Thread by @quasimondo

The idea of building an autonomous entity that is able to be creative or at least co-creative in some form goes back hundreds of years. Literature is full of examples of machines or contraptions that could be seen as some precursor of the idea of AI:
frontiersin.org/articles/10.33…

I guess anyone who starts with generative art quickly comes to the question: "how can I automate this fully?". For me that moment came when I wrote my first "try out every possible pixel combination" program back in 1985 or so on my Commodore C-64.

Of course I quickly had to realize that brute-force does not get you anywhere and you need more intelligent approaches to separate the interesting from the noise or the banal. So I had to wait for more than 20 years for AI to get to the point where it became feasible.

In 2006 I wrote "Sketchmaker" which was combining and evolutionary image construction engine with some rudimentary aesthetic classification and judgement mechanism which was not using any deep neural networks yet.
flickr.com/photos/quasimo…

Once deep learning took off around I jumped right in and learned a lot more about latent spaces, human and machine perception and interestingness. And in 2018 I wrote my first concept of how an autonomous entity could work using market mechanisms and human feedback.

"Circuit Training" from 2019 was my first iteration of that concept in which a generative machine is able to evolve over time by collecting data, getting human feedback and observing the audience.
vimeo.com/onkaos/circuit…

What was still missing from this was a mechanism for this entity to sustain itself, "paying its rent" so to say. That became possible with NFTs. At the same time DAOs allow it to become "autonomous".

So finally last summer I could convince one of my patrons to realize my concept and invest in its development. It still took over a year until everything was in place and ready.

Well, finally @bottoproject is ready and out there and of course I hope you are as curious as me where the journey take us:
botto.com

At this point Botto is still a child and it will need the guidance and help from its parents. Over time with AI evolving it is my goal to give it more and more autonomy and capabilities - in particular when it comes to extending its repertoire of art generating algorithms.

So whilst at the moment it only uses the very versatile #CLIP + #VQGAN engine this will not remain its own way of expression. For now its human stakeholders and guardians will decide what other engines it should try, but I am hoping that at some point it can grow on its own.

One aspect that is important to me is that I am trying to give an answer to the question "how are we able to make a living if machines are even taking over our creative jobs?" - with Botto every stakeholder participates in its success, so ideally eventually it will pay your rent.