On exciting and protopian futures - This is an R&D update that explains why I got excited about the future of AI and deep learning in 2005 and wrote a script about future intelligent crystalline forms as symbiotic home environments. Instead we have millions of cells in two groups-bicep and tricep-that facilitate the motion, where robots may have a few hydraulic cylinders, servos or stepping motors. A lot of us have been fixated on nerve-by-nerve and synapse-by-synapse behavior, but what if we are complicating the behavior of a much simpler structure? Perhaps instead of saying "Deep learning architectures model human neocortical architectures", we say Deep learning architectures and Human Neocortical architecture are different expressions of the same function in the same manner that robot arms and human arms are both n-degree motion actuation systems? If you go by cortical mini-columns, the number of computational units present in the neocortex decrease by two orders of magnitude, which makes sense, since we don't have one huge muscle cell pulling on our radius and ulna. Now lets look at human cortical structure. Sure it's not as comprehensive or efficient as an average human arm, but doesn't it still mathematically represent the same thing? Something like 11 axes or so of motion? Look at what it takes to make a robot arm.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |