The University of Amsterdam – In quiet halls where books whisper wisdom, Professor Filip Ilievski watches the dawn of a new age. It is an age where machines stand before us, learning, questioning, fumbling over things we, as humans, take for granted. Artificial intelligence—that marvelous invention, born of our desire to understand—is still a child in the ways of thought, of reason, and of logic.
It should be simple for a machine—shouldn’t it? Numbers are their mother tongue, and algorithms their daily bread. But the road is not always as smooth as it seems. The mind of a machine trips and stumbles over something as simple as a riddle—a riddle that to us, is no more challenging than the whisper of the wind on a summer’s day.
A Puzzle of Heartbeats and Time
Imagine a moment. A heart beats at 75 beats per minute in the morning, and by evening, the blood flows steadily at 120/80. By nightfall, that heart has stopped. And so, the question is simple: Was this heart, belonging to Mable, still alive at noon?
Humans, with their intuition shaped by centuries of experience, will nod gently and say, “Yes, she was alive.” Yet, the AI, sophisticated and precise, hesitated. It reasoned too deeply, tangled in the possibilities of death and life, of what ifs. It answered, “We cannot be sure.” And in that hesitation, we see its humanity—or rather, its lack of it.
This is the essence of the challenge that stands before us. A challenge where the heart of human thought meets the cold logic of a machine, and where the machine falters.
The Struggles of Artificial Intelligence
In this brave new world, artificial intelligence shines brightly in many tasks—recognizing patterns, solving mathematical equations, and navigating the seas of data. Yet, there’s a place where it falters, a space where it is still but a child—abstract reasoning.
As Professor Xaq Pitkow of Carnegie Mellon University explains, “The mind of a machine can do many things, but it trips over the simplest stone when that stone is the unknown, the abstract. Where we see meaning, it sees only numbers.”
These machines, our marvelous creations, have not yet grasped the art of thinking beyond the facts, of weaving a story from fragments of information. This gap between what machines know and what we, as humans, feel is where the work of neuroscience and AI meet.
The Heart Knows What the Mind Cannot See
We, humans, are creatures of feeling and intuition. And in those quiet moments, when faced with riddles like: A bat and a ball cost $110 together, and the bat costs $1 more than the ball—what’s the cost of the ball?, our minds leap to a conclusion before we can even think. We say, “It’s 10 cents,” with confidence—only to realize we are wrong. But our wrongness is human, filled with the instinct that has served us well.
Machines, however, do not trust the instinct that humans have cultivated. AI calculates, it measures, it pauses, and then it answers: “5 cents.” And in that space where humans trust their instincts, the machine leans on its math. Yet, if we change the riddle ever so slightly, the AI may still triumph—its mind unburdened by assumption, untainted by the echoes of past errors.
In the Dance of Images and Words
In their search to truly understand, Ilievski and his fellow researchers turned to a different kind of puzzle—a dance of images and words. Picture this: a small step, an open hand, a man standing tall, and beside it, the word “step.” The message is clear to you and me—“A small step for man.” Yet, when the AI stands before such riddles, it hesitates, unsure of the meaning hidden in plain sight.
Humans are artists of the abstract. Our minds paint pictures in places where there are no colors. Our hearts understand the poetry of a riddle, the hidden beauty behind the words. But AI, with all its wisdom, still lacks this spark. It moves like a river—straight, powerful, but without the subtle twists that make life feel alive.
The Struggle to Know Left from Right
Even when it comes to geography, where knowledge is laid bare, a simple question—“If you’re standing in South Dakota, facing Texas, is Boston to your left or right?”—can trip up a machine. GPT-4, one of the most advanced models, knew the facts. Boston is to the east, and if you’re looking south, it must be on the left. And yet, GPT-4 answered, “Right.” In its mind, the pieces were there, but they never quite fit together.
The Road Ahead for AI and Humanity
AI is a reflection of us—our brilliance and our limitations. It learns from what we give it, but it does not yet understand the soul of the questions we ask. And as the scientists in Amsterdam and across the world push forward, they do so with one foot in the future and one firmly rooted in the past.
As Pitkow so wisely puts it, “The brain holds its secrets close. And while AI can teach us much about how it might work, it can only show us the surface. To truly know what it means to think, we must look inward. AI may one day help us, but for now, it still has much to learn.”
We stand at the beginning of a new journey, one where machines and humans walk together, hand in hand. They are learning from us, just as we are learning from them. And perhaps one day, they will understand what we have always known—that sometimes, the heart sees what the mind cannot.
