Anthropic's Claude designs efficient quantum circuits, improving performance significantly

Claude: Claude is Anthropic’s family of AI models designed for advanced reasoning, coding, and scientific problem-solving tasks. In the news, Claude Mythos 5 worked autonomously in a closed loop with an evaluator to generate novel quantum circuit designs for primitives drawn from chemistry, optimization, physics simulation, and number theory. The model uncovered new structural insights that produced more efficient circuits than prior published work. Andi Gu: Andi Gu is a researcher who led the quantum circuit optimization project. The effort was carried out as part of Anthropic’s STEM Fellows Program under mentorship from company quantum physicists and in collaboration with Harvard researchers. Anthropic: Anthropic is an AI research company developing capable and safe AI systems, including the Claude models. It conducted this project through its STEM Fellows Program, with internal mentorship and external academic collaboration, to explore AI assistance in quantum algorithm design. The work demonstrates the company’s focus on practical AI applications across scientific domains. Alek Kemeny: Alek Kemeny is a quantum physicist at Anthropic who served as a mentor on the quantum circuit design initiative. His involvement supported the technical execution of the AI-assisted research within the company’s STEM Fellows Program. Mikhail Lukin: Mikhail Lukin is a researcher affiliated with Harvard University and QuEra Computing who collaborated on the quantum circuit design work. His involvement brought expertise from both academia and quantum hardware development to the Anthropic initiative. Susanne Yelin: Susanne Yelin is a researcher at Harvard University who collaborated on the project exploring AI-designed quantum circuits. Her participation contributed academic expertise in quantum physics to the Anthropic-led effort. AI-Assisted Discovery: AI models can autonomously iterate on quantum circuit designs by proposing, evaluating, and refining implementations with minimal human input. Quantum Resource Efficiency: Redesigning quantum circuits to use fewer qubits and gates makes algorithms for chemistry, materials science, and optimization more feasible on hardware with limited qubits and coherence times. Interdisciplinary Collaboration: Projects combining AI capabilities with quantum physics expertise often involve partnerships between AI labs and academic institutions focused on quantum simulation and hardware.