AI Tools for Material Enhancement and Discovery in Energy Applications — Prof. Eliodoro Chiavazzo at NTU

On 21 May 2025, Prof. Eliodoro Chiavazzo delivered an invited lecture titled “AI Tools for Material Enhancement and Discovery in Energy Applications” at the IAS @ NTU STEM Graduate Colloquium, jointly organised by the Institute of Advanced Studies (IAS), Energy Research Institute @ NTU (ERI@N), and the Graduate Students' Club of MSE at NTU.

In his talk, Prof. Chiavazzo shared a vision of how artificial intelligence can transform the way we discover and design materials for energy applications.

When Data Carries Human Bias

Prof. Chiavazzo highlighted a subtle yet fundamental issue in data-driven materials science: most machine-learning models are trained on expert-curated datasets, which inevitably embed the community's historical preferences and assumptions. These cross-domain biases limit how far AI can explore the vast chemical space — models tend to “re-discover” what humans already know.

To address this, his team developed a classification-based strategy to detect and quantify such biases directly within materials databases. By distinguishing between specialised and general-purpose datasets, the method reveals where predictions are trustworthy and where they become extrapolations. This approach laid the foundation for the Energy-GNoME platform — an open, AI-driven and bias-corrected database containing over 38,000 candidate materials for energy technologies.

Toward Autonomous Discovery

Beyond bias correction, Prof. Chiavazzo outlined a future in which autonomous discovery pipelines unite machine learning, high-throughput computation, text mining, and robotics. In this ecosystem, AI not only predicts promising materials but also learns from each experimental cycle, continuously refining its models — a true “living” workflow for science.

This vision moves materials research from passive data analysis to active, self-improving discovery, where algorithms, simulations, and experiments collaborate to accelerate progress in sustainable energy technologies.

Prof. Chiavazzo's lecture at NTU captured the next frontier in energy materials research: combining human insight with unbiased AI reasoning to explore what lies beyond expert intuition.