Иногда я пишу посты в LinkedIn, которые хочется процитировать и тут, но лень переводить и оформлять:

🎤 Frederik Jelinek of IBM is well known for his oft-quoted statement: “Every time I fire a linguist, the performance of the speech recognizer goes up.” And we’ve all seen what’s been happening with applied language processing systems like speech recognition, automated translation, and so on.

To some extent, this could be applied to plasma control.

The inherent complexity of most languages, multiplied by their constant evolution, has led to a desperate race to build holistic models of… in fact, the human mind. It’s no surprise that what has succeeded resembles human cognitive development more closely than algorithmic formalization.

🤔 So far, our main approach to producing fusion energy has been to make chaotic plasma even more chaotic, allowing us to achieve desired plasma parameters while sacrificing predictability and controllability—and then establish reliable control over the created complexity (feat. Einstein).

Just as in language processing, where machine learning has enabled systems to handle complexity without relying on rigid rules, the shift from rigid, rule-based systems to adaptive, data-driven models in plasma control may hold the key to unlocking more stable, efficient, and predictable performance.