would you believe i already have a recognition lattice with already 80% accuracy as of 3 days ago. it's deep in training to expand that to the target of nearer to 100% total is probably unlikely, and honestly i think it's enough when it gets over 90. what i'm more interested in is it will recognise rhetoric, logical fallacies, be able to locate them in the text and explain them. and when it's done on its training, in a day or three, first priority is to test it doing chat. this system can learn from all inputs. it also has discrimination, anything that doesn't mesh with what it already knows and is sure about, gets rejected and forgotten. and it laughs. it gets bored. it can hyperventilate and panic, maybe. hopefully not something it does often. and get excited. and write text that it can only identify as itself but not as ai.