**THOUGHTS ON BIOLOGICALLY INSPIRED AI**  EVOLUTION AND INTELLIGENCE ============================================================== Evolution does not optimize for intelligence. Human intelligence is just one of the side effects, also influenced by random events (like the dinosaur extinction caused by natural cataclysms). There are plenty of examples where "dumb" biological systems dominated for extended periods of time. I would rather say the goal of AI should be the study of general (often mathematical or computational) principles that lead to the emergence of intelligent behavior. I'm not saying we should not study biology, but that it can be distracting and lead us in wrong directions if our goal is to spawn AGI. ARTIFICIAL INTELLIGENCE DOES NOT NEED TO COPY NATURE ============================================================== The bird/airplane example is a useful analogy. In the early days of aviation airplanes even flapped their wings. We could study the structure of bird feathers for decades and make no progress. Instead the distillation of the fundamental principles behind flight (that were based on mathematical formulation of the laws of aerodynamics and computational simulations) lead to today's airplane technology. Similarly, we can spend a century trying to distill signal from noise in a biological system, focusing on irrelevant details, and make zero progress towards creating intelligent machines. We should look at examples of natural intelligence, so biology is an important subject for AI scientists to pay attention to, but real progress comes from understanding mathematical principles and computational processes that generate intelligent behavior. The space of such mathematical and algorithmic structures is far larger than the particular path that our resource limited evolutionary development followed. There are countless more optimal routes that can be explored instead. The resulting AGI will be very different from our human intelligence, or that of other species, just like airplanes fly differently from birds. I would say that is actually more exciting since it will give us alternative insights into nature of reality as well. If we copy biology we might be constraining our search space in ways that will limit the outcome. It would be far more exciting if AI could go beyond, and into orthogonal directions to our way of thinking, instead of being an artificial emulation of the type of intelligence generated by biological evolution on our planet. I think that a more substrate-free exploration with unconstrained mathematical and algorithmic ideas is a more promising and exciting approach. FINAL REMARKS ============================================================== In summary, there is no need to copy or even understand the brain and associated biological and evolutionary processes that led to its development in order to make thinking machines. However, I do believe we need other components that are still unavailable in current AI systems. Nature can provide hints to what the missing pieces might be.