How to enable quantum computing innovation through access

Two recent breakthroughs in quantum computing have generated significant excitement in the field. By using quantum computers to solve problems that classical computers could not, researchers in the United States and China have separately ushered in the era of “quantum advantage.” Yet as momentous as the demonstration of quantum advantage may be, it is the availability of more capable quantum machines that will ultimately have greater impact. Access to these machines will foster a cohort of “quantum natives” capable of solving real-world problems with quantum computers.

Both recent breakthroughs—random circuit sampling by Google in 2019 and boson sampling by the University of Science and Technology of China in 2020—are problems useful for demonstrating quantum advantage. But they do not have real-world utility and are akin to esoteric Plinko games. Neither demonstration brings us closer to identifying any near-term application for quantum computers that will drive technology development and demonstrate impact.

Although quantum computing is in its infancy, the field is already seeing significant commercial investment. The history of classical computing suggests that if this commercial activity is to continue, it is absolutely vital to identify real-world applications for near-term quantum machines, applications with real advantage over classical approaches. Doing so requires us to make quantum computing available much more widely. Fortunately, what we are also witnessing is the emergence of quantum machines sufficiently capable of engaging a broader cohort of the public—and it is this public availability that will maximize our ability to identify truly useful applications.

State of quantum computing

Decades of relentless progress in classical computing have trained technologists to expect ever-increasing compute power. Moore’s Law holds that compute power will roughly double every two years, but because of recent plateauing of pure Moore’s Law scaling, many technologists are now focused on what kind of computing comes next. Will it be application-specific computing? Hybrid approaches to computing technology? Or even entirely new computing paradigms?

Many have placed bets on quantum computing as an alternative paradigm with exponential performance gains over classical computing in solving certain types of problems. The long-awaited demonstration of quantum advantage appears to validate that belief.

The idea of creating a quantum computer, operating according to the laws of quantum mechanics, was suggested by Richard Feynman and others in the early 1980s. At the time, scientists were considering how best to simulate the chemical interactions ubiquitous in natural systems. What better way to simulate and understand a system following the laws of quantum mechanics than to use a computational system governed by those same quantum mechanical laws? Interest in quantum computing exploded in the mid-1990s when Peter Shor published an algorithm using a quantum computer to factor very large numbers exponentially faster than the best-known classical algorithms. The difficulty of factoring very large numbers is the basis for many modern encryption systems, and Shor’s algorithm solidified the idea that a quantum computer would have applications beyond quantum mechanical simulations.

The quantum mechanical properties of superposition and entanglement together create a computing system of unprecedented complexity and power, capable of completing certain calculations exponentially faster than classical computers. But with great power comes great vulnerability: Superposition and entanglement make quantum computers exquisitely sensitive to noise and difficult to control. Unlike classical computing, where small shifts are easily rounded back to true bit values, even small errors in a qubit value will be difficult to correct. Quantum information scientists believe that algorithms known as quantum error correction (QEC) will allow us to control errors and use quantum computers for real, useful calculations.

In the quarter century since Peter Shor’s factoring algorithm was published, a handful of other quantum algorithms have emerged. These include optimization, solving linear systems of equations, and approximation methods for chemical simulation. During the same period, significant hardware research has brought us to the so-called NISQ-era, that of noisy, intermediate-scale quantum machines. These small, imperfect quantum computation devices are our first opportunity to use quantum machines and test the building blocks of quantum computation. Although interesting, NISQ machines are not the large, error-corrected machines that can execute useful optimization, factoring, or simulation. We have not yet identified applications of value for near-term NISQ machines, and many scientists argue the real purpose of NISQ computers is to learn how to build better quantum computers.

Using the quantum computers of today to learn how to build the quantum computers of tomorrow may be a satisfactory scientific answer, but it is not clear that that answer is sufficient to sustain necessary interest and investment in quantum computing to continue its advancement.

Classical computing’s virtuous cycle

While the exponential growth of classical computing power is well known, perhaps less known is that the cost of producing increasingly powerful computer chips also grows exponentially. But that growing cost is underwritten by exponentially increasing revenue generated by the computing industry: Revenue generated by high-impact applications is then available to be reinvested in crucial research and development activities.

As described in the National Academies’ 2019 report Quantum Computing: Progress and Prospects, the cycle of increased revenue to increased manufacturing cost to increased performance served the classical computing industry, and our society, well for more than 50 years. Within this “virtuous cycle,” we see a second cycle: New computing applications drew more talent into the computing ecosystem, and these talented people created the next set of industry-driving applications.

The need for applications, the need for availability

In 2009, a group of quantum information scientists published a new quantum algorithm to solve linear systems of equations—think back to high school algebra and solving “N equations for N unknowns.” The ultimate impact of this algorithm was unclear, however, because the quantum efficiency would be blunted by the need to run the algorithm many times to glean all N unknowns. It took a second group of scientists, ones intimately familiar with a particular application of calculating radar cross sections, to develop a version of the algorithm in 2013 to solve for a single, concrete value—this single value being a function of all N unknowns. By needing to calculate only one value, the quantum efficiency gains are preserved.

This is what happens when we bring more and diverse voices to both quantum information and classical computational challenges. Compared to classical computing, relatively few people are thinking about quantum computing applications, which is a natural state of affairs given that quantum computers of limited capability have only recently become available.

