Bravo, Tutorial Doctor, you started from the right place. You said "We discern the difference between wrong and right", and this is the main difference between "learn" and "understand". All our current AI models, I refer to the learning machines and all the models that use neural networks, are able to learn things, not to understand them.
Let me introduce you to an example I've made in my blog (it's in Italian, but I'm just translating it in English): there are three reigns: the reign A, the reign B and the reign C. The first two are alwasys in war one each other, while the last one is unwarlike, and all its resources are driven toward scienses.
Suddenly, an earthquake shakes the ground in the realm A. There are collapsing buildings, thousands of deaths. Let us now give the "meaning" that the subjects of the three kingdoms do to the same event, namely the earthquake just happened.
In the realm A, certainly they will define it as "the catastrophe" and will remember that event certainly for so long as an "extremely negative event" The kingdom B, for its part, will define that earthquake as "a blessing from heaven" and it certainly will remember it for a long time "as the Divine that punished their warlike enemies". It 's very likely that the event has then changed the fortunes of war. So, for the subjects of the kingdom B the earthquake just happened is an "extremely positive event"
The subjects of the kingdom C, finally, define the event as an "extremely interesting event" because, with the subsequent large aftershocks that followed the big one, they have improved their understanding of the seismic event in general. Then they will remember it as an event "very positive" and appreciate the great contribution it made to their knowledge of the planet.
So, what is the real meaning of that hearthquake? Which of the three reigns gave it the effective meaning? I want to go deeper in the issue: do you really believe that a thing or an event could really be valued by logic? Is Logic able to say that an event is good or bad? Neither in thousands of years this could happen. Simply because only emotions can let use "decide" what is wrong and what is good. But good and wrong are things that we can define only with a point of view. Hearthquake, collisions of planets or the flutter of a butterly are only events for logic, nothing more.
If someone of you wants to go in deep and continue with me this topic, I'll be happy, because I'm developing (maybe) the thirst AI model that takes into account the main discovers of biology and psychology.