Following recent conversations with friends and colleagues, Prof. Cláudio G. Schön, Prof. Ernst Kozeschnik and Dr. Johannes Österreicher, I have come to the conclusion that we should be more optimistic about the future of mankind, and of science, and physics in particular. This holds even when we take into account the geopolitical challenges of the present day and the emergence of Artificial Intelligence (AI).
In materials science, the number of possible combinations of elements, structures and processing routes is astronomically larger than anything a research community could explore by hand. Machine learning changes this arithmetic. Models trained on existing data can propose candidate compounds by the million, estimate their stability and rank them for experimental attention. Not every prediction survives contact with the laboratory, and those of us at the microscope know that a computed structure is a hypothesis rather than a result. Even so, a well-ranked list of hypotheses turns an unbounded search into a tractable one.
Quantum materials will revolutionise space and fusion research
Nowhere is this more exciting than in quantum materials, where subtle electronic correlations give rise to superconductivity, exotic magnetism and topological states. These phenomena are notoriously hard to predict from first principles. AI-guided exploration may prove most valuable here, not by replacing theory, but by revealing patterns in the data that theory can then seek to explain. Pairing such models with in situ electron microscopy closes a loop that was previously open: the model proposes, the microscope disposes, and the result feeds the next round of predictions.
The dream of thermonuclear fusion is increasingly recognised as a materials challenge as much as a plasma one. A reactor wall must endure extreme heat, hydrogen isotopes and fast neutrons, and retain its integrity throughout. If AI can help us identify which corners of the vast compositional space around tungsten alloys, high-entropy materials and radiation-tolerant copper deserve our limited beam time, we shorten the road to materials that survive inside a reactor. The same reasoning applies to spacecraft and long-duration habitats, which demand materials that are light, strong and radiation-tolerant. I do not claim this makes either dream imminent, but it removes one of the more stubborn bottlenecks.
This is not a belief that algorithms will do our thinking for us.
Models are only as good as the data on which they are trained, and the value of careful experiment rises in an age of abundant prediction, because experiment separates a plausible answer from a true one. What I am confident about is the direction of travel. Fusion and space exploration have been long-standing dreams, and dreams of that kind tend to be realised gradually and then, seemingly, all at once.
We intend to help build that future, one alloy and one micrograph at a time!
