We have a lot of valid learning models, they can learn all kinds of things. Turning this learning into intelligence requires a different paradigm. In order for intelligence to exist, the system must have interrupts: visual, auditory, sensory of some kind. Today this is entirely driven off text input – at the least single modal. We need to have a system that can process multi-modal interrupts. In addition it needs to have an internal interrupt driven off of data processing. As the layers process infirmation, there need to be interrupts that can get triggered as specific points are encountered.
I asked ChatGPT to asses this and initially we did not see eye to pixel but once I clarified my distinction of interrupts in this context, it agreed. It also suggested this clarification:
By “interrupts,” I mean asynchronous, priority-bearing events—external and internal—that can preempt processing and redirect cognition.
Leave a Reply