Montessori Meets AI

(Photo courtesy of Jeroen van der Most)



My mother talked a lot about Maria Montessori. She could not afford to send me to a Montessori school, so she read all about Dr. Montessori’s work and figured out ways to apply her method at home. Looking back, I suspect that even the way she put milk into little pitcher small enough for me to pour, instead of dumping it right on my cereal, was a Montessori scheme. (It just seemed fancy to me, as fun as playing tea party. Self-reliance and control over one’s environment are learned—or unlearned—in ways invisible to us.)

Montessori’s method included sizing stuff down and give kids ways to perform everyday tasks themselves, on their own initiative. She emphasized creative freedom but within an orderly framework. Kids received the building blocks, the raw materials, and made their own project, building their own curiosity and concentration. Adults stood by, a reliable, kind, and patient presence, but bit their tongue before criticizing or interfering. The child learned to persevere.

The Spirit Machine is a new art installation, commissioned in Montessori’s honor and sited at Radboud University in The Netherlands, inside a new building named for her. It is based, not in a child’s creative intelligence, but in AI.

This intrigues me. How does the admittedly impressive language generator GPT-3 intend to rival a bright five-year-old or recreate her teacher? The concept, it turns out, is clever: All of Montessori’s work was fed into the computer, which now reacts to the university’s research news with a continuous stream of quotes “in her voice,” flowing from her ideas yet intersecting with the modern world.

Artist Jeroen van der Most and his co-creator, Gabriëlle Ras, wanted to bring a historic figure’s ideas back to life, connecting them to current breakthroughs. Were it not for the cool geometric graphics and machine aesthetic, the result would be as creepy as a séance. Semantically and philosophically, the texts do seem exactly the sort of inspiration Montessori would have offered:

“The great secret of education is to direct a child’s capacity for wonder.”

“The real creative act is not the one that produces something new but the one that alters something in our perception of reality.”

“The child must have the joy of discovery, and they must know that the more they discover, the greater their capacity for discovering.”

And surely this next one was snuck in, planted to allay criticism? “The computer is not a machine to think for us. It is a tool that helps us to think.”

My favorite is more subtle: “It is a great mistake to believe that the world has been divided into two parts, one inhabited by the wise and learned and the other by fools. The division is not so clear as this; it exists rather in every human being.” The words sum up my ambivalence toward this project. The chunk of my mind I leave unlocked wants to celebrate the fusion of past and present and future. The Luddite that is my brain’s landlord wants to whack the screens with a mallet until they shatter.

Maria Tecla Artemisia Montessori fought hard to attend medical school. Once there, she was made to dissect cadavers alone, because it was deemed improper for her to be with men in the presence of a naked body; she chain-smoked to escape the stench. Like her tiny future students, she persevered and won top honors. Specializing in pediatrics and psychiatry, she first she focused on children with mental disabilities, then adapted her methods for all children.

Montessori observed carefully, noting the ability of even very young children to fall into rapt concentration when they could pursue their own interests. She saw their need for orderly surroundings, their interest in practical activity, their tendency toward repetitive behavior. All this, she incorporated into her method, encouraging independence, thoughtfulness, and a free and lively curiosity.

Because of the concrete, hands-on way that young children learn, Montessori placed huge emphasis on the senses. She introduced kids to the slow, organic world of plants and animals and let their imagination unfold there. Would she have wanted her students to spend time watching bright letters appear on a screen? Doubtful. But the implied criticism is unfair, because the AI installation was designed to educate grown-ups, not kids.

Montessori did believe that science should transform education. She might have found this project both scientifically groundbreaking and delightfully playful. She defined Spirit as the inspiration to do research, and The Spirit Machine reacts to research in a way meant to inspire us further. Indeed, its texts have the punch of aphorism, the form W.H. Auden described as “aristocratic.”

Sure of itself, an aphorism tosses its wisdom at you in a polished little package and feels no need to blab on. I have traveled through life’s jungle swinging from one of these adages to the next, always finding (or telling myself I have found) the exact piece of wisdom I need for the challenge of the moment.

But aphorisms are not universally smart; they only sound that way because they are short. “The most attractive sentences are not perhaps the wisest, but the surest and soundest,” Henry David Thoreau told his journal. We read an aphorism and nod sagely, because that is all we can do. “The polished boundaries of aphorism signal that dialogue is unwelcome; no reply is necessary,” notes Noreen Masud on

Maybe, but I expect any self-respecting piece of AI could muster a little dialogue if somebody tried to talk back to its aphorism. The ubiquitous Alexa can parry declarations of love with more tact than most humans; surely a more sophisticated language generator ought to be able to answer questions about its texts. And I have a question, a big one: Is this just a game to break down our resistance to AI, or is it a valid process that can be used to bring the past’s distant wisdom into our time? To come close to the consistency of these results, a human being would have to inhabit a historical figure’s entire life, learn their philosophy, master their way of expressing themselves. AI can do all that in a flash. If I stifle my tendency to dismiss this as superficial, mechanical trickery, I have to admit that I would sacrifice—not my right arm, but definitely a few toes—to hear what Sojourner Truth or Lao-Tse would say about, well, AI, or what scathing witticism Oscar Wilde might level at the emoji.

Is it old-fashioned to think that you have to experience something in its own life world before you can offer relevant comment?

Aphorisms are hinged to experience, because that is how we think of wisdom. A statement might sound clever, but knowing who said it and what they endured to reach that insight will inevitably deepen our response. Someone has lived into these words, climbed and swum and nearly drowned to reach them. They are more than a clever pairing of noun and verb. They can be trusted.

Can something be profound if a human being did not come up with it? Of course, I tell myself impatiently. Sunsets inspire us. So do random coincidences, mathematical equations, the fidelity of swans, the antics of a baby koala bear. We grab wisdom wherever we can find it.

On the other hand, when I read in the artist’s manifesto that “The AI reflects like a philosopher,” I snort. No, the AI does not. Even the most bloodless philosopher has suffered; there is depth and texture to their insight. It is not just a randomized game of vocabulary and syntax and keywords and context. The artist’s explanation for the AI acting “like a true philosopher” is the system’s ability to recognize overarching themes and come up with related inspiring subjects. It proceeds by association, rather than sticking to the keywords’ literal meaning and producing a flat statement that reshuffles a few ideas. Cognitively, it is more sophisticated than many of us. “Bigger” themes that touch on human emotions or relationships receive priority. The texts are then googled to make sure the AI has not plagiarized anyone, and if they pass that test, they are put through a custom-made blacklist filter to weed out sensitive content.

How I would love to hear Maria Montessori’s reaction to this filtered sensitivity, this sleek, smart, chilly, high-concept work of art. I suppose we could ask GPT-3?


Read more by Jeannette Cooperman here.