How Quantum Computing & Machine Learning Work Together

We’re all (at least slightly) familiar with the concept of machine learning and AI by now — but just what exactly is quantum computing? If you aren’t scrolling SlashDot and TechCrunch on the daily, quantum computing may have escaped your tech dictionary. The name alone evokes some notion of a complex, sci-fi super-computer type of setup. And guess what? That’s not far off the mark.

We sat down and had a chat with two experts delving into the AI/Quantum computing sphere, Amit Bansal, Managing Director, Analytics Delivery Lead APAC & Artificial Intelligence Delivery Leader at Accenture and Vaibhav Namburi, Director of Five2One & Dveloper.io to enlighten us about what’s in store for the future.

What is quantum computing?

Let’s preface this by saying that we’re not going to go into the finer details of quantum computing here — because we could be here all day.

Bansal instead manages to outline the premise of quantum computing in a nutshell:

“Quantum computers are devices that work based on principles from quantum physics,” he starts. “The computers that we currently use are built using transistors and the data is stored in the form of binary 0 and 1. Quantum computers are built using subatomic particles called quantum bits, qubits for short, which can be in multiple states at the same time. The main advantage of quantum computers are that they can perform highly complex operations at supersonic speeds. Thus, they solve problems that are not currently feasible.”

[Note: if you’re interested in the lengthy answer to this question, visit An Interactive Introduction To Quantum Computing for a good walkthrough.]

Which type of problems can quantum computing solve?

“The most important benefit of quantum computers is the speed at which it can solve complex problems,” says Bansal. While they’re lightning quick at what they do, Bansal notes, “they don’t provide capabilities to solve problems from undecidable or NP Hard problem classes.” There is a problem set that quantum computing will be able to solve, however it’s not applicable for all computing problems.

Typically, the problem set that quantum computers are good at solving involves number or data crunching with a huge amount of inputs, such as “complex optimisation problems and communication systems analysis problems” — calculations that would typically take supercomputers days, years, even billions of years to brute force.

The application that’s regularly trotted out as an example that quantum computers will be able to instantly solve is strong RSA encryption. A recent study by the Microsoft Quantum Team suggests this could well be the case, calculating that it’d be doable with around a 2330 qubit quantum computer.

The most cutting edge quantum computers built by heavyweights like Intel, Microsoft, IBM all are currently hovering at around the 50 qubit mark, however Google have recently announced Bristlecone, their 72-qubit project. Given Moore’s law and the current speed of development of these systems, strong RSA may indeed be cracked within 10 years.

AI and machine learning

The term AI is used fairly broadly these days, however, as Namburi puts it, “AI is a distilled concept that machines will be able to execute tasks characteristic of human intelligence.”

He goes on to elaborate, “Machine Learning (ML) at its core is a simple way of achieving AI, and AI/ML can offer assistance in speeding up and parsing extremely large chunks of data whilst creating and analysing predictive models and trends that will help unravel patterns not easily determined by us.”

Machine learning is a faster way of determining and analysing these patterns (rather than using traditionally-coded algorithms) and can be used for a number of different applications, however, its application in AI is the one that’s got the whole world abuzz.

The intersection of machine learning and quantum computing

You’ve probably guessed by now that quantum computing has the possibility to make machine learning AI solutions exponentially faster at crunching their datasets than their traditional computing counterparts — although you can’t code these ML/AI algorithms in the traditional sense.

However, the intersection of these two fields goes even further than that, and it’s not just AI applications that can benefit. As Bansal explains, there is an “intersecting area where quantum computers implement machine learning algorithms and traditional machine learning methods are employed to assess the quantum computers. This area of research is developing at such blazing speeds that it has spawned an entire new field called Quantum Machine Learning.”

This interdisciplinary field is very, very new though. “Recent work has produced quantum algorithms that could act as the building blocks of machine learning programs, but the hardware and software challenges are still considerable and the development of fully functional quantum computers is still far off,” says Bansal.

The 4 approaches to machine learning, categorised by whether the system under study is classic or quantum, and whether the information processing device is classical or quantum.

The future of AI using quantum computing

The future of AI sped along by quantum computing looks bright, with real-time human-imitable behaviours almost a foregone conclusion.

As Bansal says, quantum computing will be able “to solve complex AI problems and obtain multiple solutions to complex problems simultaneously. This will result in artificial intelligence more efficiently performing complex tasks in human-like ways. Similarly, robots that can make optimised decisions in real time in practical situations will be possible once we are able to employ quantum computers based on Artificial Intelligence.”

How far away will this future be? Well, considering only a handful of the world’s top companies and universities currently are developing (physically huge) quantum computers that currently lack the processing power required, having an army of robots imitating humans running about is probably a fair way off — which may put some people at ease, and disappoint others! Building just one though? Maybe not so far off…

How will classic computing and quantum computing work together?

Quantum computers will never “replace” classic computers, simply because there are some problems that classic computers are better and/or more efficient at solving.

Bansal muses that a “likely future scenario is that quantum computing will augment subroutines of classical algorithms that can be efficiently run on quantum computers, such as sampling, to tackle specific business problems. For instance, a company seeking to find the ideal route for retail deliveries could split the problem into two parts and leverage each computer for its strengths.”

Numburi likes the use case for blockchain, suggesting that it be used to “speed permissions on the extremely laggy Proof of Work system which is necessary for the blockchain to hold true right now. Quantum computers can handle the level of processing computers in this day and age can’t.”

All we can say is stay posted folks!

Want to get involved in a quantum computing or machine learning project? Browse our job listings to find your dream role.