サイバネットシステム event: AI agents run CAE analysis

The talk came out of Cybernet Systems' CAE University Special Open Forum 2026, and its title frames the topic as learning from the stumbles of CAE education. The speaker, Yasuhiro Miyado, a senior specialist in the company's modeling and data science unit, used attempts to have AI agents operate CAE tools as the starting point. One observation the talk highlights is that CAE operation can essentially be seen as coding, which is likely why an AI that writes code can also run simulations.

The experiments went beyond simple steps. They covered nonlinear analysis and shape exploration under a time limit, and the talk notes AI actions that a CAE engineer would never take. That combination points to both the reach of autonomous execution and the gaps that remain, which is why the discussion turns to what is left for people.

The stakes appear to hinge on how the boundary is drawn between what agents handle and what humans decide. For engineers doing simulation work and for those teaching it, the practical question may be where judgment and management become the core skill, though the body does not specify how quickly such a shift would play out.