Judaism & AI: The Dybbuk in the Machine

Michael M. Rosen’s Like Silicon from Clay: What Ancient Jewish Wisdom Can Teach Us About AI suggests that traditional teachings about the golem, dybbuk, and maggid can inform how we survive and thrive as companies like OpenAI churn out further iterations of the instantly-ubiquitous ChatGPT and whatever computer-creatures come next.

Rosen, a nonresident senior fellow at the American Enterprise Institute where he specializes in intellectual property issues in the field of technology, is deeply knowledgeable about the state of the current ethical dilemmas regarding AI. He is also a committed Jew with command of canonical sources.

As Rosen explains, there have been four basic reactions to the late-2022 launch of ChatGPT, the “large language model (LLM) that has revolutionized many aspects of everyday life… helping write wedding toasts, obituaries, news summaries, and even term papers.”

Positive Autonomists, like OpenAI head Sam Altman, “regard AI’s recent advances as truly revolutionary—a difference in kind, not just degree, from previous computing technology.” They also “believe that machines have already achieved—or will soon achieve—a measure of autonomy… sentience, awareness, or consciousness.” Most importantly, “they wholeheartedly applaud these breakthroughs.”

Negative Autonomists, on the other hand, including Elon Musk and Eliezer Yudkowsky, agree that AI will deeply alter human existence and could potentially become autonomous, but interpret this possibility “overwhelmingly as dangerous, harmful, and even existentially risky” with some like Yudkowsky urging AI’s immediate and permanent deactivation.

Others disagree with this dichotomy. Positive Automatoners deem AI “no more than an extension of human capabilities—a force multiplier that reflects and implements its programmers’ own abilities, assumptions, and biases.” Constructs like ChatGPT, in this view, lack the capabilities of human creativity and independent thinking.

Negative Automatoners agree that the technology is a “mere mechanical prosthetic” but nonetheless worry that machines will harm humanity - not, like the Negative Autonomists, because of AI’s powerful potential, but rather from its ample shortcomings. Forget lacking a human soul - it often can’t even provide accurate information!

As policy-makers, engineers and CEOs navigate these diverse perspectives, Rosen offers three rabbinically-gleaned guideposts.

1. The Golem

The golem, a human-like creature mentioned in the Talmud and in legends surrounding the 17th century sage known as the Maharal of Prague, are understood to protect their creator and the Jewish community. But they could also act in an overly destructive manner. The stories about them, therefore, remind us that creatures created by humans but with the ability to operate on their own should be closely monitored.

As Rosen puts it, “the golem is the Positive Autonomist perspective personified, or at least as personified as a humanoid can be—the product of human hands, but independent, benevolent, and transformative.” But, he reminds his readers, the golem’s “operation remained dependent on its human creator, who was able and willing to terminate its existence when absolutely necessary, such as when the initially good golem turned bad.”

Thus the 20th century Israeli scholar Gershom Scholem, in a 1965 speech celebrating The Hebrew University of Jerusalem’s first computer, which he referred to as a “golem,” remarked to it, and the audience, “Develop peacefully and don’t destroy the world. Shalom.”

2. The Dybbuk

A supernatural spirit thought to possess a human form, the dybbuk expresses mischievous and often downright evil inclinations. In tales of it possessing individuals, communal sages are compelled to exorcize and banish it. Such a demonic entity, Rosen proposes, “should be especially attractive to Negative Automatoners, whose default view of robots emphasizes their dark and stultifying aspects.”

The cautionary tales of dybbuks parallel the feelings of those who consider AI a parallel phenomenon, “a dark, dangerous, and heavy reflection of humanity; a malign force that humanity must exorcise; and a burden to be cast off” if we are to remain psychologically healthy.

3. The Maggid

Much less well-known, but perhaps more valuable to the current moment, is the maggid. The seminal 16th century Jewish scholar Rabbi Joseph Karo, the author of the classic Jewish work the Shulchan Arukh, The Code of Jewish Law, believed that a subconscious voice - a positive mirror of the dybbuk - visited him and inspired his scholarship and spiritual growth. He called it a maggid, a storyteller, and similar to Sam Altman’s view of AI, it “inspired him and brought out his best.”

Our inner dybbuks—our darkest instincts—will dominate our thinking about technology and the solutions we devise if we don’t empower our inner maggids—our better angels—to overpower them.

Rabbi Karo, seemingly the first to mention such a spiritual entity, describes it in a way, that, as Rosen puts it, “resembles the dybbuk in that, like its darker cousin, it represents human urges and creativity without necessarily tran­scending them. Unlike the dybbuk, however, the maggid reflects our inner angels, positive urges, and desire to apply the best versions of ourselves to bettering the world.”

Ultimately, Rosen concludes, “our inner dybbuks—our darkest instincts—will dominate our thinking about technology and the solutions we devise if we don’t empower our inner maggids—our better angels—to overpower them.”

He proposes “warmly embracing the best that our contemporary golems have to offer by encouraging the continued development of technology that will extend and enhance human existence.” At the same time, he argues for a strategy of rigorous voluntary industry standards, informed by corporate, government and public ethical voices, “that will ensure AI development is carried out responsibly and for proper purposes.”

Part of this regulatory process should be identifying and rooting out our worst psychological biases and impulses—the dybbuks—from our machines “and imbuing them with our better angels—our maggids.” And, of course, like the creators of golems of old, having safeguards in place should what we forged threatened to escape human control.

Rosen’s final prayer, echoing Scholem, is that whatever AI develops next comes in shalom, ushering in a more peaceful existence for us all.