Is Artificial Intelligence the New ‘Golem’?

When a major main-frame computer was unveiled in the mid-1960s at the Weizmann Institute in Israel, it was introduced by a famous scholar of Kabbalah-Jewish mysticism named Gershom Scholem. Sholem named the large computer, Golem Alef (Golem1). Why did this renowned thinker and writer name the machine Golem?

The root of the word, “golem,” can be traced back to Biblical times. There in the Book of Psalms, it seems to invoke a formless mass out of which something useful might be shaped. It should be remembered that within the Torah Genesis narrative Adam himself was formed from the “adama” — the red clay of the earth.

In rabbinical times, the word “golem” combined the earlier scriptural concepts of a clay-like mass, and that of the primeval man, Adam, to give us the image of a clay robot brought to life through the mystical power of Kabbalistic Hebrew words — often invoking the Creator’s name.

Get The AJT Newsletter by email and never miss our top stories Free Sign Up

It is easy to see how this automaton might become an irresistible figure of folklore: very large and powerful, brought to life by mystically inclined rabbis, often to protect the Jewish people. These humanoids were under the control of their creators, that is, at first. Often, the golems would become monsters, no longer controlled by the sages who brought them to life, and so they would have to be destroyed. Ironically, these protectors of the Jewish people would become a danger to that very community, as well as to other people.

There were many such stories, that is, many versions of the Golem-out-of-control theme. Remember the Wise Men of Chelm and their ridiculous antics? This same Polish town – Chelm — had a golem as the story goes. The most famous (or infamous) golem however, was the Golem of Prague.

According to the folktale near the end of the 16th century, the Jewish community of Prague was protected by the famous Golem of Prague. Out of desperation it was sculpted from the clay of the city river and animated by the great rabbi of Prague at that time — Judah Loew. The story goes that the Golem was housed it the attic of the Alt Neu (Old New) Synagogue, which you can still visit — the synagogue, that is. The creature was brought to life through invoking powerful kabbalistic means to fight the Jew-haters who would incite pogroms against the poor people who lived in the ghetto. Rabbi Judah had to remove the magic spell which vivified the Golem as its behavior became violent and unpredictable, in effect turning the robot “off.” Where then is the body of the clay behemoth? No one knows.

For many, the tale of the Golem of Prague and its cousin golems of other towns are like other cautionary tales from other cultures: “The Sorcerer’s Apprentice” by Goethe, also reimagined by Walt Disney starring Micky Mouse in the movie Fantasia; Mary Shelly’s novel, “Frankenstein,” whose monster was brought unforgettably to the big screen by Boris Karloff.

Whether the underlying power accessed in the story is mystical, magical, or technological, that power can easily go wrong. In the case of the golem narratives, sometimes the problem was that the golem would take its instructions too literally — it had no sense of context. Later, in Yiddish the term, “golem,” would come to mean a big obtuse lummox, someone who could not “read the room,” oblivious to context.

A golem which cannot use common sense in carrying out its orders? This is an example of the alignment problem. How can we be sure that our robots, our artificial intelligence, our computational systems understand exactly what we are instructing, and also just as importantly what we are not intending (so that they are aligned with our interests)? How can we be sure that these optimizing technologies have the outputs that we want? And if they do not, can we correct them in time? Can we turn them off if we need to for safety, as did Rabbi Judah Loew to the Prague Golem?

Think of the foresight of Gershom Scholem 60 years ago, naming the massive computer at the Weizmann Institute in Rehovoth Golem Alef. In his remarks, Dr. Scholem ruminated philosophically about the two golems: the one which was in Prague and the (then) new one in Rehovoth that he was dedicating. With a background in mathematics and logic himself, Scholem understood the double-edged nature of this still new and very powerful technology. Six decades later, we are still wrestling with the promise and the peril of this yet new golem — artificial intelligence.

We need to proceed very carefully, wisely, and with humility, so that the story of this new emerging technology, call it Golem AI, will turn out much better than the Golem of Prague.