Artificial Intelligence for Quantum Computing

TL;DR

This paper surveys how state-of-the-art AI methods can accelerate the development and operation of quantum computers, spanning hardware design, circuit preprocessing, device control, error correction, and postprocessing. It highlights approaches such as reinforcement learning, transformers, diffusion models, and graph neural networks, applied to tasks from unitary synthesis and pulse optimization to QEC decoding and state tomography. The authors argue that AI is essential for scaling fault-tolerant quantum computing, enabling data-efficient device characterization, robust control, rapid code discovery, and efficient postprocessing, while also outlining the practical challenges of data availability and integration. They further discuss future opportunities in accelerated hybrid quantum-classical platforms, high-quality synthetic data generation, and deeper cross-disciplinary collaboration, envisioning AI-driven breakthroughs that could unlock practical QC applications and FTQC.

Abstract

Artificial intelligence (AI) advancements over the past few years have had an unprecedented and revolutionary impact across everyday application areas. Its significance also extends to technical challenges within science and engineering, including the nascent field of quantum computing (QC). The counterintuitive nature and high-dimensional mathematics of QC make it a prime candidate for AI's data-driven learning capabilities, and in fact, many of QC's biggest scaling challenges may ultimately rest on developments in AI. However, bringing leading techniques from AI to QC requires drawing on disparate expertise from arguably two of the most advanced and esoteric areas of computer science. Here we aim to encourage this cross-pollination by reviewing how state-of-the-art AI techniques are already advancing challenges across the hardware and software stack needed to develop useful QC - from device design to applications. We then close by examining its future opportunities and obstacles in this space.

Figures (7)

• Figure 1: A depiction of the sections covered in this review and how AI can be used to benefit the entire QC stack.
• Figure 2: A simple hierarchy from Artificial Intelligence to generative AI, broadly contextualizing the techniques discussed in this work. Each level is paired with a simple description.
• Figure 3: Schematic of process for training ML models, adapted from youssry2024experimental. The process starts with creating an experimental or simulated dataset by applying controls to the system and recording outputs. This dataset trains ML models, which are then use to determine optimal control settings.
• Figure 4: Workflow of the GPT-QE algorithm. During the initial Operator Preparation stage, operators are extracted (the choice of operators depends on the problem ansatz - UCCSD and QAOA being two examples), resulting in Hermitian operators $\{P_j\}_{j}$ such as Pauli strings. Additionally, a range of discrete coefficients ($\{\theta_k\}_k$) are generated. $\{P_j\}$ and $\{\theta_k\}$ are combined into different unitary pool operators ({e$^{i P_j \theta_k}\}_{j,k}$). During the next GPT token-generation and training stage, the {e$^{i P_j \theta_k}\}$ are tokenized and passed to a transformer for training. In training, the model produces sequences of tokens for which the loss function is computed. These losses are used to update the transformer parameters. Finally, after training, the model is able to generate a 'prediction' of a quantum circuit.
• Figure 5: Most quantum device architectures require specific tuning and control protocols to operate as qubits. Machine learning-based approaches allow us to automate and speed up such protocols, allowing for high-throughput characterization and optimization of quantum devices.