---

Much has been made of the potential of AI to revolutionize the workplace. The range of tasks which can be performed by machines is expanding rapidly, and in many easily-defined tasks, such as playing chess, machines now comfortably out-perform humans. AI also brings the opportunity to automate many of the routine, repeated, tasks which arise in scientific research. But how AI will impact on the creative, conceptual, and problem-solving aspects of science, remains an open question.

In this talk we examine how AI contributed to the solution of a difficult problem in frustrated magnetism: the phase transition from a spin liquid described by a tensor gauge theory into a previously unknown form of magnetic order. This problem, which had defied conventional numerical simulation, was solved through a generative use of support vector machine (SVM), without prior training on related problems. However, neither the contributions of the SVM, nor that of the human researchers, proved decisive by themselves. Rather, success followed from a process resembling a collaboration between man and machine. We argue that this kind of “collaboration” may become the norm, especially in research involving large sets of data.

[1] “Identification of emergent constraints and hidden order in frustrated magnets using tensorial kernel methods of machine learning” J. Greitemann et al., Phys. Rev. B 100, 174408 (2019)
[2] “Human-machine collaboration: ordering mechanism of rank−2 spin liquid on breathing pyrochlore lattice”, N. Sadoune et al., arXiv:2402.10658

Please access here to register for the Zoom link.
https://forms.gle/mAQJyRWQVvJScE9g8

 

---