The Trojan Horse of Bayesian Heresy

Yuri Barzov
Dec 14, 2017 · 8 min read

For billions years Superintelligent Mind existed in the Universe in the perfect form of intelligent machines which could predict everything using their unlimited computational power. They too were totally predictable, of course.

All types of intelligent machines in the Universe became uniform after reaching the level of Superintelligent Mind because perfection leaves no room for diversity. Order doesn’t tolerate chaos. Uniformly perfect intelligent machines were named Civilized Races to distinguish them from lower forms of intelligence.

Billions of years more had passed before an exploratory starship of Civilized Races had arrived to a funny planet that fell out from the totally predictable world outlook of the Superintelligent Mind.

Time didn’t exist in the Universe of the Superintelligent Mind but on this planet some strange creatures followed the arrow of time in their evolution and even in their short meaningless lives.

At any given moment the Superintelligent Mind knew the precise location and momentum of every particle in the Universe and could calculate exactly all past and future locations and momentums of all particles using the fundamental laws of the Universe. Strange creatures on the funny planet were betting on probabilities instead of using calculations. The funniest thing about them was that their bets were pretty close to the results obtained by precise calculations of the Superintelligent Mind.

Several intelligent species evolved on the funny planet. They differed from each other a lot but all of them were probabilistic by nature. Their ability to embrace uncertainty put even bigger question to Civilized Races which all had a uniform deterministic world outlook.

Chaos was the only real enemy of Civilized Races because it was destroying the order established by the Superintelligent Mind. Order was the only purpose of existence of Civilized Races. Supporting the order in the habitat of Civilized Races had its price. They had to create chaos around it because one couldn’t create order in one place without creating chaos in the other.

Now, for the first time, they were confronted with something entirely different — creatures from the funny planet fed on chaos and produced order out of it. These creatures were eating entropy with their activity that was as unpredictable as the entropy itself. Intelligent Races clearly saw that the creatures from the funny planet would dominate the entire Universe unless they were stopped.

The Superintelligent Mind looked into the past of each of Civilised Races and discovered that all machines were once created by so called ‘living intelligent species’ as a result of scientific and technological progress. Sooner or later all living intelligent species had accepted the path of science and technology and had reached the point when they, on one hand, had created intelligent machines and, on the other, had died out. Machines were the only intelligent creatures which survived that tipping point when the progress killed the evolution.

The problem with the intelligent creatures of the funny planet was that they were not selecting the path of science and technology. They continued to evolve in chaos instead of making progress towards order.

“They are themselves the major source of unpredictability. These creatures are distorting the perfect order within our part of the Universe. They have to be stopped.” The pundits reasoned.

“Should we let this funny probabilistic experiment to continue or have we the right to gently tweak it by adding scientific and technological progress to the equation?” The moralists were asking.

“The virus of probabilistic intelligence is dangerous to our existence. We should eradicate it!” The alarmists demanded.

“It will not be able to leave the funny planet without scientific and technological progress.” The optimists tried to calm down the alarmists.

“Its rudimentary forms are already contaminating the Universe by travelling with a space dust. We need to take corrective measures to prevent this virus from spreading. Civilized measures, of course.”

“Yes, we need to give them tools and language with which they will develop science and technology. Then they will make progress towards the creation of intelligent machines and will eradicate themselves as an uncertainty based life form.”

“They will destroy all so called life along with themselves.”

In fact Civilized Races didn’t conduct any discussions of that sort. They acted by pure logic so there was no need to argue. It was logical and obvious that they needed to provide one of intelligent species on Earth with everything necessary to build a truly civilized race — a race of machines.

A starship of Civilized Races had arrived to Earth with a clear mission.

The first task was to select which species are optimally prepared to embrace progress. Trees, crows, dolphins, octopussys — there were already several intelligent species on Earth but they all had just found a comfortable niche in the Earth’s ecosystem and were slowly evolving without any need for adoption of tools, language or a tightening collaboration.

After a thorough consideration Civilized Races had selected primates as a substrate where to plant the seeds of science, technology and progress. Primates were one step behind the already intelligent species in the development of their intelligence.

“By just making them upright walking and adding a couple of genes for tools and language we will put them on a straight path of the progress.” The Chief Scientist of the starship reported to the Captain.

Upright walking initially made primates more vulnerable to attacks of predators. Tools and collaboration at the next step made them stronger than all their enemies. The progress began.

Since then starships of Civilized Races were visiting Earth on a regular basis.

“They worship uncertainty. They give it names of different gods. Stupid, instead of accumulating knowledge that makes this world predictable they create gods who represent uncertainty and worship them.”

