The article examines salmon counting as an example of a field undergoing digital transformation. It argues that the problem is not purely technical: counting fish in rivers involves ecological variation, regulatory requirements, and the practical constraints of fieldwork. These factors make it a sociotechnical challenge, where the social and institutional context shapes what AI can and cannot do.
The piece highlights that while machine learning offers new tools for automation, the transition is neither smooth nor complete. Data collection remains difficult, and the reliability of models depends on conditions that humans must manage. The author stresses that human judgment and local knowledge stay essential, especially where errors carry management or conservation consequences.
Overall, the article presents a grounded view of AI adoption in a niche domain. It does not claim that AI is useless, but it cautions against expecting purely technical solutions to problems that are deeply embedded in human practices and environmental complexity.