Quantum ML: 26 of 29 top papers aren’t ready

We read the full text of 29 of the most-cited papers on quantum machine learning — ranked by citations in OpenAlex, October 2026 — and judged each one by the same rules as every brief on this site. Two different AI models read each paper independently; a paper is listed only where both reached the same verdict.

26 of the 29 carry the verdict “Not yet”. That is not a verdict of failure. Each shows real progress, but none states how far it is from a real task in a business: some are demonstrations at toy size without a statement of what a real problem would need, some improve a component without moving any estimate, some name no industrial task at all. For a reader deciding where to spend, that is the finding: the field's most influential work does not yet measure its distance to use.

2 papers put a number on that distance (“Closer”), and one claims a practical advantage that its own evidence does not support (“Category error”).

Not yet (26)

Paper	Year	Citations	How it was done
Supervised learning with quantum enhanced feature spaces1	2018	2,515	real quantum chip
The power of quantum neural networks2	2020	1,023	simulation on a classical computer
Quantum advantage in learning from experiments3	2021	509	real quantum chip
Experimental Quantum Generative Adversarial Networks for Image Generation4	2020	277	real quantum chip
Diagnosing Barren Plateaus with Tools from Quantum Optimal Control5	2021	214	calculation on paper
Exponential concentration in quantum kernel methods6	2022	189	calculation on paper
Quantum annealing versus classical machine learning applied to a simplified computational biology problem7	2018	167	real quantum chip
Machine-learning-assisted correction of correlated qubit errors in a topological code8	2017	141	simulation on a classical computer
Quantum computing models for artificial neural networks9	2021	118	calculation on paper
Higher Order Derivatives of Quantum Neural Networks with Barren Plateaus10	2020	118	calculation on paper
Decoding Small Surface Codes with Feedforward Neural Networks11	2017	117	simulation on a classical computer
Mixed Quantum-Classical Method For Fraud Detection with Quantum Feature Selection12	2022	110	simulation on a classical computer
Hybrid quantum neural network for drug response prediction13	2022	109	simulation on a classical computer
Quantum Annealing for Combinatorial Clustering14	2017	106	real quantum chip
Time Series Quantum Reservoir Computing with Weak and Projective Measurements15	2022	90	simulation on a classical computer
Application of Quantum Machine Learning using the Quantum Kernel Algorithm on High Energy Physics Analysis at the LHC16	2021	87	simulation on a classical computer
On the practical usefulness of the Hardware Efficient Ansatz17	2022	79	calculation on paper
Comparing neural network based decoders for the surface code18	2018	76	simulation on a classical computer
Financial Fraud Detection: A Comparative Study of Quantum Machine Learning Models19	2023	65	simulation on a classical computer
A semi-agnostic ansatz with variable structure for quantum machine learning20	2021	58	simulation on a classical computer
Data compression for quantum machine learning21	2022	55	simulation on a classical computer
Improved Financial Forecasting via Quantum Machine Learning22	2023	55	simulation on a classical computer
Experimental quantum kernel machine learning with nuclear spins in a solid23	2019	52	real quantum chip
Quantum Deep Hedging24	2023	34	real quantum chip
Data-driven decoding of quantum error correcting codes using graph neural networks25	2023	21	simulation on a classical computer
Convolutional neural network based decoders for surface codes26	2023	17	simulation on a classical computer

Of these, 7 ran on real quantum hardware; the rest were simulated on classical computers or worked out on paper.

Closer (2)

• A scalable and fast artificial neural network syndrome decoder for surface codes²⁷ (2021, 42 citations; simulation on a classical computer).

  “We note that the use of the ANN decoder improves the threshold to approximately 13.8%, when compared to 11.5% threshold from non-ML HDRG decoder.”

• Neural network decoder for near-term surface-code experiments²⁸ (2023, 20 citations; simulation on a classical computer).

  “When applied to the experimental data of [Google Quantum AI, Nature 614, 676 (2023)], the neural network decoder achieves logical error rates approximately 25% lower than minimum-weight perfect matching, approaching the performance of a maximum-likelihood decoder.”

Category error (1)

• Quantum Long Short-Term Memory (QLSTM) vs Classical LSTM in Time Series Forecasting: A Comparative Study in Solar Power Forecasting²⁹ (2023, 58 citations; simulation on a classical computer).

  “Quantitative results demonstrate that QLSTMs achieve up to 50% improvement in accuracy and 85.7% faster convergence compared to their classical LSTM counterparts.”

How this was made

Papers were found through OpenAlex by citation count and read in full by a language model; every quotation, number and name behind each line was checked by code against the paper's text. A second model read each paper again without seeing the first result. 3 review articles are left out, because a review makes no claim of its own to judge. 2 papers where the two readings disagreed are left out. Citation counts are as of October 2026. The verdicts and the rules behind them are set out under Method. Responsible for the rules and for this page: Georg Singer. Georg Singer holds no shares in any company or institution named in this brief (as of October 2026).

Sources

1. arxiv.org Supervised learning with quantum enhanced feature spaces
2. arxiv.org The power of quantum neural networks
3. arxiv.org Quantum advantage in learning from experiments
4. arxiv.org Experimental Quantum Generative Adversarial Networks for Image Generation
5. arxiv.org Diagnosing Barren Plateaus with Tools from Quantum Optimal Control
6. arxiv.org Exponential concentration in quantum kernel methods
7. arxiv.org Quantum annealing versus classical machine learning applied to a simplified computational biology problem
8. arxiv.org Machine-learning-assisted correction of correlated qubit errors in a topological code
9. arxiv.org Quantum computing models for artificial neural networks
10. arxiv.org Higher Order Derivatives of Quantum Neural Networks with Barren Plateaus
11. arxiv.org Decoding Small Surface Codes with Feedforward Neural Networks
12. arxiv.org Mixed Quantum-Classical Method For Fraud Detection with Quantum Feature Selection
13. arxiv.org Hybrid quantum neural network for drug response prediction
14. arxiv.org Quantum Annealing for Combinatorial Clustering
15. arxiv.org Time Series Quantum Reservoir Computing with Weak and Projective Measurements
16. arxiv.org Application of Quantum Machine Learning using the Quantum Kernel Algorithm on High Energy Physics Analysis at the LHC
17. arxiv.org On the practical usefulness of the Hardware Efficient Ansatz
18. arxiv.org Comparing neural network based decoders for the surface code
19. arxiv.org Financial Fraud Detection: A Comparative Study of Quantum Machine Learning Models
20. arxiv.org A semi-agnostic ansatz with variable structure for quantum machine learning
21. arxiv.org Data compression for quantum machine learning
22. arxiv.org Improved Financial Forecasting via Quantum Machine Learning
23. arxiv.org Experimental quantum kernel machine learning with nuclear spins in a solid
24. arxiv.org Quantum Deep Hedging
25. arxiv.org Data-driven decoding of quantum error correcting codes using graph neural networks
26. arxiv.org Convolutional neural network based decoders for surface codes
27. arxiv.org A scalable and fast artificial neural network syndrome decoder for surface codes
28. arxiv.org Neural network decoder for near-term surface-code experiments
29. arxiv.org Quantum Long Short-Term Memory (QLSTM) vs Classical LSTM in Time Series Forecasting: A Comparative Study in Solar Power Forecasting