those 512 different directions. Your mind and body are pretty opinionated and out of millions of possibilities, the next thought to think is automatically served up. It is amazing. In a lot of examples you will find show off these spaces with a random walk through them which is cool but ultimately feels like a tech demo. What motivates our directions through these spaces is really an open ended question. Can we create landmarks and promising directions in the space. I will offer up some speculation around a sort of social conversation as a motivation to find each other in the space. But this session is really more about asking the question than answering it.

Notes: https://docs.google.com/presentation/d/1-8HQwwxgVL3TAPiqrctp3Sqez2sNpQ6kWPIXj2rU6Qw/edit#slide=id.ge1f32157a0020