“Talking quantum circuits”
Interpretable and scalable quantum natural language processing
The central question that pre-occupies our team has been:
“How can quantum structures and quantum computers contribute to the effectiveness of AI?”
In previous work we have made notable advances in answering this question, and this article is based on our most recent work in the new papers [arXiv:2406.17583, arXiv:2408.06061], and most notably the experiment in [arXiv:2409.08777].
This article is one of a series that we will be publishing alongside further advances – advances that are accelerated by access to the most powerful quantum computers available.
Large language Models (LLMs) such as ChatGPT are having an impact on society across many walks of life. However, as users have become more familiar with this new technology, they have also become increasingly aware of deep-seated and systemic problems that come with AI systems built around LLM’s.
The primary problem with LLMs is that nobody knows how they work - as inscrutable “black boxes” they aren’t “interpretable”, meaning we can’t reliably or efficiently control or predict their behavior. This is unacceptable in many situations. In addition, Modern LLMs are incredibly expensive to build and run, costing serious – and potentially unsustainable –amounts of power to train and use. This is why more and more organizations, governments, and regulators are insisting on solutions.
But how can we find these solutions, when we don’t fully understand what we are dealing with now?1
At Quantinuum, we have been working on natural language processing (NLP) using quantum computers for some time now. We are excited to have recently carried out experiments [arXiv: 2409.08777] which demonstrate not only how it is possible to train a model for a quantum computer in a scalable manner, but also how to do this in a way that is interpretable for us. Moreover, we have promising theoretical indications of the usefulness of quantum computers for interpretable NLP [arXiv:2408.06061].
In order to better understand why this could be the case, one needs to understand the ways in which meanings compose together throughout a story or narrative. Our work towards capturing them in a new model of language, which we call DisCoCirc, is reported on extensively in this previous blog post from 2023.
In new work referred to in this article, we embrace “compositional interpretability” as proposed in [arXiv:2406.17583] as a solution to the problems that plague current AI. In brief, compositional interpretability boils down to being able to assign a human friendly meaning, such as natural language, to the components of a model, and then being able to understand how they fit together2.
A problem currently inherent to quantum machine learning is that of being able to train at scale. We avoid this by making use of “compositional generalization”. This means we train small, on classical computers, and then at test time evaluate much larger examples on a quantum computer. There now exist quantum computers which are impossible to simulate classically. To train models for such computers, it seems that compositional generalization currently provides the only credible path.
1. Text as circuits
DisCoCirc is a circuit-based model for natural language that turns arbitrary text into “text circuits” [arXiv:1904.03478, arXiv:2301.10595, arXiv:2311.17892]. When we say that arbitrary text becomes ‘text-circuits’ we are converting the lines of text, which live in one dimension, into text-circuits which live in two-dimensions. These dimensions are the entities of the text versus the events in time.
To see how that works, consider the following story. In the beginning there is Alex and Beau. Alex meets Beau. Later, Chris shows up, and Beau marries Chris. Alex then kicks Beau.
The content of this story can be represented as the following circuit:
2. From text circuits to quantum circuits
Such a text circuit represents how the ‘actors’ in it interact with each other, and how their states evolve by doing so. Initially, we know nothing about Alex and Beau. Once Alex meets Beau, we know something about Alex and Beau’s interaction, then Beau marries Chris, and then Alex kicks Beau, so we know quite a bit more about all three, and in particular, how they relate to each other.
Let’s now take those circuits to be quantum circuits.
In the last section we will elaborate more why this could be a very good choice. For now it’s ok to understand that we simply follow the current paradigm of using vectors for meanings, in exactly the same way that this works in LLMs. Moreover, if we then also want to faithfully represent the compositional structure in language3, we can rely on theorem 5.49 from our book Picturing Quantum Processes, which informally can be stated as follows:
If the manner in which meanings of words (represented by vectors) compose obeys linguistic structure, then those vectors compose in exactly the same way as quantum systems compose.4
In short, a quantum implementation enables us to embrace compositional interpretability, as defined in our recent paper [arXiv:2406.17583].
3. Text circuits on our quantum computer
So, what have we done? And what does it mean?
We implemented a “question-answering” experiment on our Quantinuum quantum computers, for text circuits as described above. We know from our new paper [arXiv:2408.06061] that this is very hard to do on a classical computer due to the fact that as the size of the texts get bigger they very quickly become unrealistic to even try to do this on a classical computer, however powerful it might be. This is worth emphasizing. The experiment we have completed would scale exponentially using classical computers – to the point where the approach becomes intractable.
