What Makes Quantum Machine Learning “Quantum”?

Published: 1 month ago (March 6, 2026 at 02:39 PM EST)

7 min read

Source: Towards Data Science

Introduction

I started working in quantum computing seven years ago, right after completing my master’s degree. Back then, the field was buzzing with excitement—but also with skepticism. Today, quantum computing stands alongside high‑performance computing (HPC) and artificial intelligence (AI) as an emerging technology.

The focus has shifted from purely hardware‑centric research to applications, software, and algorithms. Quantum computers are now viewed as tools that can be leveraged across many disciplines, rather than as an isolated niche. One of the most promising (yet still not fully understood) uses of quantum computers is quantum machine learning (QML).

What Is Quantum Machine Learning?

Quantum machine learning has become a catch‑all term over the past few years. A milestone in its history was the 2013 launch of the Quantum Artificial Intelligence Lab by Google and NASA, tasked with exploring how quantum computers could be used for machine learning. Since then, “QML” has appeared in research papers, startup pitches, and conference talks—often with wildly different meanings.

Interpretation	Description
Quantum‑accelerated ML	Using quantum hardware to speed up classical machine‑learning tasks.
Quantum‑inspired classical algorithms	Classical methods that borrow ideas from quantum physics.
ML on quantum hardware	Running a familiar ML workflow on unfamiliar (quantum) hardware.

Even as a researcher in quantum computing, I was initially confused. A common first question is:

What exactly makes quantum machine learning quantum?

The Core Answer

The short answer is not speed, neural networks, or vague references to “quantum advantage.”
At its core, quantum machine learning is defined by how information is represented, transformed, and read out—using the rules of quantum mechanics rather than classical computation.

What to Expect From This Series

This article aims to:

Clarify the distinction between genuine quantum approaches and hype.
Provide a clean conceptual foundation for the rest of the series.
Explore the lore of QML, as well as near‑term research results and applications.

Stay tuned for deeper dives into the theory, algorithms, and practical experiments that shape the evolving landscape of quantum machine learning.

Machine Learning Before “Quantum”

Before we get all quantum, let’s take a step back. Stripped of its modern trappings, machine learning is about learning a mapping from inputs to outputs using data. Regardless of whether the model is a linear regressor, a kernel method, or a deep neural network, the structure is more or less the same:

Data is represented numerically (vectors, matrices, tensors).
A parameterized model transforms that data.
Parameters are adjusted by optimizing a cost function.
The model is evaluated statistically on new samples.

Neural networks, GPUs, and massive datasets are implementation choices—not defining features. This abstraction matters because it lets us ask a precise question:

What changes when the data and the model live in a quantum space?

Quantum Mechanics Enters

Quantum machine learning (QML) becomes “quantum” when quantum information is the computational substrate. This manifests in three distinct ways.

1. Data Is Represented as Quantum States

Classical ML: data → bits or floating‑point numbers.
QML: data → quantum states (complex vectors) described by density matrices; transformations are unitary matrices.

Information is encoded in complex‑valued amplitudes rather than classical probabilities.
States can exist in superposition.

Note: This does not mean that every classical dataset is automatically exponentially compressed or easily accessible. Loading data into quantum states is often costly, and extracting information is fundamentally limited by measurement.

Key point: The model operates on quantum states, not on classical numbers.

2. Models Are Quantum Evolutions

Classical ML: apply deterministic functions to data.
QML: apply quantum operations (typically unitary transformations) to quantum channels.

In practice, most QML models are built from parameterized quantum circuits—sequences of quantum gates whose parameters are tuned during training, analogous to adjusting weights in a neural network.

The system starts in a state described by a matrix (often a Hamiltonian).
The applied gates dictate how the system evolves over time, which defines the model’s behavior.

Consequently, quantum models explore a hypothesis space that is structurally different from that of classical models, even when the training loop looks similar on the surface.

3. Measurement Is Part of the Learning Process

Classical ML: reading a model’s output is trivial and does not affect the model.
QML: measurement is probabilistic and destructive; it collapses the quantum state.

Outputs are obtained by repeated circuit executions called shots (running the same circuit many times to estimate probabilities).
Gradients are estimated statistically from these measurements rather than computed exactly, so training cost is often dominated by sampling noise rather than raw computation.

Bottom line: Uncertainty is built into the model itself. Any serious discussion of QML must account for the fact that learning happens through measurement, not after it.

What Doesn’t Make QML Quantum

Quantum computing—and quantum machine learning (QML) in particular)—often generate hype and misunderstanding. Many approaches that carry the “quantum” label are, in fact, not quantum in any meaningful sense. Typical examples include:

Classical ML algorithms run on quantum hardware without exploiting quantum states in a substantive way.
“Quantum‑inspired” methods that are entirely classical.
Hybrid pipelines where the quantum component can be removed without affecting the model’s behavior or performance.

Quick Test

If you encounter a claim about a QML model and are unsure how quantum it really is, ask:

“Can I replace the quantum part with a classical one without altering the model’s mathematical structure?”

Yes (or “maybe”) → The approach is probably not fundamentally quantum.
No → The method likely leverages genuine quantum properties.

Even non‑quantum approaches can be valuable, but they fall outside the core of quantum machine learning.

Where Is QML Today?

When discussing quantum computing, keep in mind that current hardware is noisy, small, and resource‑constrained. Consequently:

There is no general, proven quantum advantage for machine‑learning tasks today.
Many QML models resemble kernel methods more than deep neural networks.
Data loading and noise often dominate performance.

This isn’t a failure of the field; it reflects the present state of quantum computing. Most QML research is now exploratory: mapping model classes, understanding quantum learning theory, and pinpointing where quantum structure could provide an advantage.

Why Quantum Machine Learning Is Still Worth Studying

If near‑term speedups are unlikely, why pursue QML at all?

Fundamental insights – QML forces us to rethink foundational questions about both machine learning and quantum computing.
Defining learning on quantum data – We must answer what it truly means to learn from quantum data, how noise influences optimization, and which model classes exist in quantum systems but have no classical counterpart.

The broader perspective

Quantum machine learning is less about beating classical ML today and more about expanding the space of what “learning” can mean in a quantum world.

Scientific progress – New approaches often spark breakthroughs; exploring QML now prepares us for future hardware advances.
Future‑proofing – Even if current hardware isn’t ready, the concepts, algorithms, and theory we develop today will be ready when the technology catches up.

Final Thoughts and What Comes Next

Advances in quantum computing are accelerating. Hardware companies are racing to build a fault‑tolerant quantum computer—a machine that utilizes the full power of quantum mechanics. Meanwhile, software and application firms are exploring the problems that quantum computing can meaningfully address.

That said, today’s quantum computers are incapable of running a near‑life‑sized application, let alone a complex machine‑learning model. Still, the promise of quantum‑computing efficiency in machine learning is intriguing and worth exploring now, in parallel with hardware advancements.

In this article I focused on the definitions and boundaries of quantum machine learning to pave the way for future pieces that will explore:

How classical data is embedded into quantum states.
Variational quantum models and their limitations.
Quantum kernels and feature spaces.
Optimization challenges in noisy quantum systems.
Where quantum advantage might plausibly emerge.

Before asking whether quantum machine learning is useful, we need to be clear about what it actually is. The more we step away from the hype, the closer we can move toward genuine progress.