The next generation in AI will utilize meaning to precisely carry out what is requested. This video shows how Patom brain theory was developed with its explanation that the brain uses patterns that work in combination with others.
It uses combinatorial patterns, not computations on encoded data.
Deep Symbolics should follow the statistical LLM phase as its meaning-based representation aligns with brain observations. Why does our brain store vision separately to sound to sensory perception and to motor control? The use of pattern-atoms is demonstrated with a simple model that combines these senses, leaving a common model that aligns with known brain function.
The potential applications address today's hard problems - understanding human language plus robotics tasks.
As current approaches favor computational models, there is potential for quick wins using the new approach that aligns with brain theory. For example, driverless cars have a long-tail of problems. Many of those relate to the inability of current technology to recognize objects in real time or in detail. That one problem alone has the potential to remove many impediments to greater adoption of robotic technologies like autonomous vehicles and perhaps even "Rosie the Robot."
