These two live treasures made me laugh all morning😄.
But having said that, my opinion is that the previous separation between academia and engineering was due to the limitations of productivity tools, which resulted in the results of academic research relying on a large number of engineering practices and a large number of engineers to practice the best engineering model.
This was reasonable in the past.
From the current perspective, only a few key academic achievements in large models can be made in a year.
Looking back, the results achieved in the AI ​​field in the past two years are all academic integration of engineering.
Especially after Manus, Cursor, and Augment have popularized the application of AI agents, model manufacturers have also quickly followed up on related capabilities, including internally they have gradually blurred the boundaries between research institutes and engineers.
To put it bluntly, the requirements for researchers are higher, because it is difficult for engineers to engage in academic research.

