Ask any programmer about the future of computation, and chances are they’ll at least mention the intriguing possibility of quantum computing: a field that harnesses the strange behavior of subatomic particles to perform multiple calculations at once.
The quantum computers of tomorrow, researchers say, will run at a pace that outstrips what’s currently possible with traditional computers, bolstering humankind’s ability to achieve artificial intelligence, synthesize new materials, shore up its encryption capabilities, and more.
None of that has happened yet. But if the findings of a study published today in Nature are confirmed, Google may have just brought us one step closer to that tantalizing reality.
In the paper, the tech giant’s AI Quantum Team presents evidence that it’s built a quantum computer that needs only 200 seconds to solve a problem that would have taken IBM’s Summit, the world’s most powerful supercomputer, 10,000 years to crack. Google calls their computational feat, in which a quantum computer accomplished something a traditional computer could not, “quantum supremacy”—a bold term meant to signal the beginning of an age in which traditional computers are rendered obsolete.
Already, the claim has been disputed by IBM, which designed Summit for the Department of Energy. With just a tweak or two, IBM scientists wrote in a blog post, Summit could accomplish Google’s task in just 2.5 days.
But others in the computing community are tentatively optimistic about the breakthrough’s promise—one that’s “a big step toward kicking away any plausible argument that making a quantum computer is impossible,” Greg Kuperberg, a mathematician at the University of California, Davis who was not involved in Google’s efforts, told Adrian Cho at Science last month (when the paper was accidentally leaked on a NASA server).
To be clear, the quantum computer in question—a 53-qubit processor called Sycamore—didn’t solve a puzzle with a direct, practical application: Though the computation itself was complex, its output was just strings of random numbers. But it’s proof, Google researchers argue, that quantum mechanics can be put to use.
Traditional computers perform calculations by processing “bits” of information that each encode either a 0 or a 1, but never both. Quantum mechanics subverts this binary hurdle. Like subatomic particles, which can exist in two states at once, so-called quantum bits (or qubits) can store two numeric values at once. This property of double-occupancy can be a huge boon for efficiency. Rather than analyzing things one at a time, a quantum computer can multitask—and every additional qubit increases its processing power exponentially.
On a larger, more zoomed-out scale, quantum computers could also interpret problems differently than traditional computers do, thanks to the ways their qubits interact, and may have better ways to identify correct solutions over incorrect ones. All this goes to say, the potential of quantum computing is immense.
That’s great in theory, but actually manufacturing a qubit-based computer isn’t easy. Quantum states have to be induced, and remain extremely fragile to even the slightest perturbations.
It’s something researchers—including those at Google and IBM—have struggled with for years.
That may be why Google’s paper has provoked skepticism since it was first released.
In their blog post, the team at IBM stands by Summit’s computational prowess. They also point out a big drawback of Google’s program: It can’t correct its own errors. Summit, they say, would have accomplished Sycamore’s task with more accuracy.
Google has yet to apply their quantum computer to something “practical,” Director of IBM Research Dario Gil pointed out to Cho, a limitation he says might diminish the oomph behind their assertions. “This is not about final and absolute dominance over classical computers,” Gil told Cade Metz at The New York Times.
Google, however, stands by Sycamore’s achievement. Even incremental steps get us closer to a quantum future and everything it has to offer, they explain in a blog post. “It’s these milestones that drive progress in the field,” Google CEO Sundar Pichai told MIT Technology Review’s Gideon Lichfield. “That’s how humanity makes progress.”