Anthropic says Claude designed quantum circuits that beat best published human designs
Anthropic published research on October 16 saying its Claude model (referred to as Mythos 5 in some Chinese-language coverage) designed new quantum circuits for eight tasks spanning chemistry, optimization, physics simulation, number theory and numerical computation, working from plain-language descriptions with almost no human intervention; an automated evaluator scored each proposal on qubits × gates and Claude iterated on the feedback in a closed loop. According to Anthropic, the designs beat the best previously published circuits on all eight tasks: for elliptic-curve point addition the model spent about 53 hours, derived a series of new algebraic identities (formalized in Lean) and produced a reversible divsteps implementation that stores no branch history, using roughly 12% more qubits but about 17 times fewer gates, while in the decoding step of quantum interference measurement it swapped in a Berlekamp-Massey decoder and cut gate count roughly 12-fold. The researchers caution that circuit design is only one piece of building a fault-tolerant quantum computer, with error correction and hardware control still needing progress. Why it matters: If the results hold up to outside review, it suggests AI can supply genuinely new mathematical structure to hardware-constrained science, potentially lowering the qubit counts quantum applications need. What do you think?
Open the headline: source, AI briefing and more →Comments
…