Alchemy of Artificial Intelligence and The Proximity to Kings
To my readers it will appear as though I am writing some article on old Greek mythology, but you will soon realize that the world remains the same the more it changes. Recently Ali Rahimi, a researcher in artificial intelligence at Google, compared machine learning with alchemy. Later a few technology journalists, more than ever before, started writing about the relationship between technology and alchemy.
Alchemy is about using the “trial and error” method and coming out with a formula (mostly secret or something that cannot be deconstructed). Similarly, in machine learning a model is designed out of data, this model constantly learns and produces an output but nobody know how decisions are made.
On Alchemy and Humanity
You call it mystical or mythical, alchemy had played a very important role in discovering the relationship between man and universe in the field of science. Alchemy was aimed to transmutation of “base metals” (e.g., lead) into “noble metals” (particularly gold).
How do we know what works best in the interest of humanity?
1️⃣ Science which is a process;
2️⃣ Alchemy which is based on “trial and errors”;
3️⃣ Expert advice that is nothing better than dart-throwing monkeys.
But if monkeys can do a much better job than experts, then it appears the “trial and error” method might work out better than science that is based on “reproducible facts” and not “unproven beliefs”.
Considering all this, it appears that soon we will be living in a society ruled by black boxes where multiple secret algorithms will control our commerce, culture and nation, until we get a master algorithm that will take over the world
Mystery of AI Alchemy
European leaders and kings were always after the secrets of gold making. In those times it was dangerous for alchemists to open up because they were aware that their secret science (which was actually fraud) might put them in some danger.
Manufactures of gold were always worried that once this substance (gold), which is precious, became available abundantly, then they all would go bankrupt.
Matthew Effect and Artificial Intelligence Mongers
‘For to all those who have, more will be given’ The Gospel of St Matthew states (Matthew 25:29)
Huge extents of our data that belong to Google, Facebook and others, is further used to train their new algorithms to make quicker, improved and money making black-boxes. All this success will make these technology giants stronger, while legal or regulatory responses to such impending problems are either absent or insufficient.
With more alchemy power (secret black boxes) in their hands, they will enjoy even greater proximity to government authorities.
The Proximity to Kings
As alchemists enjoyed closed relationships with emperors and kings, now artificial intelligence vendors are helping government in making policies that affect our daily lives.
Emperors, princes, and priests all witnessed the apparent miracle of astounding metallic metamorphosis. Many great alchemists occupied higher positions while their critics remained obscure. In 1404, King Henry IV of England issued an act that declared the multiplying of metal was a crime against the crown.
In the 17th century, there used to be official alchemists to the Emperors and the love-hate relationships between them was interesting. For my interested readers I am taking this selection from a book 🔗The Chemical Choir: A History of Alchemy Reprint Edition by P. G. Maxwell-Stuart
In 1644 and 1646, Christian IV of Denmark minted coins from gold he said he had transmuted from base metals by his Court alchemist, Casper Harback; in 1647, one J.P. Hofman transmuted base metal into gold in the presence of the Holy Roman Emperor, Ferdinand III; and the following year another equally obscure alchemist called Busardier had some powder (presumably the stone, as it turned out), stolen virtually from his deathbed by a friend, recorded only as ‘Richthausen’ who then used it to perform a transmutation of mercury in the presence of Ferdinand III and the Count von Rutz, the Imperial Director of Mines. Every precaution, which arc assured, was taken to prevent fraud, and at the conclusion of the demonstration the Master of the Imperial Mint declared that the gold was the finest he had ever tested. The Emperor was impressed and had a medal struck to commemorate the occasion: Prague, 15 January 1648. Two years later the Emperor himself carried out a successful transmutation with some of Richthausen’s powder, and in 1658 the experiment with mercury was repeated, again successfully, in the presence of the Elector of Mainz. In 1675 Emperor Leopold I witnessed the transmutation of a copper vessel into gold by a monk from Bohemia, VenzeI Seyler — a man, however, who was discovered cheating and sent back to his cloister for a couple of years; and in 1704 the Emperor was treated to further demonstrations, this time by a Neapolitan, Domenico Cactano, who had astonished the Court of Madrid in a similar fashion. But Cactano also turned out to be a fraud, and was hanged in 1709, just like the official alchemist to the Margrave of Bayreuth, who had successfully swindled his master for nine years between 1677 and 1686 before being uncovered at last and ending on the gallows.
Let us assume Artificial Intelligence is Alchemy and Google, Amazon and Facebook are top vendors selling red powder (AI black box) and also note that these technology giants spent nearly $50 million to influence the U.S. government just in 2017 alone. It is no doubt that these companies have become very influential more like governments.
Gold, without control by the state, could result in the bankruptcy of the state, so “multiplying of metals” got banned in 1404 by King Henry IV of the England. Artificial intelligence’s self-replicating, self-adapting and self-competitive features with complex black-box optimization, “trial-and-error” methods, are all good (though not perfect), and better than any human made manual methods. 🔗AI will contribute $15.7 trillion to the global economy by 2030, but the real problem is accountability. However, many mistakes and ethical dilemmas arise, like a wrongly diagnosed patient versus a doctor’s judgement, or self-driving cars’ 🔗who will you kill type scenario.
I am still not sufficiently experienced in machine learning to understand “estimating the expected value of partial perfect information”, (whatever that means) but using MIT’s 🔗Moral Machine ( A platform for gathering a human perspective on moral decisions made by machine intelligence, such as self-driving cars ) I just saved more humans than pets, more fit people than large people, and more males than female.
My Moral Machine result shows I am already biased but should I take it seriously and mend myself by testing my moral decisions again and again for a machine?
No, because every time I am out in the street I see that potential scenarios of moral consequences are infinite and machines can’t decide it for me, it’s not happening for me. Not today, at least.