AI Meets Quantum in the Real World— with SandboxAQ | Ep. 75

T
The Post-Quantum World • Apr 03, 2024

Audio Brief

Show transcript
This episode explores the intersection of artificial intelligence and quantum technology, focusing on how AI is enabling near-term quantum applications well before full-scale quantum computers arrive. There are three key takeaways from the conversation with Paul Kassebaum of SandboxAQ. First, AI is currently acting as the critical enabler for quantum sensing technologies. Second, a hybrid approach called the sandwich strategy is accelerating complex simulations in material science. Third, organizations must begin automated inventorying of cryptographic assets immediately to prepare for future security threats. Regarding the first takeaway, the immediate value of AI lies in solving the noise problem inherent in quantum sensing. While quantum sensors are incredibly sensitive, they are easily disrupted by environmental interference. AI algorithms are now being used to filter out this magnetic noise, isolating weak signals effectively. This development allows technologies like magnetic navigation to function as a GPS backup, or magnetic cardiography to operate in busy emergency rooms without the need for magnetically shielded labs. On the second point, SandboxAQ utilizes a method known as the sandwich strategy to solve problems in drug discovery and material science without relying on powerful quantum hardware. This approach uses classical physics simulations as a bottom layer to generate high-quality data. That data then trains AI models, the top layer, to extrapolate solutions much faster than traditional methods. By incorporating active learning, the AI identifies exactly which missing data points offer the highest learning gain, directing the simulation to compute only what is strictly necessary, thereby saving massive computational resources. Finally, regarding security, the transition to post-quantum cryptography requires immediate situational awareness. Most organizations lack visibility into where their encryption lives or how it is implemented across their networks. The conversation highlights that cryptographic vulnerabilities are rarely isolated files but rather complex entanglements within code. Therefore, companies should not wait for final regulatory standards to begin using automated tools to create dynamic inventories of their keys and algorithms. As these technologies mature, leaders should anticipate that regulatory approvals from bodies like the FDA or FAA will likely be the primary bottleneck for deployment, rather than engineering hurdles.

Episode Overview

  • This episode explores the intersection of artificial intelligence (AI) and quantum technology, specifically focusing on how AI is currently enabling near-term quantum applications before full-scale quantum computers arrive.
  • Paul Kassebaum from SandboxAQ explains how his company is leveraging AI to enhance quantum sensing, quantum simulation, and post-quantum cryptography (PQC) for immediate commercial and public sector use.
  • The conversation covers specific use cases such as magnetic navigation as a GPS backup, magnetic cardiography for heart monitoring, and the urgent need for organizations to begin inventorying their cryptographic assets ahead of future quantum threats.

Key Concepts

  • AI as a Quantum Enabler: While much discussion focuses on how quantum computing will boost AI, the current reality is the reverse: AI is solving the "noise" problem in quantum sensing. AI algorithms filter out environmental magnetic interference to isolate weak signals from quantum sensors, making technologies like magnetic navigation feasible outside of controlled lab environments.
  • The "Sandwich" Strategy in Simulation: To solve complex problems like drug discovery or material science without powerful quantum computers, SandboxAQ uses a "sandwich" approach. They use classical physics simulations (the bottom layer) to generate high-quality data, which then trains AI models (the top layer) to extrapolate and solve problems faster than traditional methods could alone.
  • Active Learning: A specific subset of AI called "active learning" is used to optimize computational resources. Instead of blindly generating data, the AI identifies exactly what piece of missing information would provide the biggest learning gain, directing the physics simulation to compute only that specific data point.
  • Cryptographic Inventorying: The first step in preparing for the quantum threat is situational awareness. Most organizations do not know where their encryption lives or how it is implemented. Modern tools provide a "control center" view to discover keys and algorithms across networks, file systems, and code, moving beyond manual spreadsheet tracking.
  • Regulatory Realities vs. Technological Readiness: In quantum sensing applications like medical diagnostics or aviation navigation, the technology often matures faster than the regulations. Delays in deployment (e.g., waiting 5 years for widespread use) are frequently due to FDA or FAA certification processes rather than engineering hurdles.

Quotes

  • At 2:05 - "Currently, I'd say that the AI in our name is the key enabler for many of the quantum technologies. This is especially true for our sensing division." - explaining the immediate symbiotic relationship where AI makes noisy quantum sensors commercially viable.
  • At 4:01 - "Unlike a MRI machine, which will have a dedicated room that's meant to be a very magnetically quiet space, our device is designed to work in emergency rooms... picking apart the signal from the noise using AI." - highlighting the practical application of AI in moving quantum medical technology from the lab to the real world.
  • At 8:29 - "We pair AI methods with physics simulation methods. We use the physics simulation methods to synthesize new data... to expand the training set of AI systems." - clarifying the hybrid approach used to solve complex simulation problems using today's classical high-performance computing.
  • At 15:53 - "It's very rarely the case that a cryptographic vulnerability is isolated to swapping out a single file. It's much more likely to be a big hairy Gordian knot." - illustrating why automated discovery and remediation tools are necessary for post-quantum migration, as manual fixes are often insufficient.
  • At 18:25 - "We need to start creating standard APIs for software developers to implement common cryptographic patterns and stop giving them only have the option of having enough levers and rope to hang themselves by." - advocating for a shift in how developers interact with cryptography to prevent future security vulnerabilities.

Takeaways

  • Conduct a Cryptographic Inventory Now: Do not wait for final NIST standards to begin discovering where encryption exists in your infrastructure. Use automated tools to create a dynamic inventory of keys and algorithms across networks and code to prepare for the inevitable migration to post-quantum cryptography.
  • Leverage AI for Data Efficiency: When dealing with expensive computational simulations (like material science or chemistry), apply active learning techniques to identify high-value data points rather than brute-forcing calculations, significantly reducing resource costs and time.
  • Plan for Regulatory Lead Times: If developing or adopting quantum sensing hardware for critical sectors like healthcare or transportation, factor in multi-year timelines for regulatory approval (FDA, FAA) that will likely exceed the time required for technical development.