Quantum machine learning is the research field that explores how quantum computers could be used for machine learning – and, conversely, how machine learning can help build and control quantum computers. Its central question is whether quantum hardware can learn from data faster or better than classical computers.
The promise
Quantum computers manipulate qubits, which can be in superpositions of states and become entangled, letting algorithms exploit interference. In 2009 Aram Harrow, Avinatan Hassidim and Seth Lloyd described a quantum algorithm for solving certain systems of linear equations exponentially faster than known classical methods under strict conditions, which inspired proposals for quantum versions of recommendation systems, support vector machines and neural networks. A widely cited 2017 review in Nature by Jacob Biamonte and colleagues set out the field’s agenda.
Sobering results
Many claimed speedups shrank under scrutiny. In 2018 Ewin Tang, then an undergraduate, showed that a classical algorithm could match the quantum recommendation algorithm’s speed up to polynomial factors, triggering a wave of such “dequantizations”. Loading large classical datasets into quantum states is itself costly, and training of quantum circuits can stall on flat landscapes known as “barren plateaus”. Current machines remain small and error-prone.
Progress in hardware
Hardware has advanced: Google’s Willow chip (December 2024) demonstrated error correction that improves as more qubits are added, and in October 2025 Google reported a “verifiable quantum advantage” on a physics benchmark with an algorithm it called Quantum Echoes; its first molecular application did not yet beat classical methods. Most researchers expect the first practical uses in simulating chemistry and materials rather than in mainstream AI. Meanwhile, classical AI is helping quantum computing, for example by decoding errors.
Why it matters
Quantum machine learning is a frequent subject of hype; a sober view separates real but narrow results from claims that quantum computers will soon transform AI (see Hype Cycle).
Written with Claude, made by Anthropic, which does not build quantum computers; Google, whose quantum results are described here, is an Anthropic investor and competitor, and investors Amazon, Microsoft and Nvidia also run quantum computing programs.
Related
Hype Cycle · Materials Discovery · AI Accelerator · Neural Network · Support Vector Machine · Neuromorphic Computing
Sources: Jacob Biamonte et al.: Quantum machine learning (Nature) (2017) · Ewin Tang: A quantum-inspired classical algorithm for recommendation systems (arXiv) (2018) · Google: Our Quantum Echoes algorithm is a big step toward real-world applications for quantum computing (October 2025)
Leave a Reply