Creatures is a speculative model for a new paradigm in designing artificial intelligence.
The core pretext of this new system is based on a tried-and-true method: evolution. It proposes seeding base intelligence in sensor-rich hardware, which are then propagated throughout various environments. This will allow the creatures to develop in response to dynamic physical and social landscape. Through embodying these new modes of intelligence as physical manifestations, humans and AI will both benefit from a level of communication and interaction that is unseen in our current methods.
Ultimately, it seeks to address the hermetic and disembodied nature of Human-AI interaction that has a tendency to breed distrust and spread misinformation. While the technologies that could enable these new explorations may still be a long way away, the fundamental principles of transparency and openness can be applied to our current research of AI systems.
Project collaborator: Carmen Schweizer
Faculty Advisor: Paolo Cardini, with special thanks to Bathsheba Demuth and Dr. Elizabeth Brainerd