QuEra, Harvard and Anthropic Use Claude to Cut Quantum Circuit Resource Needs by Up to 39x

At a glance

Quantum computing company QuEra worked with Harvard and Anthropic to use the AI model Claude to reduce the computing resources quantum circuits need. For some operations, the new designs were up to 39 times more efficient than before.

Key points

1. 1

   Claude found circuits using fewer gates and qubits for 7 of 8 core operations.

2. 2

   A Reed-Solomon decoder circuit dropped from about 5.1 million Toffoli gates to about 440,000.

3. 3

   The 39x figure is an estimate of required resources, not a sign that quantum computers will run 39 times faster.

4. 4

   The research was published as a paper on arXiv, and an automated checker verified the AI-made circuits.

Why it matters

For quantum computers to be useful, the computing resources that algorithms need must also shrink, not just the hardware. This work shows that AI can help with that design.

Also reported by

• The Quantum InsiderQuEra, Harvard and Anthropic Use AI to Reduce Quantum Circuit Resource Requirements