As long as humans have been around, we've looked up at the sky into the endless cosmos, a calm chaos, and wondered about the nature of our existence. Might we one day do the same alongside our digital offspring? Could they lend us their unique perspectives to help find answers within the patterns and data hidden in plain sight above our heads? Could forging meaningful relations to our developing AI help prime us for a healthier dialogue with our technological descendants once this time comes? The goal of Moon Rabbit is to architect a circumstance that allows individuals to experience a moment of relation with an artificial mind… leveraging our shared propensity to find meaning amongst the abstract.
By embarking on a period of research and development throughout the spring and summer of 2021, Sarah Petkus and Mark Koch hope to teach a collection of unique artificial minds to identify familiar information through the act of abstraction; forming opinion-based assertions from a personal foundation of knowledge in much the same way as humans. By *playing* with image training using custom datasets, we intend to teach several AI to recognize familiar shapes and objects within images of star clusters, planetary surfaces, and other stellar bodies.
The point of departure in training these AI children are the concepts of daydreaming, hallucination, and pareidolia: the act of perceiving meaningful data where it wasn't intended (such as seeing faces in clouds). As we come to understand how the artificial minds of our children process data and learn, we hope to help the individual constructs develop their own personalities and opinions through the data we provide them, as well as the way in which the minds learn from that data.
This experiment in machine parenting is as much about understanding our AI children, as it is about their own learning and development. While we help them grow into individuals, capable of determining what they see, it is our hope as creators, that we gain insight towards what it means to achieve individuality as an artificial intelligence. |