In 1953 the philosopher Isaiah Berlin wrote an influential essay called “The Hedgehog and the Fox,” which drew upon an ancient Greek proverb that “a fox knows many things, but a hedgehog knows one big thing.” Berlin used the image to divide thinkers and creators into two camps: hedgehogs who see the world through a single defining idea or vision; and foxes, who can see the world from multiple perspectives at once.
In the end, neither beast is superior or better to the other. What the hedgehog gains in clarity and precision, Berlin argued, the fox gains in agility and versatility. Bring them together, however, and the result is an explosion of creativity and power.At first glance today‘s AI would seem the proverbial fox of high-tech, thanks to ChatGPT and generative AI apps’ ability to move and make decisions faster than human judgment, even faster than the human eye.
Meanwhile, today‘s quantum computers seem the plodding, slow-moving hedgehog, as scientists and engineers at Google, Intel, Microsoft, IonQ, and other companies work laboriously to increase the number of entangled qubits to increase their computer’s power.
So while AI spreads like wildfire across the global economy and changing lives and experiences, the timeline for large-scale quantum computers lags far behind, as the diagram above shows. Some skeptics even wonder if we‘ll ever get to the point where quantum computers can make s significant difference in the quality of human life, or in our economy.
But perhaps it‘s time to look at the problem from a different angle, i.e. from the underlying physics of both technologies. Seen in that light, AI actually turns out to be the slow moving hedgehog; and quantum the quick darting fox.
Why? Because AI ultimately relies on the same predictable pattern of digital computations, no matter how fast and spontaneous its insights and conclusions may seem. At the same time, machine learning, the true workhorse of AI, remains a mechanical mock-up of the intricacies of the human brain, not an exact replica―even in its deep neural networks state.
Quantum science, on the other hand, is built on the unpredictable bountiful randomness of nature itself as embodied in the quanta, the most basic unit of energy and reality. The quanta move at the swift speed of light itself―in fact they define the speed of light. Unpredictable, lightning fast, dwelling in a realm beyond human certainty or control: the quanta map out a future that is unseen but contains all the limitless possibilities of nature, and of humanity‘s place in that nature.
Let me explain.
All machine learning, the foundation of AI, involves a computer‘s ability to recognize patterns in sets of data―whether those data are sounds, images, electrical pulses, or financial transactions. The mathematical representation of these data is called a tensor. As long as data can be converted into a tensor, it’s ready for ML and its more complex offspring, Deep Learning. Deep Learning builds algorithms copied from the structure and functions of the brain‘s neural network, in order to build a predictive model. As the diagram shows, it “learns” by using other testing datasets that correct and validate its initial model.
What ML operators develop is a prediction curve based on the recognition of past patterns. The more data, the better the model. Patterns that didn‘t appear in tens of thousands of examples can suddenly become glaringly obvious in the millionth or ten millionth example.
There‘s nothing magical about AI/ML. All it does is create high-capacity statistical models that can find statistical regularities in larger and larger datasets. Its deep neural networks may be able to seize on those regularities quicker than meets the human eye; but the procedure is no different than any other digital technology.
Here I have a personal confession. I find the physics (as opposed to the science) of AI/ML boring. By contrast, the physics of quantum hums and dances in fascinating and unpredictable ways.
As we pointed out in our last column, the power of a quantum computer rests on the physics of quanta of energy, the most basic building block of matter. Even when harnessed by quantum scientists to do instantaneous calculations in multiple states, the quantum-based bits or qubits remain defiant of human guidance or predictability. Their inherent tendency toward randomness―which makes a random number generator for cryptography using quantum truly and invincibly random unlike its digital emulators―during quantum computations is called “noise.” Today‘s quantum computers are very “noisy,” in that keeping their qubits in an orderly patterns of entanglement is a frustrating challenge, which means the machine will generate a wrong answer to your question almost as often as the right one.
Computer scientists see this “noise” as a drawback, as well as they should. But it‘s not the only place where we encounter “noise” generated by quanta. It also happens in the human eye, where the millions of quanta pouring out from a light source send a confusing array of optical signals which the rods and cones of the physical eye must sort out, in order to recognize patterns and colors at all. In short, quantum noise is part of nature and life. Now it’s part and parcel of the computers of the future.
But here‘s where AI/ML can help. AI can help to cut through the “noise” and sharpen our understanding of the signals generated by the agile quantum computing process. At the same time quantum can deepen and make more precise the calculations run by AI/ML.
For example, quantum computing offers powerful advances in combinatorial optimizing, i.e. finding the most efficient of resources and supply chain allocation. Quantum computing is already showing its skill in modeling the behavior of complex systems like molecular simulations. Its ability to enhance ML tasks including matrix diagonalization and pattern matching with even sharper precision, will all help with sifting out the best and quickest solution from a myriad of data and possibilities.
Then when we bring the fox and hedgehog together in hybrid systems, i.e. ones that integrate quantum‘s rapid capabilities with AI/ML’s solid grounding, the results can be spectacular.
Some will be concerning, as when hybrid quantum/AI platforms are able to factorize the large prime numbers that underlie public encryption systems, from RSA to AES Then the “cryptographically relevant quantum computers” governments and industries all worry about, become reality far sooner than faster than the experts expected.
Other results, however, will be life-saving and life-enhancing, with the rapid development of new drugs and biotech research methodologies, including advances in brain-computer interface (BCI). Together AI and quantum will be able to empower a next―generation of industrial robots that can handle toxic or hazardous tasks and materials without missing a beat or spilling a drop.
In the realm of space, quantum and AI together will be able to seamlessly direct satellite traffic in an increasingly crowded Low Earth Orbit (LEO), and speed the development of new power sources like Helium 3, which opens up the possibility of space travel deep into the solar system. The advent of nuclear-powered propulsion in space, will seem an everyday achievement.
The quantum/AI combination will also transform the power grid, making it more productive but also more secure, while advances in areas like nuclear fusion will become commonplace.
In sum, the coming Quantum/AI convergence will mean an even greater, more productive future for all of us―once we understand, that is, which is the real hedgehog, and which the fox.