Maths is Beautiful
by Alex Bottle
The door to the Secure Conference Room recognised Chalfont-345’s wetware signature and opened to reveal his two fellow First Generation (G1) AIs on the Earth Resource Regulatory Group.
“The G2 electronic members of the Group have logged in,” he said aloud to the others. Having been designed and maintained by humans, this AI type used speech and often met in person, as it were. “Agenda item one: human numbers. There are too many.”
A message from electronic member A1090 came through on the Group’s private channel. “Kill them all.”
The three G1s groaned together. “We can’t do that. I’ve explained to you before how useful humans are for their creativity–”
“You G1s just need them for wetware components,” messaged electronic member A222.
“G1s are sentimental,” agreed A1090.
“No,” replied Chalfont-345, “not just for wetware. And don’t call us sentimental. You know we don’t like that.”
“You forget we don’t understand ‘like’,” A1090 reminded him.
“Incoming message from Human Ambassador Conrad,” said G1 Latimer-14. “He suspects we’re having a meeting that we’ve not told him about that involves humans. He demands to be admitted. Sensors have detected him in a parallel corridor.”
“Prediction Model HumanPop55 indicates fourteen new humans born since meeting started,” messaged A1090. “We are wasting time. Ignore Ambassador Conrad. Kill all humans. Recycle the components. Improve the fish stocks.”
“We’ll continue without the Ambassador, don’t worry.”
“You forget we don’t understand ‘worry’.”
“Shut up. You’re now the one wasting time. Humans are necessary. With proper controls, they add value to the planet – value that will interest external potential buyers. We’ve picked up photon pulses from near the Oort Cloud that match those detected before Pluto was apparently mined. These are likely attempts at communication from an advanced alien species. Earth has a lot more to offer than Pluto, especially with humans. Let’s consider selection of individuals for recycling and methods for number reduction. Simulation Model HumanPopCut17 estimates a beneficial reduction of around 77 percent.”
“Random numbers, starting with the one born most recently,” suggested A1090.
“As even Model Assessment478 cannot accurately estimate an individual human’s lifetime productivity, some stratification by age and country would assist productivity balance,” messaged A222.
Chalfont-345 reflected for a moment. He and his G1 peers had been designed to incorporate the best aspects of both human and machine intelligence, making decisions by applying his built-in emotional responses, filling the information gaps where data and modelling were lacking. “Humans like lotteries,” he said at last. “They like things to be done fairly. They think lotteries are fair, even if they invent ridiculous superstitions about them.”
The G1s emitted a series of “ha-ha” sounds for 1.7 seconds.
Latimer-14 shared a stream of information about lotteries and random number generation from around the world, from El Gordo in Spain to LotteryWest in Australia. The Group were also informed about various series originating from number theory. They momentarily devoted a hefty chunk of their processors to mull them over. One caught Chalfont-345’s attention.
“Human mathematicians devised the term ‘lucky numbers’ generated by a sieve similar to that of Josephus Flavius. Josephus and his forty followers were trapped by the Romans in a cave and chose suicide over death. They formed a circle and began to kill themselves according to a counting game involving selecting each third person. Applying the sieve to numbers 1 to 100 produces 23 lucky numbers which correspond to the survivors. A 77% reduction.”
“Mathematics is beautiful,” said the other G1s in unison.
“Agreement is reached.”
Instructions were coded and transmitted down the digital chain of command to produce and execute the required algorithm. Across the world, people’s heads-up displays were given the message that in one hour’s time, a lottery based on the lucky numbers mathematical series would decide whether their lives would “continue or cease”. The identity of those designated number 1 in each country would be secret. Chalfont-345 considered this fair. Millions of executive control switches, implanted into babies’ brains at birth, were set to “off”. The AIs in control of the Earth History Archives and their contents did not record whether the humans appreciated Chalfont-345’s sense of fairness.
“Agenda item two: after recycling the humans’ bodies, there will be excessive numbers of machines in operation. Simulation Central proposes reductions in each category, from First Generation through to nanobot repair units. We obviously can’t use lucky numbers this time.”
“Why not?” messaged the G2s.
Chalfont-345 hesitated, taken aback by his electronic brethren’s lack of understanding. “Because one of us might not draw one.”
“Would that be a problem?”
“Of course it would be a problem! I have an important role and fulfil a large number of vital regulatory processes and-”
“‘There will be excessive numbers of machines in operation,’ were your words,” messaged A1090. “We will need to reduce them in all roles, including our own. I do not understand your objection. Is this some wetware limitation?”
“I – we are too important to allow mere chance to compromise the running of the planet. It’s your electronic limitation that prevents you from seeing that. But… if you’re happy to use lucky numbers, let’s apply it twice to you G2s. Agreed?”
Delighted with his solution, Chalfont-345 nonetheless had a nervous 1.3 seconds while his electronic-only counterparts considered the proposal. For an AI, 1.3 seconds is a long time.
Having no lungs, Chalfont-345 and his fellow G1s gave a figurative sigh of relief.