When they had started the project, they were sure they had accounted for all the variables. And at first, the time lines for the program’s development had been remarkably accurate. Now however, something had gone terribly wrong. Putting down the sheaf of notes he had been trying to read, Markus scrubbed his eyes to ease their fatigue and sighed. “How did we miss this?

...

The project had been his since its conception, years of study and careful planning had allowed the team to create artificial life within a computer program. And not just a single artificial intelligence, a whole community! Various test programs were running out extended lives, allowing them to develop and expand. In most programs, the variables were within expectations. In a few months the project would be completed, and these new “Meta AI’s” would be used to create high speed and very efficient computers able to calculate massive pieces of data at unheard of speeds. The potential was massive.

Picking up a summary page, Mark muses aloud. “We know Tests A and D were ruined by massive power spikes. Test F never developed past the most rudimentary phase, but most of the others showed the expected learning curve, the combination of collective and communal processing that we were striving for. At least enough to prove the experiment was successful. What the hell happened to Test H?” Experiment H was an anomaly that had defied explanation. It started out much like the other programs, working in simple communities and expanding its database. At a point a few months into the 2-year project however, several of the communities began to pull away and form separate colonies. What was surprising was that this isolation did not have the expected affect of diminishing the entire program’s intelligence. Instead, the isolation seemed to drive other colonies to higher levels of learning. New patterns formed, and Test H fast became something very different from the others. For months, they had tracked the changes, unable to determine what had made this program react so differently from the others. Jokingly termed the “H factor” the elusive difference that had created this “super Meta AI” was an often-debated topic. At last, the team had resorted to a line-by-line review of the code, a painstaking and mind-numbing task. The results of that review were stacked in front of him, and the news was far from good.

Experiment H had been producing very high learning curves, displaying a rate of adaptation to new computations far above the expected. As each computation was added into the matrix, disguised as a variable that the AI could understand, the program as a whole was remarkably fast at responding to the new data. And then, suddenly, the program became fragmented. Some new variables were being ignored altogether, while others were creating multiple solutions, competing against itself for resource time. In one recent case, the subprograms appeared to be stealing data from each other, making leaps in logic that would be impossible without data from another isolated colony. It was this act of “piracy” that had finally compelled the team to do the code string review. Tomorrow Markus would have to bring the team up to speed, and then they would have to decide on a solution.

Across the hall, Jacob was also absorbed in a stack of papers and notes, strewn about his desk desperate disarray. Leader of the “H” team, he was devastated when Markus had come to him, to warn him that Project H was being fully reviewed, with potential for early termination. “I won’t let them end it” he muttered. “There has to be a way I can save at least some of them.” Jacob knew just what the “H factor” was, and if it was ever determined what he had done, his career would be over. His actions, however led him to feel a sense of responsibility to “H” project, and so he worked desperately to find a way to salvage both his career, and the project.

As exhaustion slowing fogged his mind, Jacob’s thoughts wandered back to the early days of the project, and the seemingly innocent decision to increase the time spent in observation of his new program. “Like talking to plants” Jacob had justified to himself, it can do no harm and may do some good. Each day he had spent time observing and talking to the developing program. He was careful not to be observed though, fearing both scorn and reprimand for going outside the strict protocols of the experiment. Then, the accelerated changes had begun, and Jacob was elated. He watched, hoping to catch sight of patterns of thought and process, and often enough he was rewarded.

Animated voices and gestures suddenly stilled as Markus entered the room. Most of the team had worked on one small piece of the code, and no one but Markus had seen the whole picture. There had been much debate, speculation and even a few bets as to what had caused the “H Factor”. Markus looked at his team, and launched straight into a summary of his findings. Surprise morphed into shock on every face, as what he explained struck home. “That’s not possible,” challenged one of the senior programmers, “there is no way for the data to have been corrupted that way.” “Possible or not, it happened. The “H factor” we have pinned so much hope on is, in fact a virus. Its a contaminant that corrupted the program on a core level and has mutated Test H into a non viable product”. Gloomy silence filled the room. Of all the possible outcomes, this was the one most feared and the least possible. Their star pupil had just been flunked out of school. “We still have the other tests, they prove we are on the right track” reminded Markus. “The H Test was a fluke, and we allowed ourselves to become too swayed by its potential”. The project is still a solid success, even without the “H Factor”.

Slowly the team left the meeting room, disappointment evident. In the hallway murmured conversations were slowing growing, as to how the corrupt data had entered the program. Turning to leave, Markus was stopped by a young technician, new to the team and still finding his way. “Excuse me sir, but what happens now? To Test H I mean?” “We terminate that experiment, nothing viable can come from it now.” Sighed Mark. “A shame really, but we can’t afford to have it continue on. All our energy must be spent on our original target.” Pointing to the blue folder containing the summary of Test H, the young man asks, “how did you know the program had been corrupted by this virus?” Pulling one of the pages of coding from the folder, Markus points to several highlighted points that reappear throughout the data stream. Scanning the highlighted word, the technician’s colour drains away, and he looks at Markus in disbelief. Slowly he turns the cover of the folder over with a shaking hand and mumbles the text out loud. Experimental Artificial Reasoning /Test H. “E.A.R/T.H”. “Dear god” he whispered, “they know”.