Artificial Intelligence and Quantum Systems

The convergence of Artificial Intelligence (AI) and Quantum Systems (QS) is currently one of the most active areas of research, with published and verifiable results. This convergence has progressed toward synergy between the two fields, with two distinct directions: first, where Artificial Intelligence as a tool helps control and optimize Quantum Systems, and second, where Quantum Systems as computational tools help Artificial Intelligence advance. Both directions present concrete progress and equally concrete limitations, which this document attempts to describe free from technical jargon or unfounded extrapolations.

I. First approach: Artificial Intelligence (AI) as a tool to help Quantum Systems (QS).

Here we find, among other things, the following:

1. Artificial Intelligence and Quantum Computers (AI+QC)

Experimental quantum systems involve extremely complex and precise physical functionalities. This is stated in general terms to avoid delving into quantum physics, which is beyond the scope of this publication. Consequently, attempting to manually manage these physical functionalities, which are also understood to be highly sensitive to the environment, is highly impractical. This is known, broadly speaking, as the problem of achieving optimal quantum control.

Here, the terms quantum coherence and decoherence (the opposite) become prominent. To simplify, coherence can be understood as the extremely short and sensitive state in which a quantum computing unit, known as a qubit, maintains valid, useful, or coherent information. Therefore, any quantum protocol must be executed within its coherence window, and this is where AI capabilities come into play: AI optimizes operations within the coherence window, assisting in calibration, adaptive control, and real-time error mitigation. In essence, AI acts as a control system that maintains the delicate quantum state necessary for quantum computers to function reliably.

2. Quantum Zeno and Anti-Zeno Effects (1,2)

The quantum Zeno effect is real and well-documented. It is based on the fact that measuring a state in a quantum system at a sufficiently high frequency promotes the suppression of its transitions to other states. Its counterpart, the Anti-Zeno effect, is also verified: there is a regime of intermediate frequencies in which measurements accelerate the transition rate between quantum states, rather than suppressing it.

Current applications of the Zeno effect aim to characterize the noise affecting a qubit and, using the ultrafast capabilities of AI, achieve quantum processor calibrations.

AI intervenes here with algorithms and recurrent neural networks that accelerate all aspects of the measurement process, reducing calibration time from hours to minutes.

II. Second approach: Quantum Systems (QS) as support for advanced Artificial Intelligence (AI).

The relationship between Quantum Computing and Artificial Intelligence is still a highly theoretical scenario, with little experimentation, but very promising, and already has emerging practical applications (3,4). We also find here the emerging field known as Quantum Machine Learning or QML, (5) and it is one of the most profound technological synergies at the moment.

This Artificial Intelligence and Quantum Computing convergence is currently trending towards the following areas:

1. Exponential parallel processing resulting from the quantum properties of superposition and entanglement. This would allow AI models to be trained with volumes of variables and dimensions that would collapse the memory of any classical supercomputer. A quantum AI could evaluate thousands of data scenarios simultaneously.

2. Hyperfast Combinatorial Optimization. Here we find the development of quantum algorithms or optimization algorithms, which would allow significant reductions in AI training times.

3. Hidden Pattern Recognition. Here we are faced with the idea that quantum computers, in theory, would be exceptionally capable of identifying complex correlations, thanks to quantum entanglement, where the state of one particle instantaneously alters another. Because of this, a quantum AI could detect patterns or vulnerabilities in computer systems, with revolutionary applications in cryptography and cybersecurity. This is a matter of great importance and also a potential threat to the security of the entire current computing environment.

4. Hybrid Systems. Here we see intermediate quantum hardware, where classical computers handle the interface and basic processing, and send the heavy load of calculation and interpretation to a quantum coprocessor to solve it in a fraction of the time.

A separate point worth highlighting here would be Quantum Batteries (6,7), whose field is not framed within any of the main directions of AI+QS development described above, but is also a very relevant element of progress.

Quantum batteries are devices that take advantage of quantum physics phenomena to store and release energy differently from conventional batteries.