The experiment consisted of teaching (or training) the quantum computer to answer a question about a story, where both the story and question are presented as text-circuits. To test our model, we created longer stories in the same style as those used in training and questioned these. In our experiment, our stories were about people moving around, and we questioned the quantum computer about who was moving in the same direction at the end of the stories. A harder alternative one could imagine, would be having a murder mystery story and then asking the computer who was the murderer.
And remember - the training in our experiment constitutes the assigning of quantum states and gates to words that occur in the text.
4. Compositional generalization
The major reason for our excitement is that the training of our circuits enjoys compositional generalization. That is, we can do the training on small-scale ordinary computers, and do the testing, or asking the important questions, on quantum computers that can operate in ways not possible classically. Figure 4 shows how, despite only being trained on stories with up to 8 actors, the test accuracy remains high, even for much longer stories involving up to 30 actors.
Training large circuits directly in quantum machine learning, leads to difficulties which in many cases undo the potential advantage. Critically - compositional generalization allows us to bypass these issues.
5. Real-world comparison: ChatGPT
We can compare the results of our experiment on a quantum computer, to the success of a classical LLM ChatGPT (GPT-4) when asked the same questions.
What we are considering here is a story about a collection of characters that walk in a number of different directions, and sometimes follow each other. These are just some initial test examples, but it does show that this kind of reasoning is not particularly easy for LLMs.
The input to ChatGPT was:
What we got from ChatGPT:
Can you see where ChatGPT went wrong?
ChatGPT’s score (in terms of accuracy) oscillated around 50% (equivalent to random guessing). Our text circuits consistently outperformed ChatGPT on these tasks. Future work in this area would involve looking at prompt engineering – for example how the phrasing of the instructions can affect the output, and therefore the overall score.
Of course, we note that ChatGPT and other LLM’s will issue new versions that may or may not be marginally better with ‘question-answering’ tasks, and we also note that our own work may become far more effective as quantum computers rapidly become more powerful.
6. What’s next?
We have now turned our attention to work that will show that using vectors to represent meaning and requiring compositional interpretability for natural language takes us mathematically natively into the quantum formalism. This does not mean that there doesn't exist an efficient classical method for solving specific tasks, and it may be hard to prove traditional hardness results whenever there is some machine learning involved. This could be something we might have to come to terms with, just as in classical machine learning.
At Quantinuum we possess the most powerful quantum computers currently available. Our recently published roadmap is going to deliver more computationally powerful quantum computers in the short and medium term, as we extend our lead and push towards universal, fault tolerant quantum computers by the end of the decade. We expect to show even better (and larger scale) results when implementing our work on those machines. In short, we foresee a period of rapid innovation as powerful quantum computers that cannot be classically simulated become more readily available. This will likely be disruptive, as more and more use cases, including ones that we might not be currently thinking about, come into play.
Interestingly and intriguingly, we are also pioneering the use of powerful quantum computers in a hybrid system that has been described as a ‘quantum supercomputer’ where quantum computers, HPC and AI work together in an integrated fashion and look forward to using these systems to advance our work in language processing that can help solve the problem with LLM’s that we highlighted at the start of this article.
1 And where do we go next, when we don’t even understand what we are dealing with now? On previous occasions in the history of science and technology, when efficient models without a clear interpretation have been developed, such as the Babylonian lunar theory or Ptolemy’s model of epicycles, these initially highly successful technologies vanished, making way for something else.
2 Note that our conception of compositionality is more general than the usual one adopted in linguistics, which is due to Frege. A discussion can be found in [arXiv: 2110.05327].
3 For example, using pregroups here as linguistic structure, which are the cups and caps of PQP.
4 That is, using the tensor product of the corresponding vector spaces.
About Quantinuum
Quantinuum, the world’s largest integrated quantum company, pioneers powerful quantum computers and advanced software solutions. Quantinuum’s technology drives breakthroughs in materials discovery, cybersecurity, and next-gen quantum AI. With over 500 employees, including 370+ scientists and engineers, Quantinuum leads the quantum computing revolution across continents.
The most common question in the public discourse around quantum computers has been, “When will they be useful?” We have an answer.
