The quantum here-and-now – POLITICO

by

Quantum computing falls very much on the future side of this newsletter’s editorial ambit, but that future is already here for a small, yet burgeoning, group of companies.

As lawmakers discuss reauthorizing the $1.2 billion National Quantum Initiative Act, companies selling quantum computing products on the commercial market are urging Congress to focus not just on R&D, but on how the tech can be used right now.

According to them, if that doesn’t happen America risks losing the technological edge and private-sector strength that powered its global domination of the smartphone era. In various conversations over the past month business, policy, and research voices in commercial quantum have had a common refrain: When it comes to quantum, the federal government risks slipping behind globally if it prioritizes R&D over current applications.

“We spent $1.3 trillion on an infrastructure package, but none of it looked at how we can use quantum computing to do this in a better manner,” said Allison Schwartz, the global government relations and public affairs lead at quantum computing company D-Wave. (She cited examples like her company’s collaboration with the Japanese government to optimize evacuation routes during natural disasters, or Australia’s plan to use quantum to modernize its transportation network, as possible models.)

Those projects are intriguing, but they invite the cardinal question in quantum development and decision-making: Can quantum computers reliably outdo their classical counterparts that we use today? So far, the answer is no. Just look at the National Institute of Standards and Technology, which recently held a competition to develop (hopefully) “quantum-proof” cryptography standards and ultimately selected four classical algorithms to do the job.

“Classical computing is just unbelievably good. It’s the crown jewel of our civilization,” said Scott Anderson, director of the Quantum Information Center at the University of Texas at Austin and one of the foremost experts on quantum computing, when we spoke earlier this month.

“Even when a general problem like simulating a large quantum system is intractable for a classical computer, the chemists and the physicists [who require massively powerful computing] don’t just give up. They resort to other heuristics and approximation methods that often get them there in practice,” he continued.

Quantum computing is struggling to overcome the problem of “decoherence,” or the inability to keep a “qubit” — essentially a quantum transistor — stable enough to perform calculations, a process that faces extreme scientific and logistical challenges. (For a fairly effective explainer for the layperson, or a journalist, of what this sophisticated physics problem means, read here.) Companies like D-Wave skirt the fact that this hasn’t really been solved yet by taking different approaches to the technology than the scientists trying to crack the decoherence problem at IBM or Google are.

Quantum Computing Inc. (or QCi), for example, uses a technique that accepts the inherent instability of quantum systems, and then uses photons to leverage that instability into something capable of computation, as described in a presentation by the company’s COO and CTO Bill McGann. D-Wave uses a process called quantum annealing that boasts greater stability than traditional, or “gate-based” quantum systems, but for which the applications are limited to quantitative problems in industries like medicine or finance. (They announced in 2021 that they’re now developing gate-based quantum computers as well.)

Who’s buying these systems, and what are they used for? QCi announced last year a partnership with a Dutch bank to use its “entropy” computers to detect fraud. Early D-Wave systems were purchased by Lockheed Martin to solve computational problems, and purchased by Google itself in 2013 to power a collaboration with NASA.

Despite promising recent developments, however, it remains the case that in 2023 there’s no definitive proof of a quantum computer doing something a classical computer cannot do. That means even quantum computers’ current practical applications are orienting more toward pushing the field forward, or familiarizing forward-thinking companies with the technology so they have a leg up over competitors when the quantum future does arrive. (Last week I met with Sergio Gago Huerta, head of quantum at Moody’s and author of the Quantum Pirates newsletter, who described his work largely as educating quants at financial firms on how the technology can fit into their deeply number-crunching-heavy trade.)

And it’s not as if the government is doing nothing when it comes to quantum’s applications: Among the Biden administration’s just-announced slate of regional tech hubs mandated by last year’s CHIPS and Science Act is a project called Elevate Quantum Colorado, which seeks to “increase infrastructure resilience and strengthen the quantum hardware supply chain” by building on the region’s robust research infrastructure.

Still, commercial quantum boosters argue that government-assisted synergy with the private sector is necessary to keep both research and application thriving — and that similar to the case made in a report earlier this month from the Center For Data Innovation, it’s vital for keeping America in the driver’s seat globally.

“The technology is advanced enough. The U.K. is engaging, the EU is engaging, Germany, Australia, Japan, but also China and Russia,” said D-Wave’s Schwartz. “We need to work together with our allies, so the U.S. doesn’t fall behind.”

An architect of this year’s red-teaming challenge at the DEF CON hacker conference defended it from recent criticism that the AI troubleshooting practice “faces substantial limits in mitigating real-world harms.”

Writing to DFD in response to a report from the nonprofit Data & Society that criticized the DEF CON challenge, Austin Carson argued that he and his fellow organizers understand full well that red-teaming is just one tool in a broader diagnostic toolbox. Carson is founder of the nonprofit SeedAI and an organizer of the red-teaming challenge meant to test the security and safety boundaries of popular AI systems.

“Stating we believed that the red-teaming process would solve all of the concerns around AI certainly misunderstood or misrepresented our effort. If I’m being honest, anyone pushing red teaming as a fix-all is either being disingenuous or negligent. I’d like to believe the authors of the report know we’re neither,” he wrote.

“Our event at DEF CON wasn’t solely about red-teaming and finding vulnerability disclosures. It was a first step in opening the floodgates to generate awareness of AI and expose more people to it through the lens of AI security,” he said. He also writes that he doesn’t necessarily disagree with Data & Society’s premises — saying, in keeping with their concerns, that his team is planning to take the findings from the DEF CON exercise across the country to hear from “a broader swath of community members” on what they want from AI systems.

“We agree with the report authors’ point about the importance of constructing red-teaming exercises in a way that results in actionable insight and includes a transparent disclosure and accountability process… Moving forward, we are working to define ‘what is red-teaming’ and how we differentiate between things that reflect individual or societal values with safety issues,” he wrote.

For international college admissions, ChatGPT is increasingly “the elephant in the room” as one admissions counselor put it.

That’s according to this morning’s POLITICO China Watcher, which reported on the first full admissions cycle after China’s zero-COVID-19 policy expired this year. The education community finds itself split on whether Chinese students who use ChatGPT to write English-language essays are overcoming a language barrier, or simply cheating like their English-speaking counterparts.

“ChatGPT’s cheerleaders hail it as a democratizing agent… But application fraud was a major issue among Chinese applicants even before ChatGPT,” the authors write, noting that a robust cheating infrastructure in the country might be supercharged by the software.

But possibly more significant than the risk of cheating, they note, is the growing technological gap between American and Chinese students. The latter can only access ChatGPT through a VPN, meaning it’s barred to them in official university settings where they might otherwise use it for valid educational purposes.