Quantum computers are not supercharged versions of classical computers. Rather, they manipulate information in very different ways. Exploiting the power of quantum complexity for calculations of value requires more scientists familiar with quantum information processing—people who, perhaps, also possess intimate knowledge of existing computational challenges or who have innovative insights into novel capabilities not provided by classical computers.

Identifying a full complement of real-world applications for near-term quantum computers will come only with widespread access to quantum computers. Access is necessary in order to create a cohort of practitioners with the right interests and sufficient expertise to connect quantum capabilities to relevant computational problems. Broader access also creates a larger pool of innovators thinking about quantum information. Only by creating a generation of quantum hackers and tinkerers can we reach serendipitous discoveries. We may even see practical breakthroughs well before abstract understanding, as has occurred in the field of deep learning, where the ability to exploit technology has far outpaced theoretical understanding of it. 

In the summer of 2019, spurred by cloud-based public access to a subset of IBM’s Quantum Experience, the Johns Hopkins Applied Physics Laboratory formed an intern cohort to join the lab’s quantum information team. We playfully referred to the students as our “How to Train Your Dragon” cohort—just as the younger generations of Vikings in that film harnessed the power of dragons for good, we hoped that our students would do the same for quantum computing. Composed of undergraduates with no quantum background, the cohort was exposed to a range of mathematics and physics concepts that enabled them to perform cutting-edge research within months. Their testing and evaluation of noise mitigation protocols on real quantum processors will be featured in a forthcoming technical journal publication, constituting a novel contribution to the field of quantum information. My colleagues extended this internship program the next summer to include talented high school students and saw a similar, rapid learning process, creating a new cohort of “Quantum Natives.” With access to cloud computing resources, these young scientists were able to do meaningful, breakthrough work. Imagine the breakthroughs possible if these resources were more broadly available.

Discussions of quantum computing often focus on the exotic, nonintuitive aspects of the paradigm, but our experience introducing students to quantum machines indicates that creating the quantum generation may not be so different from creating the digital generation. Early access to computers shaped the first generation of classical computing entrepreneurs. Bill Gates spent thousands of hours programming as a teenager, developing an expertise from which sprang Microsoft. Stories of serendipitous access to novel technology are all but ubiquitous among scientists, albeit with results less dramatic than Microsoft. My own earliest exposure to a computer was in the home of a friend. His father had a TRS-80 hooked up to the television with a cassette deck as the memory. We spent hours carefully programming the machine to play blackjack—our wonder tempered only slightly by unfamiliarity with the card game itself. Others have more intense reports of spending long afternoons programming university mainframes; in the 1970s, apparently, no one locked laboratory doors to tweens.

Access to the quantum cloud

Quantum computers are highly specialized machines currently not suited for life outside research laboratories, so public access to these machines currently occurs via the cloud—online access over existing internet infrastructure. The availability of early quantum computing resources will not need to be mediated by purchase of specialized equipment, creating a real opportunity for broad-based access to rather exotic technology.

Clearly, access to cloud-based quantum computing will require digital access considered ubiquitous in modern society. But virtual schooling, driven by the COVID-19 pandemic, has laid bare socioeconomic divides in technology and internet resources—resources not nearly as universal as often imagined. Quantum computing serves as one more reminder that universal digital infrastructure is absolutely vital for the education of today, the workforce of tomorrow, and participation in the technology development of the future.

Cloud-based access to quantum computing makes it easier to bring diverse voices to quantum information development and encourages cross-pollination of scientific domains, as seen in our earlier discussion of the synergy between the quantum linear systems algorithm and challenges in calculating radar cross sections. The contrast in access to quantum computers, compared to classical ones, may even sidestep phenomena partially responsible for the gender gap in computer science.

Early computer programming was frequently dominated by women. Before the 1980s, the number of women studying computer science at universities was growing faster than the number of men. This trend reversed itself in the mid-1980s, with women’s participation falling precipitously. Scholarship has traced this inflection, counterintuitively, to the emergence of personal computers. As computer hardware became more affordable and available, it was marketed very specifically to boys, creating a large computer literacy gap between men and women when they arrived at colleges, establishing a heightened expectation for existing computer skills in even introductory computer science courses. Widespread cloud access to quantum may allow us to avoid creating a similar gendered gap in quantum computer literacy.

Achieving real-world impact

While current quantum computing machines give us the opportunity to directly explore quantum algorithms and applications, these machines have not demonstrated quantum advantage with real-world impact, and we are not confident that we have identified an application that will show true advantage in the short term. The need for continued interest and investment in quantum technology demands that we identify such applications. Broader access to quantum computing resources and, with that access, broader participation will be key to formulating quantum applications.

Current public access to certain quantum computing resources puts us in a strong initial position. The classical computing industry has long maintained a beneficent relationship with higher education, contributing resources that range from student fellowships to in-kind hardware grants to free software licenses. Commercial players in quantum computing must ensure similar avenues for access for both educational and research purposes. The National Q-12 Education Partnership, a public-private effort headed by the National Science Foundation and the White House Office for Science and Technology Policy, commits to bringing quantum education to precollege students. These sorts of initiatives will bring together industry, academia, and government to expand the quantum ecosystem, within which vibrant cycles of research and development will drive progress in quantum computing.

Joan A. Hoffmann is a principal physicist within the Research and Exploratory Development Department at the Johns Hopkins University Applied Physics Laboratory.

IBM and Microsoft provide financial support to the Brookings Institution, a nonprofit organization devoted to rigorous, independent, in-depth public policy research.