Free Into the Darkness: Crimson Worlds Refugees I by Jay Allan Page B

pointing to a series of waves on the screen. “I’ve done the calculations a hundred different ways. There is no discernible logical pattern in how these machines process information. It almost looks…random.”
    She stared at the ‘pad for a few seconds. He was right. The patterns he’d isolated weren’t patterns at all, at least not in the sense that they had any predictability to them. “But it can’t be random. That doesn’t make sense.”
    “No, not random. I’m sure these crazy waves make sense…but only if you can see the input data that generated the recorded patterns.” He looked up at her, and he could see she wasn’t following. “Don’t you see? These First Imperium intelligences—at least the lower level ones in these robots—aren’t truly sentient. But they able to comprehend how sentient beings act and react.”
    He looked up and sighed again, clearly frustrated that she wasn’t understanding his point. “Imagine you are in a trench fighting a battle. You look across at your opponent, who is also in a trench. If you project that he will behave rationally, you can assign probabilities to various actions he might initiate. For example, you might determine that he is most likely to remain in place, since charging correlates with a large advantage for you defending in your trench—and a high likelihood of death for him.
    “But how do you account for a madman unconcerned with physical harm? Or a religious fanatic who believes death in battle leads to paradise? Or simply a soldier acting irrationally due to battle fatigue or rage at the loss of a friend? How do you assess the potential of irrational actions by an opponent? Our own AIs do this by assigning probabilities based on historical data programmed into their memory banks. For example, we know that certain Caliphate units exhibit fanatical behavior in battle, and our AIs know this, so they assign a higher probability of such behavior to their forces. But the First Imperium has no historical data of this sort for us…or at least very little.”
    He looked up at her. “I postulate that the First Imperium AIs do not address emotionally-driven and similar illogical decisions simply by assigning likelihoods of a specific action based on static historical data. I believe they understand irrationality and emotion, even if they themselves do not actually experience such traits.”
    Anastasia began nodding. She understood what he was getting at…sort of. “Is it possible to…understand…irrationality when you yourself are not capable of truly experiencing it?” She had forgotten all about Volga and superpower pride. Unraveling the secrets of the amazing machines that ran the First Imperium was the most fascinating task she’d ever attempted—and one that could have profound impact, both on the fate of the fleet, and indeed, on all humanity. Perhaps with an improved understanding they could communicate with the First Imperium, negotiate a peace of some kind. Or, if not, a true comprehension of their enemies could be weaponized too. Understanding how they “thought” was the first step toward defeating them.
    “We’re hypothesizing, of course, but I would say no, not fully. I believe the First Imperium intelligences are able to do a credible job of anticipating and responding to the irrational or random behavior of their adversaries. Thus, they are not utterly confused by actions driven by emotion. However, I feel this is where their weakness exists as well. Not being able to truly experience the kinds of emotions that lead to irrationality, they are only partially capable of crafting targeted responses. This explains the success of commanders like General Cain or Admirals Garret and Compton in actions against them. These men are highly skilled and, for the most part, logical. But they all rely on intuition as well, and on other emotions--stubbornness, anger at the losses they suffer, pride in their units and warriors. The First Imperium