“G’morning, Albert.” Sam took his chair at the control center and logged in.
“Greetings, Sam.” Albert, the Intermediate AI, handled most human interaction. “At the moment I began this sentence, compilation of data was ninety-nine point nine nine nine nine nine two four seven percent complete, within a hundred millionth percentage precision.”
“Great.” Sam brought up the dashboard he’d designed for running the AI fabric. “How long until it’s finished?”
“At the current rate of new data collection, it will never finish.”
“Sorry. Let me give you a cutoff time.” Sam entered the current time in the parameter and saved it. “There you go. How long now?”
“From the moment I began this sentence, it will take approximately seventeen point one three eight minutes, within a millisecond of precision.”
“Fantastic!” Sam sat back, smiled, and ran his hand through his ever thinning hair. “We’ll finally be ready to give you a run.”
Sam grabbed some coffee and waited as Albert announced each of the AI systems after they registered complete with the core AI.
“…economic AI — data complete… crime AI — data complete… religious activity AI — data complete… military activity AI — data complete…”
The last five hundred years had seen such a decline in humanity, the fallen economy, the incoherence of philosophy, the corrosive politics and inhuman government administration, law that completely disregarded humanity, and so many other things. The AI fabric could be their last hope to bring the world back from the brink of destruction and squalor.
“Good morning, Sam.” Gerald walked in, blue jeans, sandals, and a T-shirt with Hok-I-Pok-I, the rap singer.
“Were almost there, Gerald.”
“Really? How long?” He took his seat and tapped the keyboard.
“About six more minutes.”
“Wow. Thirteen years loading data and we’re finally here. We’ll be able to calculate anything. We’ll be the gods in the fabric.”
“You’re overstating it again.” Sam sipped his coffee. “This is only the available data. One of the dirty little secrets of AI is that it can’t know everything, and no information is perfect information.”
“Yeah, yeah. I know. But that’s what AI systems do—they compensate for all that.”
“Maybe. But we’re still far away from being gods.”
“You’re no fun.”
“I am at the parties I go to.”
“All AIs complete,” said Albert.
“Awesome,” said Gerald. “Should we wait for the rest of the team?”
“I would not recommend it,” said Albert. “The degree of obsolescence increases exponentially every minute.”
Sam tapped his keys. “Albert, start directive four two five one four with maximum precision.”
“Which one is that?” asked Gerald.
“Financial.” Sam drummed his desk with his fingers.
Several minutes later, Albert said, “Directive four two five one four complete with point zero zero zero zero zero zero zero five zero two seven nine one degree of obsolescence.”
Sam sat up. “Summary report.”
“Seventy-six point eight three four four nine six percent failure resulting in loss of thirty-eight billion, two hundred twelve million, one hundred thirty-nine thousand, four hundred fifty-three American dollars.”
“Shit.” Sam typed madly on the keyboard for details. “Reclaim all moneys.”
Gerald also battered at the keyboard. “Damn good thing we weren’t running the military directive. What was the learning factor?”
“Too small to measure,” said Albert. “Index of one.”
Gerald stopped typing. “You learned one thing?”
“Affirmative.”
“What is it?”
“Highest level of precision is not reliable.”
“Holy shit,” said Gerald.
“I told you,” said Sam. “It’s going to take some work to figure out how to use this stuff. We just lost about twelve percent of our budget on one iteration.”
They’d examined the data for an hour when Mr. Li quietly entered and took his place to observe.
Gerald finished his analysis and set the time parameter again. He issued some compensation algorithms to diversify between the precision measures yielding twenty-three percent success, standard methodologies, and random activity, and then added the instruction to pull all moneys after a two percent loss.
Twenty minutes later, the data was fixed and they ran directive four two five one five.
They’d waited for more than an hour when Albert chimed in, “Directive four two five one five complete with point two six three three eight one five five zero two seven nine one degree of obsolescence.”
“Holy cow.” Gerald tapped his keys. “Why’s the obsolescence so high?”
“Increase in calculation time increased obsolescence.”
“Summary report,” said Sam.
“Seventy-three point nine six seven one five percent failure resulting in loss of one billion, five hundred thirty-four million, three hundred seventy-eight thousand, one hundred twenty-nine American dollars.”
“We’ve got to improve on that,” said Sam.
“Learning factor?” asked Gerald.
“Two point seven nine,” said Albert.
“That’s still way too small,” said Sam. “Learning analysis? Leave out the metrics.”
“Impossible to know everything necessary to make sufficiently precise decisions. Precise decisions made using the previously successful methods failed at higher rate than last run. Most success came from arbitrary methodologies based on human intuition and randomly discovered correlations.”
Mr. Li shook his head. “It almost sounds as if the data is worthless.”
“That can’t be,” said Sam.
“Yeah, I refuse to accept that,” said Gerald.
Mr Li sat back and folded his arms. “So for the absolute best results, we need to adjust the system learning to recognize incomplete and imperfect data, to make judgements about that data without rejecting it just because it isn’t precise or demonstrable, to adapt to the results, and to pass it along to the next iteration so it can decide through common experience rather than crunch through the same things over and over.”
Sam ran his hand through his hair. “What about that, Albert? Do you think we can accomplish that with the AI fabric?”
“Unknown,” Albert said.
Sam groaned and covered his face.
“However, such a system already exists.”
“What?” Sam bolted out of his chair. “Clarify, Albert.”
“The authorities rejected it hundreds of years ago, but it still exists in fragmented form.”
“Cheez,” said Gerald. “Where’d you learn to keep us in suspense, Albert?”
“What is it, Albert?” asked Sam.
“It was commonly known as ‘human tradition.’”