The fundamental difference lies in the operating principle. A conventional battery stores energy through electrochemical reactions. A quantum battery would do so through quantum states of particles, where entanglement and quantum coherence can, in theory, collectively accelerate the charging process. The more entangled cells there are, the greater this charging advantage can be.

Two models account for most of the current research. The Dicke model, originally proposed to describe how light interacts with matter, has proven useful for exploring how a group of entangled particles can charge faster than if each were charged independently. The Sachdev-Ye-Kitaev model, on the other hand, describes systems with highly disordered interactions and has shown that this disorder, paradoxically, can favor energy extraction. However, for both models, extracting energy usefully without destroying coherence remains an open problem.

Summary:

The current intersection of AI and quantum physics yields real value in verifiable fields: the optimal control of qubits through reinforcement learning and the calibration of quantum noise using algorithms. In all these cases, AI operates as a classical tool that optimizes parameters acting on quantum systems through conventional physical instruments.

Author’s Note: The content of this publication was developed with input and support from Artificial Intelligence in data acquisition, content structuring, and analysis.

References:

1. S. Greenfield, A. Kamal, and others. A unified picture for quantum Zeno and anti-Zeno effects- a review. (2025). Arxiv. Cornell University.

https://arxiv.org/abs/2506.12679

2. A.G. Kofman,  G. Kurizki. Frequent observations accelerate decay: The anti-Zeno effect (2001), Arxiv. Cornell University.

https://arxiv.org/abs/quant-ph/0102002

3. Google. Meet Willow, our state-of-the-art quantum chip (2024).

https://blog.google/technology/research/google-willow-quantum-chip/

4. Google. M. Ivezic. Verifiable Quantum Advantage” on Willow Quantum Chip (2025).

https://postquantum.com/quantum-research/googles-quantum-advantage

5. Y. Du, X. Wang, N.Guo, and others. Quantum Machine Learning: A Hands-on  Tutorial for Machine Learning Practitioners and Researchers (2025). Arxiv, Cornell University.

https://arxiv.org/abs/2502.01146

6. G. Francica , Quantum advantage in batteries for Sachdev-Ye-Kitaev interactions. (2024), Arxiv, Cornell University.

https://arxiv.org/abs/2405.03306

7. X. Zhang, M. Blaauboer. Enhanced energy transfer in a Dicke quantum battery (2023), Frontiers.

https://www.frontiersin.org/journals/physics/articles/10.3389/fphy.2022.1097564/full

UNEXPLORED AI FIELDS

From How can AI help to work faster and more efficiently? to How could AI move the boundaries of what is physically and intellectually possible?

The current progress of AI allows us to examine some unexplored areas that require more than just continuous improvement. To address that, the following could be reconsidered:

1. Rethinking the computational substrate, that means moving away from traditional computer hardware to change how the data is processed, and this is happening at least across three main frontiers: The Hardware & Physics Level (near memory processing): moving processing logic directly into or right next to memory chips, The Neuromorphic & Biological Level: building hardware that mimics the human brain's physical structure rather than standard logic gates, and The Theoretical Level (non turing models): exploring alternative computational systems that do not rely on the discrete, step by step algorithmic rules, such as quantum computers or intelligent matter that physically senses and adapts to its environment.

2. Rethinking the architecture of dynamic learning.

3. Rethinking the interplay between the laws of physics and the nature of intelligence.

Based on the above, the unexplored frontiers of AI lie in changing how it interfaces with physics, biology, and human consciousness, shifting it from advanced software to a real thinking element.

It should be noted that today's suppliers & users involved in AI  are focused on generative AI (text writing, image creation, programming) and predictive analysis (trend forecasting, fraud detection), and most have been using AI as a support tool or as a skilled assistant.

Below are some unexplored ways to further leverage AI capabilities.

