Schmidt pulled his still very drunk colleague tightly against himself, kissing her deeply. Right now he didn’t like her very much, even though the breath mint had made it possible to enjoy kissing her — too much. He could already see how this was going to be all his fault in the morning. The other side of that coin was that if he was going to be blamed anyway, he might as well make the most of it. Her mouth was fresh and cool. After her thirty years of killer professional training and experience, no pun intended, Cally O’Neal’s full frontal attentions packed one hell of a wallop. The way he saw it, he could take his life into his hands with her and her pissed-off grandfather in the morning, or with her pissed off self right now if he pushed her away. Under the circumstances, he wisely chose immediate gratification and deferred risk. Not that he wouldn’t have anyway, he admitted.
He had taken advantage of straight men’s notorious lack of decorating taste to avoid spending the money refurbishing his cover apartment. It was pretty awful; he usually hated staying here. Tonight promised to be an exception to that rule. It was a good thing she was at least buzzed. The dirty white shag carpeting and beat up faux-wood paneling inside would have put any sober woman off. He unlocked the door behind her and, without breaking the kiss, maneuvered her back through it.
Considering a male operative had to be competent to extract required information from a source in whatever way it took, including any female source who took a liking to him, he wasn’t too shabby at this game, himself. He was, therefore, stunned to find himself sitting on his ass on the floor, after about a minute and a half of practically doing each other in the doorway.
“I hate white shag carpeting,” she spat at him, whirling and slamming the door behind her.
He stared at the empty space where she had been a moment before, butt stinging where he had landed, reflecting that he had never understood a woman less in his whole life.
The battle began as a war between two artificial intelligences. The aim was not to destroy property nor yet to take lives. The aim, at the beginning, was to try to drive each one crazy.
Fortunately, buckleys were pretty much always that way.
Clarty was a good Africa hand and loved technology. But he did not understand it. Take for example the “IR sensors” scattered around the perimeter.
Clarty knew that they picked up on infrared emissions, heat that is, from the warmth of any mammalian critter. What he did not know was how they worked.
Any large mammal generates an awful lot of heat. In the case of humans, enough to melt fifty pounds of ice in one day. However, because of IR sensors, human soldiers, spies and burglars had long before come up with IR defeating systems. Heat cloaks, thermo-paint, IR static generators, they were all designed to reduce the IR signature of a person to that of, say, a rabbit.
And there were many rabbit-sized creatures in Africa or any other “natural” area.
For that matter there were many creatures that produced as much IR signature as a human. The very simplest systems would then scan for human contour and outline but a ghillie cloak changed that and the system never could tell the difference between a human and, say, a baboon.
So the makers of the IR sensors had a choice between a system that would produce thousands of false positives or a system that couldn’t spot a male human wearing the simplest of disguises.
Unless they threw in an AI. AIs could make “rational” judgments about whether there was a real threat or a hopping bunny.
But then there existed the question of just how to integrate the AI.
Still the only human AI, buckleys were the only choice. The IT geeks at the manufacturer understood the problems of buckleys far too well. Buckleys were notoriously unstable. Simply throwing a buckley onto the control interface was a sure recipe for disaster. And depending upon the number of IR sensors scattered around, one buckley might not be able to do all the decision making.
So Clarty’s system worked like this.
There were IR sensors. They were not “smart.” They were not “brilliant.” They didn’t have any processing to them at all beyond that necessary to convert IR into something a system could read.
Behind them, at the second tier, was a “smart” system that converted one or more sensors into data the buckley or buckleys could read.
At the very top of the hierarchy was a system that, based upon the number of sensors, generated medium emulation buckleys. These AIs would then consider the sensor data and determine if it was a real threat or not. They would sit there, day in and day out, looking at sensor data and deciding whether to cause an alarm.
Occasionally, they would get bored and cause a false alarm. The longer they were left in place, the more false alarms they would generate from sheer boredom. Each time the user reset the system, telling a “smart” program that it was a false alarm, it would be considered by a non-AI algorithm. When the buckleys got to a certain level of false positive, determined by the user or the overriding “smart” system, they would be reset and forget they had ever been there. And the cycle would start all over again. They would also be automatically reset if the “smart” system determined they were going AI gaga or if they started fighting incessantly, which was common.
None of this occurred at a level the user could see.
Clarty did not understand how his system worked.
DAG did.
That was the first of many differences of quality between the two groups. Differences of quality that had made coming up with an attack plan such a pain in the ass for Mosovich.
This was his first “real world” action with DAG. He wanted it to be professional, precise and good training. Because the best that could be said about Clarty’s unit of “pirates” and his setup was that it was going to be a good training op for DAG.
There were so many many choices. It really became a question of what sort of command personality Mosovich wanted to project.
He could start with orbital battle lasers, normally used to take out heavy Posleen infestations but fully on-call for an op like this, to take out the sentries. Then DAG would come in right behind them in choppers or even shuttles and hammer into the middle of the compound. Good estimate of take-down of the entire compound was one minute twenty-three seconds. Snipers scattered around to take any leakers. Satellite and UAV surveillance to make sure nobody got away.
Simple, brutal, effective.
Training level? Minimal. Joie de vivre level? Zero. Coolness level? In the negatives.
He had no intention of taking so much as one casualty, whatever he did. But with such a simple op, making it interesting had real command benefits.
So he decided to start with causing a nervous breakdown in the “automated” system.
Buckley Generated Personality 6.104.327.068 was beyond bored. He’d been looking at really boring African countryside for nearly seven boring hours which was, to an AI, approximately a gazillion years by his calculation. He’d calculated pi to a googleplex decimal points. He’d tried to log onto a MMORPG and gotten kicked for being an AI, the bastards. He’d gotten into a three point two second argument, about thirty years to a human, with Buckley Personality 4.127.531.144 over whether a sensor reading was a monkey or an abat. Since all they had were these stupid ZamarTech IR sensors, who knew? He couldn’t even ask anyone to check it out and adjust the system without setting of a bagillion alarms. They could have put in an interface that let the AI simply ask somebody to go tell them what something was, thereby increasing their functionality but noooo…
Now he was looking at another IR hit. The buckley did not “see” this as a human would; he did not see a smear of white on a black background. What the buckley received and processed was a large number of metrics. Horizontal area of total generated heat. Precise numerics of shape, thermal output fall-off, calculations of three-dimensional shape, vectors not only of the total blob but of portions. It then took all this information and compared it to a database of notable IR hits, ran all that through a complicated algorithm assigning a valid numeric likelihood of it being positive for a hostile human or animal then, at the last, applied “AI logic” to the situation.