How quantum computing may train AI and autonomous vehicles

In less than a century since its inception in 1936, traditional computing has revolutionized nearly every industry in the world. And as traditional computing gives way to new frontiers, Artificial Intelligence (AI) will likely be the source of the next wave of innovation.

Given the progress of innovation in the AI field, the limitations of traditional computing quickly come to light, as AI continually pushes the existing boundaries of computing speed and power efficiency. Complex software applications—like those used in autonomous vehicles, chemical reaction simulation, and 3D dynamic simulations—are continually pushing the capabilities of traditional computing technology.

Here, we'll discuss the fundamental differences between quantum and traditional computing and examine quantum computing's potential to improve AI algorithms in autonomous vehicles and other compute-intensive software applications.

What Is Quantum Computing?

Quantum computing is not intended to replace traditional computing. Traditional computers are actually better than quantum computers for most computing tasks. Quantum computers, however, are much better at solving challenging computational problems.

While traditional computers use binary transistor "bits" (i.e., 0 or 1) to process data, quantum computers use quantum bits, also called qubits, representing three different states. The third state of a qubit is called "superposition"; it allows qubits to represent both a 0 and a 1 simultaneously.

Given the third superposition state, quantum computing power increases exponentially with the number of qubits, whereas computer power increases in a 1:1 ratio with the number of transistor bits. For example, if you had two classical bits, the total possible number of value combinations is four (00, 01, 10, 11).

But two qubits can exist in a superposition of four states, meaning the total combinations of two qubits is eight different states. Therefore, the information stored in "N" qubits is equal to the data stored in "2^N" classical bits.

To translate this information into more relatable data storage numbers, 3 qubits is equal to 8 bits (or 1byte), while 13 qubits are equal to 8192 bits (1kilobyte). If you were to store 1terbyte (8.8x10^12 bits) of information using qubits, you would only need 43 of them!

Given this drastic advantage in computing power, quantum computers are better suited for complex computation problems. These include optimization challenges like simulating chemical reactions, drug identification, multi-stop logistics, and artificial intelligence algorithms. In the world of autonomous vehicle training, this means much more training data could be stored on and processed by a quantum computer than a traditional data center, allowing more data to be used to educate the model and thus yield a more accurate model, faster.

In 2019, Google experimented with a 54-qubit processor named "Sycamore," which aimed to complete a computational experiment that would have taken the world's fastest supercomputer 10,000 years to complete. Sycamore conducted the experiment in 200 seconds. In an autonomous vehicle AI model training application, a quantum computer would be able to process 10,000 years of training data in 200 seconds, allowing for infinitely more opportunities for AV model trainers.

What Is Quantum AI?

AI is the modern frontier of software. While there are many facets and sub-genres of AI today, it's still in its infancy relative to its full potential, and the development of AI is often slow-moving.

Artificial intelligence training models are incredibly complex to develop, use, and iterate upon. Often, the execution of these algorithms can take hours, days, weeks, or even months, depending on their complexity. Today, autonomous vehicles ‘Race to Full Autonomy’ is considered the pinnacle of AI and has long been the proving ground for many AI training models. Quantum AI may be the solution that super-charges autonomous vehicle AI training models across the finish line.

Quantum AI aims to alleviate modern AI of the constraints of traditional computing. Optimization algorithms and models are fundamental to most AI, but traditional computation is limited in its ability to complete many training models efficiently.

Soon, quantum computing may be used to provide faster computation for critically challenging computing processes and create new insights into existing computation foundations.

For example, training a Tesla self-driving car neural network requires incredible amounts of parallel computing and incredibly vast data sets. In fact, Tesla collects terabytes of data per day as a means of training their autonomous vehicle algorithms. The data centers that train these modern self-driving car neural networks can often only train subsections of their neural network, given the limitations of the hardware’s compute throughout.

In this example, it is likely that fully retraining Tesla’s entire neural network would be far too inefficient. However, since a quantum computer can achieve record-setting parallel computation speeds that no data center on the planet can match, naturally, quantum computing can provide value to training complex neural networks such as Tesla’s in a matter of minutes.

The Quantum Challenge

Today, the software for quantum computing is far behind the hardware's capabilities. This is because the infancy of quantum-based AI is limited by classical training algorithms and computing methodologies that have been slowly developed since 1936. The inherent value of these training algorithms cannot be realized using a quantum computer given the fundamental differences in computation methods (bits vs qubits).

Therefore, new quantum algorithms must be created to utilize quantum computers' computing power. There has been significant progress toward machine learning-based quantum algorithms, such as the HLL Algorithm (quantum algorithm for linear systems of equations).

Research groups have only recently begun working on quantum deep neural network training algorithms that may someday prove more useful. One thing is for sure, however: once quantum AI training algorithms have been developed and iterated upon, quantum computers will provide incredible break-neck processing speed that may help train the next generation of autonomous vehicles, computer-based chemistry, and even the simulation of future quantum systems.


Related news articles

Latest News

Sorry, your filter selection returned no results.

We've updated our privacy policy. Please take a moment to review these changes. By clicking I Agree to Arrow Electronics Terms Of Use  and have read and understand the Privacy Policy and Cookie Policy.

Our website places cookies on your device to improve your experience and to improve our site. Read more about the cookies we use and how to disable them here. Cookies and tracking technologies may be used for marketing purposes.
By clicking “Accept”, you are consenting to placement of cookies on your device and to our use of tracking technologies. Click “Read More” below for more information and instructions on how to disable cookies and tracking technologies. While acceptance of cookies and tracking technologies is voluntary, disabling them may result in the website not working properly, and certain advertisements may be less relevant to you.
We respect your privacy. Read our privacy policy here