Artificial intelligence is becoming increasingly powerful at finding patterns, making predictions and accelerating scientific discovery. Yet one important question remains: can AI explain how—and why—it reaches its conclusions?
This question lies at the heart of HEXA – Harnessing EXplainable AI for Next-Generation Quantum Devices, a project led by Prof. Karolina Słowik. HEXA brings together researchers with expertise ranging from quantum physics and computational chemistry to machine learning, pharmacy and economics. Its goal is to develop new materials and nanostructures for future photonic and quantum technologies while exploring how explainable AI can transform the way scientific discoveries are made.
The challenge extends beyond the laboratory. As society becomes increasingly dependent on digital technologies, the demand for energy continues to grow. Data centres, communication networks and AI systems consume vast amounts of power, making energy efficiency one of the major technological challenges of the coming decades.
Photonics offers a promising solution. Instead of relying solely on electricity, photonic technologies use light to transmit and process information. Because light can carry information rapidly and with relatively low losses, photonic devices have the potential to operate faster and more efficiently than conventional electronic systems. Researchers in HEXA design nanoscale structures capable of manipulating light in highly specific ways, creating components such as photodetectors, optical amplifiers, quantum light sources and thermal metasurfaces.
Importantly, energy efficiency does not have to come at the expense of performance. One of the project’s guiding assumptions is that better-designed materials can simultaneously consume fewer resources, waste less energy and deliver improved functionality. Understanding how a material’s microscopic properties influence its behaviour is therefore a key step toward developing more sustainable technologies.
This is precisely where artificial intelligence becomes valuable. Yet the team is not interested in AI as a black box that simply produces answers.
We do not want AI to act like a fortune teller saying: ‘This material will be good’. We want it to tell us what makes the material good. — prof. Słowik
For the researchers, explainability is particularly important because AI is increasingly used in areas that influence real-world decisions, from industry and infrastructure to healthcare and scientific research. Predictions alone are not enough; understanding the reasoning behind them is essential if knowledge is to be trusted, verified and applied.
The same philosophy guides the team’s work on quantum light sources. These devices generate light in an exceptionally controlled way, even photon by photon, making them promising components for future quantum communication systems. Because any attempt to intercept quantum information changes its state, quantum communication could provide fundamentally new approaches to secure data transmission and information processing.
At the same time, the project reaches beyond quantum technologies themselves. HEXA is developing AI workflows that can be transferred to other fields, including medicine and economics. While the data may differ, many scientific challenges involve similar steps: identifying patterns, building predictive models and understanding which factors drive the outcome.
The data and the final application may change, but the workflow—data preparation, feature extraction, model training, interpretation, and feedback—can remain similar. — prof. Słowik
Looking ahead, HEXA aims to contribute to a future shaped by energy-efficient photonic technologies, secure quantum communication, advanced sensors and smarter materials. Yet the broader ambition is equally important: creating a reusable framework that combines physical understanding with explainable AI to accelerate innovation across disciplines.
One achievement already stands out. The team’s theoretical designs have moved beyond simulations and into laboratory fabrication, bringing future devices one step closer to reality. For the researchers, this transition from concept to physical structure is a powerful reminder that understanding and innovation can advance hand in hand.
