Quantum computing is becoming more accessible thanks to a new AI tool developed by researchers at Pasqal, a quantum computing startup based in Paris. The autonomous agent can translate plain English descriptions into working quantum computing code and execute experiments on real quantum computers.
The concept, described in a preprint posted on the arXiv server, brings 'vibe coding' - where users describe what they want in natural language and AI generates the code - to the specialized world of quantum computing.
Christophe Jurczak, a co-founder of Pasqal, demonstrated the tool's potential by running experiments that would typically require a team of specialized quantum physicists. 'I can do it on my own, from my couch in Dallas, Texas,' Jurczak said.
The AI agent works by first learning the technical specifications of Pasqal's quantum computers, which encode information in arrays of atoms trapped with laser light. When researchers describe a physical phenomenon they want to simulate, the agent translates this into quantum code, tests it on virtual versions of the hardware, and then runs it on actual quantum computers in Saudi Arabia or Canada.
In three test cases, the agent successfully simulated materials with magnetic properties - a task that normally requires expertise in both the physics of materials and quantum computing. The agent demonstrated what the researchers called 'a firm grasp of the hardware constraints,' even correctly explaining when a simulation was too complex for the available machines.
However, the researchers note that the AI still sometimes requires human guidance. In one test case, the agent needed significant hand-holding from human researchers to achieve what they described as a 'physically accurate implementation.'
The tool focuses on quantum simulations, one of the most promising applications of quantum computers. These simulations tune quantum computers to model the behavior of other physical systems like catalysts or materials with unusual magnetic properties - calculations that would overwhelm classical computers.
The development could accelerate research in materials science, drug discovery, and other fields by making quantum computing accessible to scientists who specialize in their domain rather than quantum programming.
