Ten lectures on turbulence: 2. AI and turbulence

Artificial Intelligence (AI) has emerged as a significant paradigm in turbulence research, potentially constituting a “fourth methodology” alongside traditional numerical simulation, experimental observation, and theoretical analysis. While the advantages of AI are well-documented, this paper focuses on its current limitations. The following conclusions are proposed: 1) AI-driven research must transcend existing conceptual frameworks; otherwise, it remains unable to advance the fundamental understanding of turbulence regardless of the specific results. 2) A primary deficiency is the over-reliance on mathematical methodologies at the expense of physical deductive processes; consequently, AI fails to capture the underlying dynamical mechanisms of turbulence generation. 3) Without a major breakthrough in current methodological approaches, the discovery of novel turbulence physics remains highly improbable.