“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.
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:
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].
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.
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.
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.
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.
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.
Last year, we joined forces with RIKEN, Japan's largest comprehensive research institution, to install our hardware at RIKEN’s campus in Wako, Saitama. This deployment is part of RIKEN’s project to build a quantum-HPC hybrid platform consisting of high-performance computing systems, such as the supercomputer Fugaku and Quantinuum Systems.
Today, a paper published in Physical Review Research marks the first of many breakthroughs coming from this international supercomputing partnership. The team from RIKEN and Quantinuum joined up with researchers from Keio University to show that quantum information can be delocalized (scrambled) using a quantum circuit modeled after periodically driven systems.
"Scrambling" of quantum information happens in many quantum systems, from those found in complex materials to black holes. Understanding information scrambling will help researchers better understand things like thermalization and chaos, both of which have wide reaching implications.
To visualize scrambling, imagine a set of particles (say bits in a memory), where one particle holds specific information that you want to know. As time marches on, the quantum information will spread out across the other bits, making it harder and harder to recover the original information from local (few-bit) measurements.
While many classical techniques exist for studying complex scrambling dynamics, quantum computing has been known as a promising tool for these types of studies, due to its inherently quantum nature and ease with implementing quantum elements like entanglement. The joint team proved that to be true with their latest result, which shows that not only can scrambling states be generated on a quantum computer, but that they behave as expected and are ripe for further study.
Thanks to this new understanding, we now know that the preparation, verification, and application of a scrambling state, a key quantum information state, can be consistently realized using currently available quantum computers. Read the paper here, and read more about our partnership with RIKEN here.
In our increasingly connected, data-driven world, cybersecurity threats are more frequent and sophisticated than ever. To safeguard modern life, government and business leaders are turning to quantum randomness.
The term to know: quantum random number generators (QRNGs).
QRNGs exploit quantum mechanics to generate truly random numbers, providing the highest level of cryptographic security. This supports, among many things:
Quantum technologies, including QRNGs, could protect up to $1 trillion in digital assets annually, according to a recent report by the World Economic Forum and Accenture.
The World Economic Forum report identifies five industry groups where QRNGs offer high business value and clear commercialization potential within the next few years. Those include:
In line with these trends, recent research by The Quantum Insider projects the quantum security market will grow from approximately $0.7 billion today to $10 billion by 2030.
Quantum randomness is already being deployed commercially:
Recognizing the value of QRNGs, the financial services sector is accelerating its path to commercialization.
On the basis of the latter achievement, we aim to broaden our cybersecurity portfolio with the addition of a certified randomness product in 2025.
The National Institute of Standards and Technology (NIST) defines the cryptographic regulations used in the U.S. and other countries.
This week, we announced Quantum Origin received NIST SP 800-90B Entropy Source validation, marking the first software QRNG approved for use in regulated industries.
This means Quantum Origin is now available for high-security cryptographic systems and integrates seamlessly with NIST-approved solutions without requiring recertification.
The NIST validation, combined with our peer-reviewed papers, further establishes Quantum Origin as the leading QRNG on the market.
--
It is paramount for governments, commercial enterprises, and critical infrastructure to stay ahead of evolving cybersecurity threats to maintain societal and economic security.
Quantinuum delivers the highest quality quantum randomness, enabling our customers to confront the most advanced cybersecurity challenges present today.
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.
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”.
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.