Very recently in Nature we announced a successful demonstration of a quantum computer generating certifiable randomness, a critical underpinning of our modern digital infrastructure. We explained how we will be taking a product to market this year, based on that advance – one that could only be achieved because we have the world’s most powerful quantum computer.
Today, we have made another huge leap in a different domain, providing fresh evidence that our quantum computers are the best in the world. In this case, we have shown that our quantum computers can be a useful tool for advancing scientific discovery.
Understanding magnetism
Our latest paper shows how our quantum computer rivals the best classical approaches in expanding our understanding of magnetism. This provides an entry point that could lead directly to innovations in fields from biochemistry, to defense, to new materials. These are tangible and meaningful advances that will deliver real world impact.
To achieve this, we partnered with researchers from Caltech, Fermioniq, EPFL, and the Technical University of Munich. The team used Quantinuum’s System Model H2 to simulate quantum magnetism at a scale and level of accuracy that pushes the boundaries of what we know to be possible.
As the authors of the paper state:
“We believe the quantum data provided by System Model H2 should be regarded as complementary to classical numerical methods, and is arguably the most convincing standard to which they should be compared.”
Our computer simulated the quantum Ising model, a model for quantum magnetism that describes a set of magnets (physicists call them ‘spins’) on a lattice that can point up or down, and prefer to point the same way as their neighbors. The model is inherently “quantum” because the spins can move between up and down configurations by a process known as “quantum tunneling”.
Gaining material insights
Researchers have struggled to simulate the dynamics of the Ising model at larger scales due to the enormous computational cost of doing so. Nobel laureate physicist Richard Feynman, who is widely considered to be the progenitor of quantum computing, once said, “it is impossible to represent the results of quantum mechanics with a classical universal device.” When attempting to simulate quantum systems at comparable scales on classical computers, the computational demands can quickly become overwhelming. It is the inherent ‘quantumness’ of these problems that makes them so hard classically, and conversely, so well-suited for quantum computing.
These inherently quantum problems also lie at the heart of many complex and useful material properties. The quantum Ising model is an entry point to confront some of the deepest mysteries in the study of interacting quantum magnets. While rooted in fundamental physics, its relevance extends to wide-ranging commercial and defense applications, including medical test equipment, quantum sensors, and the study of exotic states of matter like superconductivity.
Instead of tailored demonstrations that claim ‘quantum advantage’ in contrived scenarios, our breakthroughs announced this week prove that we can tackle complex, meaningful scientific questions difficult for classical methods to address. In the work described in this paper, we have proved that quantum computing could be the gold standard for materials simulations. These developments are critical steps toward realizing the potential of quantum computers.
With only 56 qubits in our commercially available System Model H2, the most powerful quantum system in the world today, we are already testing the limits of classical methods, and in some cases, exceeding them. Later this year, we will introduce our massively more powerful 96-qubit Helios system - breaching the boundaries of what until recently was deemed possible.
The marriage of AI and quantum computing is going to have a widespread and meaningful impact in many aspects of our lives, combining the strengths of both fields to tackle complex problems.
Quantum and AI are the ideal partners. At Quantinuum, we are developing tools to accelerate AI with quantum computers, and quantum computers with AI. According to recent independent analysis, our quantum computers are the world’s most powerful, enabling state-of-the-art approaches like Generative Quantum AI (Gen QAI), where we train classical AI models with data generated from a quantum computer.
We harness AI methods to accelerate the development and performance of our full quantum computing stack as opposed to simply theorizing from the sidelines. A paper in Nature Machine Intelligence reveals the results of a recent collaboration between Quantinuum and Google DeepMind to tackle the hard problem of quantum compilation.
The work shows a classical AI model supporting quantum computing by demonstrating its potential for quantum circuit optimization. An AI approach like this has the potential to lead to more effective control at the hardware level, to a richer suite of middleware tools for quantum circuit compilation, error mitigation and correction, even to novel high-level quantum software primitives and quantum algorithms.
An AI power-up for circuit optimization
The joint Quantinuum-Google DeepMind team of researchers tackled one of quantum computing’s most pressing challenges: minimizing the number of highly expensive but essential T-gates required for universal quantum computation. This is important specifically for the fault-tolerant regime, which is becoming increasingly relevant as quantum error correction protocols are being explored on rapidly developing quantum hardware. The joint team of researchers adapted AlphaTensor, Google DeepMind’s reinforcement learning AI system for algorithm discovery, which was introduced to improve the efficiency of linear algebra computations. The team introduced AlphaTensor-Quantum, which takes as input a quantum circuit and returns a new, more efficient one in terms of number of T-gates, with exactly the same functionality!