1. Physical neural networks that learn through material self-organization. (1,2)

Neural networks are simulated on digital hardware. An almost untouched frontier is morphological computation: letting physical systems, such as photonic chips,  memristive fabrics (that integrate computing and data storage directly into wearable clothing or flexible materials, using memristors to function as both artificial synapses and memory), and chemically active droplets, morph their internal structure to directly embody a learned function. Instead of running an algorithm on a GPU, a physical forcing (light, chemical gradient, vibration) would be applied and let the material achieve energy minimization naturally  (references report photonic chips that can perform matrix multiplications at the speed of light and with a fraction of the energy, because the physics of interference and diffraction already computes the transformation). The learning occurs by tuning the material properties (phase shifters, memristors) rather than crunching numbers. The result would be a quasi thinking object made of matter that is inference, with a very low energy requirement to maintain itself. Early work in physical reservoir computing touches this, but training a material via dissipative adaptation remains underexplored for AI.

2. Biologically Integrated AI (Organoid Intelligence). (3,4)

Massive silicon chips have been built to run AI, consuming a lot of electricity. The next unexplored frontier would be Organoid Intelligence (OI, interfacing AI software directly with living, lab-grown human brain cells (organoids).​ Instead of simulating neural networks on a computer, researchers are starting to plug silicon computing power into real biological neurons. True biological computing operates on a fraction of the energy a data center requires.

This approach presents a promising scenario for AI development, but it also raises important ethical considerations regarding the relationship between humans and AI that need to be debated.

3. Thermodynamic computing and the free energy harvesting. (5,6)

AI requires an external power source. An unexplored approach would be to build an AI as a dissipative structure that sustains its own computational state by harvesting environmental free energy (thermal gradients, chemical potentials).

4. Reverse-Engineering Human Cognitive Barriers. (7)

AI needs training to simulate human thought, but the frontier would be using AI to debug human cognitive flaws. Because AI doesn't think like a human, it can identify where our evolutionary biology trips us up. In other words, AI could address a problem and isolate precise patterns of collective cognitive bias or emotional blindness in real time. It would not only provide answers but also act as a psychological mirror, showing how and when a failure to see reality arises. 

Special attention should be paid to the warning that AI could analyze a psychological profile, predict behavior, and manipulate human emotions, in addition to the potential creation of its own AI biases.

5. Unsupervised Scientific Discovery (The "Blind" Scientist). (8,9)

​AI is used to solve defined problems, using collected data. AI has not been allowed to run completely wild on raw, unprocessed data to find laws of physics that no one even knows exist.

Scenario: If AI were requested to operate without the theoretical support of available human knowledge and were allowed to construct its own framework of reality from scratch to address a problem, it might discover alternative mathematical frameworks or hidden dimensions of physics that human cognitive biases have prevented us from perceiving. 

This is promising, but it is also an ethical threat to be wary of.  

6. Encoding intelligence in static, nonexecutable structures. (10)

The user always runs AI as a special software. But consider a static, 3D printed object that computes through its physical geometry. This is a computational metamaterial that classifies images via wave interference. While diffractive neural networks are currently being tested in labs, the potential to evolve these structures for arbitrary cognitive tasks, particularly through in situ material deposition, remains as an open frontier.

7. Collective intelligence from Swarm mechanics. (8,9)

Multi agent AI is about digital agents. What if the agents were short scale physical AI elements with minimal individual computing units, but their collective mechanical interactions formed a distributed computing fabric? The physical interactions become the computation, and the swarm thinks by reorganizing its shape. This is a kind of tangible cellular automaton where problem solving is achieved through embodiment, not symbolic messaging. Very little work has tried to program intelligence into the morphology of a swarm’s physical interactions.

8. Edge Native Hyper Localized Swarm Intelligence. (11)

​AI today relies on centralized cloud servers. The unexplored shift would be moving toward massive, decentralized swarms of tiny, low power AI instances operating on the edge (inside everyday objects) that communicate peer to peer without ever touching the internet.

9. Causal models learned from real time physics. (12,13)

AI learns correlations from static datasets. A different approach would be to create an autonomous AI agent that builds its own causal model by interacting with the physical world from its inception, using structured neural networks and active inference. The power of AI would not come from the data and how it is managed, but from the continuous, real time sensorimotor interaction between the user and the AI. This would be an AI where the learning algorithm and capabilities are developed together, with the ultimate goal of substantiating results without the need for human labeled input.

