We’d like to hear from parents and teachers who’re introducing AI to younger children and find out how they’re instructing them how to use it
Across the world since the release of ChatGPT, we have seen younger and younger people make use of artificial intelligence chatbots and image generators for their schoolwork and their lives at home. First, students in universities adopted AI, then teenagers in secondary schools, often with their parents unaware of it. Now, we’d like to know how the youngest among us are learning about AI.
AI companies typically prohibit children under 13 from using their products in terms of service agreements. Some parents and teachers, however, are introducing ChatGPT and other AI to their children and their students in the hopes that an early familiarity with the technology will prepare them to succeed in the future.
Unexpected behaviour at phase transitions between classical and quantum magnetism has been observed in different quantum simulators operated by two independent groups. One investigation was led by researchers at Harvard University and used Rydberg atom as quantum bits (qubits). The other study was led by scientists at Google Research and involved superconducting qubits. Both projects revealed unexpected deviations from the canonical mechanisms of magnetic freezing, with unexpected oscillations near the phase transition.
A classical magnetic material can be understood as a fluid mixture of magnetic domains that are oriented in opposite directions, with the domain walls in constant motion. As a strengthening magnetic field is applied to the system, the energy associated with a domain wall increases, so the magnetic domains themselves become larger and less mobile. At some point, when the magnetism becomes sufficiently strong, a quantum phase transition occurs, causing the magnetism of the material to become fixed and crystalline: “A good analogy is like water freezing,” says Mikhail Lukin of Harvard University.
The traditional quantitative model for these transitions is the Kibble–Zurek mechanism, which was first formulated to describe cosmological phase transitions in the early universe. It predicts that the dynamics of a system begin to “freeze” when the system gets so close to the transition point that the domains crystallize more quickly than they can come to equilibrium.
“There are some very good theories of various types of quantum phase transitions that have been developed,” says Lukin, “but typically these theories make some approximations. In many cases they’re fantastic approximations that allow you to get very good results, but they make some assumptions which may or may not be correct.”
Highly reconfigurable platform
In their work, Lukin and colleagues utilized a highly reconfigurable platform using Rydberg atom qubits. The system was pioneered by Lukin and others in 2016 to study a specific type of magnetic quantum phase transition in detail. They used a laser to simulate the effect of a magnetic field on the Rydberg atoms, and adjusted the laser frequency to tune the field strength.
The researchers found that, rather than simply becoming progressively larger and less mobile as the field strength increased (a phenomenon called coarsening), the domain sizes underwent unexpected oscillations around the phase transition.
“We were really quite puzzled,” says Lukin. “Eventually we figured out that this oscillation is a sign of a special type of excitation mode similar to the Higgs mode in high-energy physics. This is something we did not anticipate…That’s an example where doing quantum simulations on quantum devices really can lead to new discoveries.”
Meanwhile, the Google-led study used a new approach to quantum simulation with superconducting qubits. Such qubits have proved extremely successful and scalable because they use solid-state technology – and they are used in most of the world’s leading commercial quantum computers such as IBM’s Osprey and Google’s own Willow chips. Much of the previous work using such chips, however, has focused on sequential “digital” quantum logic in which one set of gates is activated only after the previous set has concluded. The long times needed for such calculations allows the effects of noise to accumulate, resulting in computational errors.
Hybrid approach
In the new work, the Google team developed a hybrid analogue–digital approach in which a digital universal quantum gate set was used to prepare well-defined input qubit states. They then switched the processor to analogue mode, using capacitive couplers to tune the interactions between the qubits. In this mode, all the qubits were allowed to operate on each other simultaneously, without the quantum logic being shoehorned into a linear set of gate operations. Finally, the researchers characterized the output by switching back to digital mode.
The researchers used a 69-qubit superconducting system to simulate a similar, but non-identical, magnetic quantum phase transition to that studied by Lukin’s group. They were also puzzled by similar unexpected behaviour in their system. The groups’ subsequently became aware of each other’s work, as Google Research’s Trond Anderson explains: “It’s very exciting to see consistent observations from the Lukin group. This not only provides supporting evidence, but also demonstrates that the phenomenon appears in several contexts, making it extra important to understand”.
Both groups are now seeking to push their research deeper into the exploration of complex many-body quantum physics. The Google group estimates that, to conduct its simulations of the highly entangled quantum states involved with the same level of experimental fidelity would take the US Department of Energy’s Frontier supercomputer – one of the world’s most powerful – more than a million years. The researchers now want to look at problems that are completely intractable classically, such as magnetic frustration. “The analogue–digital approach really combines the best of both worlds, and we’re very excited about this as a new promising direction towards making discoveries in systems that are too complex for classical computers,” says Anderson.
The Harvard researchers are also looking to push their system to study more and more complex quantum systems. “There are many interesting processes where dynamics – especially across a quantum phase transition – remains poorly understood,” says Lukin. “And it ranges from the science of complex quantum materials to systems in high-energy physics such as lattice gauge theories, which are notorious for being hard to simulate classically to the point where people literally give up…We want to apply these kinds of simulators to real open quantum problems and really use them to study the dynamics of these systems.”
The research is described in side-by-side papers in Nature. The Google paper is here and the Harvard paper here.
Watch this short video filmed at the APS March Meeting in 2024, where Mark Elo, chief marketing officer of Tabor Quantum Solutions, introduces the Echo-5Q, which he explains is an industry collaboration between FormFactor and Tabor Quantum Solutions, using the QuantWare quantum processing unit (QPU).
Elo points out that it is an out-of-the-box solution, allowing customers to order a full-stack system, including the software, refrigeration, control electronics and the actual QPU. With the Echo-5, it gets delivered and installed, so that the customer can start doing quantum measurements immediately. He explains that the Echo-5Q is designed at a price and feature point that increases the accessibility for on-site quantum computing.
Brandon Boiko, senior applications engineer with FormFactor, describes the how FormFactor developed the dilution refrigeration technology that the qubits get installed into. Boiko explains that the product has been designed to reduce the cost of entry into the quantum field – made accessible through FormFactor’s test-and- measurement programme, which allows people to bring their samples on site to take measurements.
Alessandro Bruno is founder and CEO of QuantWare, which provides the quantum processor for the Echo-5Q, the part that sits at the milli Kelvin stage of the dilution refrigerator, and that hosts five qubits. Bruno hopes that the Echo-5Q will democratize access to quantum devices – for education, academic research and start-ups.
Errors caused by interactions with the environment – noise – are the Achilles heel of every quantum computer, and correcting them has been called a “defining challenge” for the technology. These two teams, working with very different quantum systems, took significant steps towards overcoming this challenge. In doing so, they made it far more likely that quantum computers will become practical problem-solving machines, not just noisy, intermediate-scale tools for scientific research.
Quantum error correction works by distributing one quantum bit of information – called a logical qubit – across several different physical qubits such as superconducting circuits or trapped atoms. While each physical qubit is noisy, they work together to preserve the quantum state of the logical qubit – at least for long enough to do a computation.
Formidable task
Error correction should become more effective as the number of physical qubits in a logical qubit increases. However, integrating large numbers of physical qubits to create a processor with multiple logical qubits is a formidable task. Furthermore, adding more physical qubits to a logical qubit also adds more noise – and it is not clear whether making logical qubits bigger would make them significantly better. This year’s winners of our Breakthrough of the Year have made significant progress in addressing these issues.
The team led by Lukin and Bluvstein created a quantum processor with 48 logical qubits that can execute algorithms while correcting errors in real time. At the heart of their processor are arrays of neutral atoms. These are grids of ultracold rubidium atoms trapped by optical tweezers. These atoms can be put into highly excited Rydberg states, which enables the atoms to act as physical qubits that can exchange quantum information.
What is more, the atoms can be moved about within an array to entangle them with other atoms. According to Bluvstein, moving groups of atoms around the processor was critical for their success at addressing a major challenge in using logical qubits: how to get logical qubits to interact with each other to perform quantum operations. He describes the system as a “living organism that changes during a computation”.
Their processor used about 300 physical qubits to create up to 48 logical qubits, which were used to perform logical operations. In contrast, similar attempts using superconducting or trapped-ion qubits have only managed to perform logical operations using 1–3 logical qubits.
Willow quantum processor
Meanwhile, the team led by Hartmut Neven made a significant advance in how physical qubits can be combined to create a logical qubit. Using Google’s new Willow quantum processor – which offers up to 105 superconducting physical qubits – they showed that the noise in their logical qubit remained below a maximum threshold as they increased the number of qubits. This means that the logical error rate is suppressed exponentially as the number of physical qubits per logical qubit is increased.
Neven told Physics World that the Google system is “the most convincing prototype of a logical qubit built today”. He said that that Google is on track to develop a quantum processor with 100 or even 1000 logical qubits by 2030. He says that a 1000 logical qubit device could do useful calculations for the development of new drugs or new materials for batteries.
Bluvstein, Lukin and colleagues are already exploring how their processor could be used to study an effect called quantum scrambling. This could shed light on properties of black holes and even provide important clues about the nature of quantum gravity.
You can listen to Neven talk about his team’s research in this podcast. Bluvstein and Lukin talk about their group’s work in this podcast.
The Breakthrough of the Year was chosen by the Physics World editorial team. We looked back at all the scientific discoveries we have reported on since 1 January and picked the most important. In addition to being reported in Physics World in 2024, the breakthrough must meet the following criteria:
Significant advance in knowledge or understanding
Importance of work for scientific progress and/or development of real-world applications
Of general interest to Physics World readers
Before we picked our winners, we released the Physics World Top 10 Breakthroughs for 2024, which served as our shortlist. The other nine breakthroughs are listed below in no particular order.
Achieving optical transparency First author Zihao Ou holds a vial of the common yellow food dye tartrazine in solution. By applying a mixture of water and tartrazine, Ou and colleagues made the skin on the skulls and abdomens of live mice transparent. (Courtesy: University of Texas at Dallas)
To a team of researchers at Stanford University in the US for developing a method to make the skin of live mice temporarily transparent. One of the challenges of imaging biological tissue using optical techniques is that tissue scatters light, which makes it opaque. The team, led by Zihao Ou (now at The University of Texas at Dallas), Mark Brongersma and Guosong Hong, found that the common yellow food dye tartrazine strongly absorbs near-ultraviolet and blue light and can help make biological tissue transparent. Applying the dye onto the abdomen, scalp and hindlimbs of live mice enabled the researchers to see internal organs, such as the liver, small intestine and bladder, through the skin without requiring any surgery. They could also visualize blood flow in the rodents’ brains and the fine structure of muscle sarcomere fibres in their hind limbs. The effect can be reversed by simply rinsing off the dye. This “optical clearing” technique has so far only been conducted on animals. But if extended to humans, it could help make some types of invasive biopsies a thing of the past.
To the AEgIS collaboration at CERN, and Kosuke Yoshioka and colleagues at the University of Tokyo, for independently demonstrating laser cooling of positronium. Positronium, an atom-like bound state of an electron and a positron, is created in the lab to allow physicists to study antimatter. Currently, it is created in “warm” clouds in which the atoms have a large distribution of velocities, making precision spectroscopy difficult. Cooling positronium to low temperatures could open up novel ways to study the properties of antimatter. It also enables researchers to produce one to two orders of magnitude more antihydrogen – an antiatom comprising a positron and an antiproton that’s of great interest to physicists. The research also paves the way to use positronium to test current aspects of the Standard Model of particle physics, such as quantum electrodynamics, which predicts specific spectral lines, and to probe the effects of gravity on antimatter.
To Roman Bauer at the University of Surrey, UK, Marco Durante from the GSI Helmholtz Centre for Heavy Ion Research, Germany, and Nicolò Cogno from GSI and Massachusetts General Hospital/Harvard Medical School, US, for creating a computational model that could improve radiotherapy outcomes for patients with lung cancer. Radiotherapy is an effective treatment for lung cancer but can harm healthy tissue. To minimize radiation damage and help personalize treatment, the team combined a model of lung tissue with a Monte Carlo simulator to simulate irradiation of alveoli (the tiny air sacs within the lungs) at microscopic and nanoscopic scales. Based on the radiation dose delivered to each cell and its distribution, the model predicts whether each cell will live or die, and determines the severity of radiation damage hours, days, months or even years after treatment. Importantly, the researchers found that their model delivered results that matched experimental observations from various labs and hospitals, suggesting that it could, in principle, be used within a clinical setting.
Epigraphene on a chip: the team’s graphene device was grown on a silicon carbide substrate. (Courtesy: Georgia Institute of Technology)
To Walter de Heer, Lei Ma and colleagues at Tianjin University and the Georgia Institute of Technology, and independently to Marcelo Lozada-Hidalgo of the University of Manchester and a multinational team of colleagues, for creating a functional semiconductor made from graphene, and for using graphene to make a switch that supports both memory and logic functions, respectively. The Manchester-led team’s achievement was to harness graphene’s ability to conduct both protons and electrons in a device that performs logic operations with a proton current while simultaneously encoding a bit of memory with an electron current. These functions are normally performed by separate circuit elements, which increases data transfer times and power consumption. Conversely, de Heer, Ma and colleagues engineered a form of graphene that does not conduct as easily. Their new “epigraphene” has a bandgap that, like silicon, could allow it to be made into a transistor, but with favourable properties that silicon lacks, such as high thermal conductivity.
To David Moore, Jiaxiang Wang and colleagues at Yale University, US, for detecting the nuclear decay of individual helium nuclei by embedding radioactive lead-212 atoms in a micron-sized silica sphere and measuring the sphere’s recoil as nuclei escape from it. Their technique relies on the conservation of momentum, and it can gauge forces as small as 10-20 N and accelerations as tiny as 10-7 g, where g is the local acceleration due to the Earth’s gravitational pull. The researchers hope that a similar technique may one day be used to detect neutrinos, which are much less massive than helium nuclei but are likewise emitted as decay products in certain nuclear reactions.
To Andrew Denniston at the Massachusetts Institute of Technology in the US, Tomáš Ježo at Germany’s University of Münster and an international team for being the first to unify two distinct descriptions of atomic nuclei. They have combined the particle physics perspective – where nuclei comprise quarks and gluons – with the traditional nuclear physics view that treats nuclei as collections of interacting nucleons (protons and neutrons). The team has provided fresh insights into short-range correlated nucleon pairs – which are fleeting interactions where two nucleons come exceptionally close and engage in strong interactions for mere femtoseconds. The model was tested and refined using experimental data from scattering experiments involving 19 different nuclei with very different masses (from helium-3 to lead-208). The work represents a major step forward in our understanding of nuclear structure and strong interactions.
To Jelena Vučković, Joshua Yang, Kasper Van Gasse, Daniil Lukin, and colleagues at Stanford University in the US for developing a compact, integrated titanium:sapphire laser that needs only a simple green LED as a pump source. They have reduced the cost and footprint of a titanium:sapphire laser by three orders of magnitude and the power consumption by two. Traditional titanium:sapphire lasers have to be pumped with high-powered lasers – and therefore cost in excess of $100,000. In contrast, the team was able to pump its device using a $37 green laser diode. The researchers also achieved two things that had not been possible before with a titanium:sapphire laser. They were able to adjust the wavelength of the laser light and they were able to create a titanium:sapphire laser amplifier. Their device represents a key step towards the democratization of a laser type that plays important roles in scientific research and industry.
To two related teams for their clever use of entangled photons in imaging. Both groups include Chloé Vernière and Hugo Defienne of Sorbonne University in France, who as duo used quantum entanglement to encode an image into a beam of light. The impressive thing is that the image is only visible to an observer using a single-photon sensitive camera – otherwise the image is hidden from view. The technique could be used to create optical systems with reduced sensitivity to scattering. This could be useful for imaging biological tissues and long-range optical communications. In separate work, Vernière and Defienne teamed up with Patrick Cameron at the UK’s University of Glasgow and others to use entangled photons to enhance adaptive optical imaging. The team showed that the technique can be used to produce higher-resolution images than conventional bright-field microscopy. Looking to the future, this adaptive optics technique could play a major role in the development of quantum microscopes.
To the China National Space Administration for the first-ever retrieval of material from the Moon’s far side, confirming China as one of the world’s leading space nations. Landing on the lunar far side – which always faces away from Earth – is difficult due to its distance and terrain of giant craters with few flat surfaces. At the same time, scientists are interested in the unexplored far side and why it looks so different from the near side. The Chang’e-6 mission was launched on 3 May consisting of four parts: an ascender, lander, returner and orbiter. The ascender and lander successfully touched down on 1 June in the Apollo basin, which lies in the north-eastern side of the South Pole-Aitken Basin. The lander used its robotic scoop and drill to obtain about 1.9 kg of materials within 48 h. The ascender then lifted off from the top of the lander and docked with the returner-orbiter before the returner headed back to Earth, landing in Inner Mongolia on 25 June. In November, scientists released the first results from the mission finding that fragments of basalt – a type of volcanic rock – date back to 2.8 billion years ago, indicating that the lunar far side was volcanically active at that time. Further scientific discoveries can be expected in the coming months and years ahead as scientists analyze more fragments.
Physics World‘s coverage of the Breakthrough of the Year is supported by Reports on Progress in Physics, which offers unparalleled visibility for your ground-breaking research.
In this episode of the Physics World Weekly podcast, Bluvstein and Lukin explain the crucial role that error correction is playing in the development of practical quantum computers. They also describe how atoms are moved around their quantum processor and why this coordinated motion allowed them to create logical qubits and use those qubits to perform quantum computations.
Physics World‘s coverage of the Breakthrough of the Year is supported by Reports on Progress in Physics, which offers unparalleled visibility for your ground-breaking research.
In this episode of the Physics World Weekly podcast, Neven talks about Google’s new Willow quantum processor, which integrates 105 superconducting physical qubits. He also explains how his team used these qubits to create logical qubits with error rates that dropped exponentially with the number of physical qubits used. He also outlines Googles ambitious plan to create a processor with 100, or even 1000, logical qubits by 2030.
Physics World‘s coverage of the Breakthrough of the Year is supported by Reports on Progress in Physics, which offers unparalleled visibility for your ground-breaking research.
Researchers at Google Quantum AI and collaborators have developed a quantum processor with error rates that get progressively smaller as the number of quantum bits (qubits) grows larger. This achievement is a milestone for quantum error correction, as it could, in principle, lead to an unlimited increase in qubit quality, and ultimately to an unlimited increase in the length and complexity of the algorithms that quantum computers can run.
Noise is an inherent feature of all physical systems, including computers. The bits in classical computers are protected from this noise by redundancy: some of the data is held in more than one place, so if an error occurs, it is easily identified and remedied. However, the no-cloning theorem of quantum mechanics dictates that once a quantum state is measured – a first step towards copying it – it is destroyed. “For a little bit, people were surprised that quantum error correction could exist at all,” observes Michael Newman, a staff research scientist at Google Quantum AI.
Beginning in the mid-1990s, however, information theorists showed that this barrier is not insurmountable, and several codes for correcting qubit errors were developed. The principle underlying all of them is that multiple physical qubits (such as individual atomic energy levels or states in superconducting circuits) can be networked to create a single logical qubit that collectively holds the quantum information. It is then possible to use “measure” qubits to determine whether an error occurred on one of the “data” qubits without affecting the state of the latter.
“In quantum error correction, we basically track the state,” Newman explains. “We say ‘Okay, what errors are happening?’ We figure that out on the fly, and then when we do a measurement of the logical information – which gives us our answer – we can reinterpret our measurement according to our understanding of what errors have happened.”
Keeping error rates low
In principle, this procedure makes it possible for infinitely stable qubits to perform indefinitely long calculations – but only if error rates remain low enough. The problem is that each additional physical qubit introduces a fresh source of error. Increasing the number of physical qubits in each logical qubit is therefore a double-edged sword, and the logical qubit’s continued stability depends on several factors. These include the ability of the quantum processor’s (classical) software to detect and interpret errors; the specific error-correction code used; and, importantly, the fidelity of the physical qubits themselves.
In 2023, Newman and colleagues at Google Quantum AI showed that an error-correction code called the surface code (which Newman describes as having “one of the highest error-suppression factors of any quantum code”) made it just about possible to “win” at error correction by adding more physical qubits to the system. Specifically, they showed that a distance-5 array logical qubit made from 49 superconducting transmon qubits had a slightly lower error rate than a distance-3 array qubit made from 17 such qubits. But the margin was slim. “We knew that…this wouldn’t persist,” Newman says.
“Convincing, exponential error suppression”
In the latest work, which is published in Nature, a Google Quantum AI team led by Hartmut Neven unveil a new superconducting processor called Willow with several improvements over the previous Sycamore chip. These include gates (the building blocks of logical operations) that retain their “quantumness” five times longer and a Google Deepmind-developed machine learning algorithm that interprets errors in real time. When the team used this new tech to create nine surface code distance-3 arrays, four distance-5 arrays and one 101-qubit distance-7 array on their 105-qubit processor, the error rate was suppressed by a factor of 2.4 as additional qubits were added.
How it works: Surface code logical qubits for processors of increasing size. Each larger processor can correct more errors than its predecessor. The encoded quantum state is stored on the array of data qubits (gold). Measure qubits (red, cyan, blue) check for errors on the neighbouring data qubits. (Courtesy: Google Quantum AI)
“This is the first time we have seen convincing, exponential error suppression in the logical qubits as we increase the number of physical qubits,” says Newman. “That’s something people have been trying to do for about 30 years.”
With gates that remain stable for hours on end, quantum computers should be able to run the large, complex algorithms people have always hoped for. “We still have a long way to go, we still need to do this at scale,” Newman acknowledges. “But the first time we pushed the button on this Willow chip and I saw the lattice getting larger and larger and the error rate going down and down, I thought ‘Wow! Quantum error correction is really going to work…Quantum computing is really going to work!’”
Mikhail Lukin, a physicist at Harvard University in the US who also works on quantum error correction, calls the Google Quantum AI result “a very important step forward in the field”. While Lukin’s own group previously demonstrated improved quantum logic operations between multiple error-corrected atomic qubits, he notes that the present work showed better logical qubit performance after multiple cycles of error correction. “In practice, you’d like to see both of these things come together to enable deep, complex quantum circuits,” he says. “It’s very early, there are a lot of challenges remaining, but it’s clear that – in different platforms and moving in different directions – the fundamental principles of error correction have now been demonstrated. It’s very exciting.”
Connexion