AlphaTensor-Quantum outperformed current state-of-the art optimization methods and matched the best human-designed solutions across multiple circuits in a thoroughly curated set of circuits, chosen for their prevalence in many applications, from quantum arithmetic to quantum chemistry. This breakthrough shows the potential for AI to automate the process of finding the most efficient quantum circuit. This is the first time that such an AI model has been put to the problem of T-count reduction at such a large scale.
A quantum power-up for machine learning
The symbiotic relationship between quantum and AI works both ways. When AI and quantum computing work together, quantum computers could dramatically accelerate machine learning algorithms, whether by the development and application of natively quantum algorithms, or by offering quantum-generated training data that can be used to train a classical AI model.
Our recent announcement about Generative Quantum AI (Gen QAI) spells out our commitment to unlocking the value of the data generated by our H2 quantum computer. This value arises from the world’s leading fidelity and computational power of our System Model H2, making it impossible to exactly simulate on any classical computer, and therefore the data it generates – that we can use to train AI – is inaccessible by any other means. Quantinuum’s Chief Scientist for Algorithms and Innovation, Prof. Harry Buhrman, has likened accessing the first truly quantum-generated training data to the invention of the modern microscope in the seventeenth century, which revealed an entirely new world of tiny organisms thriving unseen within a single drop of water.
Recently, we announced a wide-ranging partnership with NVIDIA. It charts a course to commercial scale applications arising from the partnership between high-performance classical computers, powerful AI systems, and quantum computers that breach the boundaries of what previously could and could not be done. Our President & CEO, Dr. Raj Hazra spoke to CNBC recently about our partnership. Watch the video here.
As we prepare for the next stage of quantum processor development, with the launch of our Helios system in 2025, we’re excited to see how AI can help write more efficient code for quantum computers – and how our quantum processors, the most powerful in the world, can provide a backend for AI computations.
As in any truly symbiotic relationship, the addition of AI to quantum computing equally benefits both sides of the equation.
To read more about Quantinuum and Google DeepMind’s collaboration, please read the scientific paper here.
Few things are more important to the smooth functioning of our digital economies than trustworthy security. From finance to healthcare, from government to defense, quantum computers provide a means of building trust in a secure future.
Quantinuum and its partners JPMorganChase, Oak Ridge National Laboratory, Argonne National Laboratory and the University of Texas used quantum computers to solve a known industry challenge, generating the “random seeds” that are essential for the cryptography behind all types of secure communication. As our partner and collaborator, JPMorganChase explain in this blog post that true randomness is a scarce and valuable commodity.
This year, Quantinuum will introduce a new product based on this development that has long been anticipated, but until now thought to be some years away from reality.
It represents a major milestone for quantum computing that will reshape commercial technology and cybersecurity: Solving a critical industry challenge by successfully generating certifiable randomness.
Building on the extraordinary computational capabilities of Quantinuum’s H2 System – the highest-performing quantum computer in the world – our team has implemented a groundbreaking approach that is ready-made for industrial adoption. Nature today reported the results of a proof of concept with JPMorganChase, Oak Ridge National Laboratory, Argonne National Laboratory, and the University of Texas alongside Quantinuum. It lays out a new quantum path to enhanced security that can provide early benefits for applications in cryptography, fairness, and privacy.
By harnessing the powerful properties of quantum mechanics, we’ve shown how to generate the truly random seeds critical to secure electronic communication, establishing a practical use-case that was unattainable before the fidelity and scalability of the H2 quantum computer made it reliable. So reliable, in fact, that it is now possible to turn this into a commercial product.
Quantinuum will integrate quantum-generated certifiable randomness into our commercial portfolio later this year. Alongside Generative Quantum AI and our upcoming Helios system – capable of tackling problems a trillion times more computationally complex than H2 – Quantinuum is further cementing its leadership in the rapidly-advancing quantum computing industry.
This Matters Because Cybersecurity Matters
Cryptographic security, a bedrock of the modern economy, relies on two essential ingredients: standardized algorithms and reliable sources of randomness – the stronger the better. Non-deterministic physical processes, such as those governed by quantum mechanics, are ideal sources of randomness, offering near-total unpredictability and therefore, the highest cryptographic protection. Google, when it originally announced it had achieved quantum supremacy, speculated on the possibility of using the random circuit sampling (RCS) protocol for the commercial production of certifiable random numbers. RCS has been used ever since to demonstrate the performance of quantum computers, including a milestone achievement in June 2024 by Quantinuum and JPMorganChase, demonstrating their first quantum computer to defy classical simulation. More recently RCS was used again by Google for the launch of its Willow processor.
In today’s announcement, our joint team used the world’s highest-performing quantum and classical computers to generate certified randomness via RCS. The work was based on advanced research by Shih-Han Hung and Scott Aaronson of the University of Texas at Austin, who are co-authors on today’s paper.
Following a string of major advances in 2024 – solving the scaling challenge, breaking new records for reliability in partnership with Microsoft, and unveiling a hardware roadmap, today proves how quantum technology is capable of creating tangible business value beyond what is available with classical supercomputers alone.
What follows is intended as a non-technical explainer of the results in today’s Nature paper.
Certified Randomness: The First Commercial Application for Quantum Computers
For security sensitive applications, classical random number generation is unsuitable because it is not fundamentally random and there is a risk it can be “cracked”. The holy grail is randomness whose source is truly unpredictable, and Nature provides just the solution: quantum mechanics. Randomness is built into the bones of quantum mechanics, where determinism is thrown out the door and outcomes can be true coin flips.
At Quantinuum, we have a strong track record in developing methods for generating certifiable randomness using a quantum computer. In 2021, we introduced Quantum Origin to the market, as a quantum-generated source of entropy targeted at hardening classically-generated encryption keys, using well known quantum technologies that prior to that it had not been possible to use.
In their theory paper, “Certified Randomness from Quantum Supremacy”, Hung and Aaronson ask the question: is it possible to repurpose RCS, and use it to build an application that moves beyond quantum technologies and takes advantage of the power of a quantum computer running quantum circuits?
This was the inspiration for the collaboration team led by JPMorganChase and Quantinuum to draw up plans to execute the proposal using real-world technology. Here’s how it worked:
- The team sent random circuits to Quantinuum’s H2, the world’s highest performing commercially available quantum computer.
- The quantum computer executed each circuit and returned the corresponding sample. The response times were remarkably short, and it could be proven that the circuits could not have been simulated classically within those times, even using the best-known techniques on computing resources greater than those available in the world’s most powerful classical supercomputer.
- The randomness of the returned sample was mathematically certified using Frontier, the world’s most powerful classical supercomputer, establishing it achieved a “passing threshold” on a measure known as the “cross-entropy benchmark”. The better your quantum computer, the higher you can set the “passing threshold”. When the threshold is sufficiently high, "spoofing" the cross-entropy benchmark using only classical methods becomes inefficient.
- Therefore, if the samples are returned quickly and meet the high threshold, the team could be confident that they were generated by a quantum computer – and thus be truly random.
This confirmed that Quantinuum’s quantum computer is not only incapable of being matched by classical computers but can also be used reliably to produce a certifiably random seed from a quantum computer without the need to build your own device, or even trust the device you are accessing.
Looking ahead
The use of randomness in critical cybersecurity environments will gravitate towards quantum resources, as the security demands of end users grows in the face of ongoing cyber threats.
The era of quantum utility offers the promise of radical new approaches to solving substantial and hard problems for businesses and governments.
Quantinuum’s H2 has now demonstrated practical value for cybersecurity vendors and customers alike, where non-deterministic sources of encryption may in time be overtaken by nature’s own source of randomness.
In 2025, we will launch our Helios device, capable of supporting at least 50 high-fidelity logical qubits – and further extending our lead in the quantum computing sector. We thus continue our track record of disclosing our objectives and then meeting or surpassing them. This commitment is essential, as it generates faith and conviction among our partners and collaborators, that empirical results such as those reported today can lead to successful commercial applications.
Helios, which is already in its late testing phase, ahead of being commercially available later this year, brings higher fidelity, greater scale, and greater reliability. It promises to bring a wider set of hybrid quantum-supercomputing opportunities to our customers – making quantum computing more valuable and more accessible than ever before.
And in 2025 we look forward to adding yet another product, building out our cybersecurity portfolio with a quantum source of certifiably random seeds for a wide range of customers who require this foundational element to protect their businesses and organizations.