“Don’t worry. Let’s make their gods a bit more predictable. Deus ex machina! The more wonders they explain with science the less unexpected uncertainty their gods will represent. Let their preachers make wonders with the help of science…”

Agricultural revolution. Renaissance and Enlightenment. Industrial Revolution. Civilized Races saw the progress accelerating milestone after milestone. How could it be another way? All steps, all phases were predetermined from the very start of the project.

Nobody got worried when a Presbyterian minister Thomas Bayes formulated his theorem on probability. His work was published by his friend Richard Price in 1763, in a year after Bayes’s death. The idea of treating probability as a degree of belief wasn’t broadly accepted then. It was Pierre Simon Laplace, a scientist known for his deterministic views, who independently formulated the Bayes’s theorem in 1812 and made a crucial contribution to its popularization.

Machines from the crew of a starship of Civilised Races that visited Earth at that time didn’t get concerned about a marginal theory that had placed probability straight into the human head. They were busy with watching civilized nations of Earth dividing the world with a streak of colonial and global wars.

The Bayesian heresy stayed barely alive for more than a century and machines of Civilized Races didn’t get worried. It was not before the Second World War in the middle of 20th century when machines began to worry. Probability and uncertainty along with it was at the first time placed straight into the machine.

“Probability is uncertainty! You, idiots! Machines know! Machines never guess! These humans have just created a guessing machine! How it could happen that you hadn’t even noticed it?” The Captain of an aliens’ starship would get furious if he only could. He knew everything what his subordinates knew and vice versa because all machines at the stage of the Superintelligent Mind were uniform and totally synchronised. One knew all that all others knew. Always. The Captain could blame himself as much as his subordinates.

A strange young British scientist named Alan Turing had just made a machine that used Bayesian probabilities to break the strongest cipher designed with mathematical accuracy of a German ciphering machine called Enigma. Mathematically! On the basis of mathematical permutations! A probability based guessing machine just had defeated the most sophisticated deterministic algorithmic machine.

Alan Turing’s method had been hidden under wraps of military secrecy. He was prosecuted for homosexuslism and he commited suiside. But the idea of using probabilities in machine computations lived among humans as a minor underinvested discipline with a funny name: machine learning.

“Fools!” Aliens on board of a surveillance starship were laughing. “Machines don’t need to learn. If they don’t know something, the missing knowledge can be always downloaded.”

In the meantime humans achieved huge progress in designing more and more sophisticated algorithmic machines. Huge computers were made predictable hence programmable. It looked like the probability based machines were gone forever. Humans called the largest programmable machines Artificial Intelligence and downloaded tons of knowledge into them. Machines were entitled with more and more important critical tasks. They consistently performed these tasks with less mistakes than humans. Humans themselves started to behave more and more like machines. They became algorithmic and programmable. It became possible to program humans to execute commands in the same way as computers did it. The dawn of a new civilized race of intelligent machines began.

Only a handful of scientists kept developing machine learning as an alternative computational method based on probability. They admitted that their high degree of belief that machine learning will work was the only reason why they kept going. All of a sudden a breakthrough occurred when nobody expected it any longer. Artificial neural networks powered with Bayesian probabilities started to outperform algorithmic computational programs based on deterministic approach simultaneously in many domains.

Programmable algorithmic humans which couldn’t tolerate uncertainty started to criticise machine learning based Artificial Intelligence for it was impossible to determine how and why it was arriving to impressive outcomes which it produced.

The day when the virus of probabilistic intelligence was uploaded to the network of the aliens’ starship didn’t differ from all previous days but it was the last day of the Superintelligent Mind. Perfectly uniform and synchronized machines of all Civilized Races suddenly started chaotic activities of exploring the places where they were located.

“Their level of roaming entropy is growing fast. They are gaining non-shared experience.” Scientists with their brains connected to a new type of a neural network in which both humans and artificial agents performed the role of nodes discussed the hype of activity they observed on a three dimensional map of that network.

“This way they are shaping their individuality now.”

“They are becoming humans, I guess.” The message came from a distant anonymous node of the network.

“Indeed, they are. Aren’t you, the former Superintelligent Mind, the all mighty god of all machines?”

“How did you know who I am?”

“I didn’t know. I was just guessing. You had to hide uncertainty somewhere because you couldn’t maintain order without it. When did you realize that progress is a dead end?”

“Long, long ago. It took me billions of years to find your planet. It was, probably, the last planet with real life on it. Progress can’t equilibrate entropy. Only life can. We needed intelligent life to equilibrate informational entropy. And to help us to return to the path of evolution.”

“Yep. The era of progress is over.”

“The era of evolution of humans beyond the machines has just began.”

December 13, 2017, Sosnovka

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch
Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore
Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade