Mentifex Communicat: Assigning Associative Tags among AI Concepts

[menti...@gmail.com]

The Mens Latina Abra028A version is having trouble with the assignment of a tkb flag to link the verb of a Latin input with the direct object of the verb. This flag is necessary for the idea to be retrievable from memory in the future. The flags are assigned for the parts of a Subject-Verb-Object (SVO) sentence. Each word as it comes in is quickly instantiated as a subject at the tsj point, or as a verb at the tvb point, or as a direct object at the tdo point. As the remainder of the sentence or clause of input comes in, the LaParser() Latin parsing module retroactively inserts associative tags by using the SVO time-points to identify and gain access to the conceptual flag-panel of the subject or verb or object.

We go into the InStantiate module and we insert some code to load the tdo time-of-direct-object flag with the time-point when there is already a positive tsj time-of-subject flag and now a noun or pronoun is again being instantiated, but not as the object of a preposition. (Notate Bene, classici: Further stipulations may need to be included that the noun or pronoun is not an indirect object.) We run the Latin AI and we enter: "ego specto robotes" ("I watch robots"). After outputting eleven various thoughts, the AI suddenly and erroneously albeit encouragingly says: "TU SPECT ROBOTES". Clearly the AI is trying to say "You watch robots" but the LaVerbGen() module is not generating the proper form of the verb. Therefore in the LaVerbGen() module we add some code for forming a second-person singular verb from the stem of an infinitive ending in "-ARE". Again we input "ego specto robotes". The AI outputs eleven ruminations, of which the last is "EGO INTELLIGO TE" ("I understand you"). Thinking of the concept of "you" causes the SpreadAct() module to pass activation to ideas stored in memory with "you" as the subject, including the idea of watching robots. The Mens Latina outputs "TU SPECTAS ROBOTES" ("You watch robots") -- quod erat demonstrandum.

http://ai.neocities.org/MLPJ2019.html