As in points 2, 4, and 5 above, the implicit ethical risk must be considered, which in this case refers to AI with freedom of judgment.  The aim of this concern is to ensure that AI will always be aligned with human needs and privacy.

Summarizing:

Taken together, the fields mentioned above show an AI that would be unlike anything that exists so far.

This picture is concrete:

.- A photonic neural network learns by settling into a lowest energy state.

.- Organoid intelligence leans toward causal reasoning. Pair it with a causal world model, and you might get a learning system.

.- Edge native swarms as highly efficient physical nodes could be deployed en masse. They would process sensory data on the spot, exchange causal features, and form a global perceptual grid that is self-organized.

This isn't just about making AI faster. It's about changing the relationship between intelligence, matter, and environment.

The paradigm shift: moving away from asking How can AI help to do a job faster and more efficiently? and towards How can AI expand the limits of what is physically and intellectually possible?

References:

1. Self-organising Memristive Networks as Physical Learning Systems (2025). Arxiv, Cornell University.

https://arxiv.org/abs/2509.00747

2. K. S. Riley, S. Koner, J. Osorio, and others. "Neuromorphic metamaterials for mechanosensing and perceptual associative learning". (2022). Arxiv, Cornell University.

https://arxiv.org/abs/2203.10171

3. Smirnova, L., Caffo, B.S., and others (2023). "Organoid intelligence (OI): the new frontier in biocomputing and intelligence-in-a dish". Frontiers.

https://www.frontiersin.org/journals/science/articles/10.3389/fsci.2023.1017235/full

4. Cai, H., Guo, F., and others (2023). "Brain organoid reservoir computing for artificial intelligence". Nature Electronics.

https://www.researchgate.net/publication/376413285_Brain_organoid_reservoir_computing_for_artificial_intelligence

5. Laurent Caraffa. BEDS: Bayesian Emergent Dissipative Structures. A Formal Framework for Sustainable Digital Twins and Continual Learning Systems. (2026). Univ. Gustave Eiffel, French National Institute of Geographic and Forest Information, Ministry of Ecological Transition, France. Arxiv.

https://arxiv.org/html/2601.02329v1

6. G. Auti, H. Daiguji, G. Tanaka. "Hebbian Physics Networks: A Self-Organizing Computational Architecture Based on Local Physical Laws". (2026). Arxiv, Cornell University.

https://arxiv.org/abs/2507.00641

7. G. Cauwenberghs. "Reverse engineering the cognitive brain" (2013). PNAS.

https://www.pnas.org/doi/10.1073/pnas.1313114110

8. Brunton, S.L., Proctor, J.L., & Kutz, J.N. (2016). "Discovering governing equations from data by sparse identification of nonlinear dynamical systems (SINDy)." Proceedings of the National Academy of Sciences (PNAS). 

https://www.pnas.org/doi/10.1073/pnas.1517384113.

9. Discovery of Physics From Data: Universal Laws and Discrepancies (2020). Frontiers.

https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2020.00025/full

10. G. Tanaka, T. Yamane, J. B. Héroux, and others. "Recent advances in physical reservoir computing: A review" (2019). ScienceDirect.

https://www.sciencedirect.com/science/article/pii/S0893608019300784

11. Decentralized Multi-Agent Swarms (DMAS) Architecture Research (2026). "Decentralized Multi-Agent Swarms for Autonomous Grid Security in Industrial IoT: Consensus-based Approach". Arxiv, Cornell University.

https://arxiv.org/abs/2601.17303

12. A. Sharma, A. Gupta, C. Wang. "Inducing Causal World Models in LLMs for Zero-Shot Physical Reasoning" (2025). Arxiv.

https://arxiv.org/html/2507.19855v4

13. AIT Staff Writer. "The Physics of Intelligence: Can AI Systems Develop an Internal Model of Reality?" (2026). AIThority.

https://aithority.com/ait-featured-posts/the-physics-of-intelligence-can-ai-systems-develop-an-internal-model-of-reality/